Startups & SMB

Empowering SMBs to do more with Gemini

Thu, 08 Oct 2026 12:00:00 +0000

Today, we announced the new Gemini agent, your single, universal agent for work. For small and midsize businesses, with lean teams and margins, Gemini will be a critical tool for you to operate efficiently, deliver excellent customer experiences, and grow.

Millions of small and midsize businesses, or SMBs, already trust Google AI across multiple services, like Gemini Enterprise, Google Ads, and Google Workspace. And they are scaling fast: Their usage of Google Cloud’s AI tools increased more than fivefold year over year.

Now, SMBs can benefit from all of the agentic capabilities of Gemini.

The agentic era for SMBs has arrived

Gemini functions as a personal assistant and team member, connects to the systems and data you already use, operates across the interfaces you’re already familiar with, and offers the model choice, security, and cost controls you expect from Google Cloud.

It answers your questions, handles your knowledge work, creates your images and media, and even writes and runs code, all from a single prompt box. You delegate an outcome and come back to finished work. Gemini can act as a flexible team member, playing a specific role such as a project manager, bookkeeper, inventory manager, or marketing consultant.

And, Gemini is architected to be helpful out of the box, with benefits including:

Connects to popular SMB tools: Gemini connects to the systems you regularly use, like Asana, Box, Clay, Docusign, Dropbox, GitHub, LegalZoom, Notion, Salesforce, Shopify, Slack, Wix, and more. It utilizes context from these tools, your business data, and work history, and gets smarter and more helpful the more you use it.
Provides model choice: Gemini is the agent, and the model underneath it is a separate choice. It runs each job on the model that fits best—orchestrating across our Gemini family of models and Claude models today, and other leading private and open models in the future. Matching the model to the work raises accuracy on the difficult jobs and lowers cost on the simple ones. And the leading model changes every few months, so keeping that choice open means that even if you switch models, your context, your skills, and your data stay put.
Accessible everywhere you work: Gemini works across the surfaces that you use every day, including on the web, Android and iOS mobile devices, Mac and Windows desktops, and applications like Google Workspace, Microsoft 365, and Slack. Because it runs in the cloud, it keeps working even when you step away from your devices.
Keeps costs under control: We’ve also implemented leading cost controls into Gemini, to ensure you and your employees can confidently put Gemini to work. This includes multi-model orchestration to choose the right model for the job, smart routing that automatically triages workloads so each one runs on the model that delivers maximum performance at the lowest possible cost, and real-time project-level spend caps in the Cloud Billing Console.

Gemini is in early access with SMB customers today and will be available to all customers soon.

How SMBs are driving results with Google Cloud AI today

Globally, SMBs are already using Gemini Enterprise to drive their businesses forward, here are a few examples:

Aerotech, a U.S. industrial automation company, leverages Gemini Enterprise to scale its internal AI problem-solving platform across its workforce, reducing troubleshooting times to increase production capacity.
Angelus, a Brazilian dental and healthcare manufacturing company, uses Gemini Enterprise to streamline project management across its research and development department. This enables its teams to automatically pull technical project data into pre-approved templates aligned with the company’s brand identity and regulatory requirements.
EdSpan Global, an international education platform, uses Gemini Enterprise to automate the standardization of thousands of digital learning objects, reducing time spent on repetitive editing to accelerate course launches.
Eficode, a software development and DevOps firm in Finland, is using Gemini Enterprise to create a multi-platform support engine that quickly indexes complex internal wikis, so employees can provide customers with faster technical support.
KLog.co, a Chilean logistics technology company digitizing freight forwarding across Latin America, uses Gemini Enterprise to automate cargo tracking and shipping paperwork, cutting manual data entry errors by more than 90% and increasing document processing capacity tenfold.
Koufu, one of Singapore's largest food and beverage companies, is using Gemini Enterprise to automate sales reporting and data analysis. This reduces spreadsheet bottlenecks so shop managers can focus on front-of-house service quality.
La Maison du Whisky (whisky.fr), a specialized spirits and beverages retailer in France and internationally, chose Gemini Enterprise to create a 'Digital Sommelier' that transforms technical product data into high-end storytelling and marketing copy in seconds.
NEEOH, a leading Brazilian out-of-home advertising communication platform, uses Gemini Enterprise to standardize secure AI usage across its organization, enabling teams to generate campaign copy and build pitch proposals faster while keeping corporate client data secure.
Quadrant Health Group, a U.S.-based healthcare and clinical operations provider, uses Gemini Enterprise to unify its fragmented technology systems, automate repetitive back-office tasks, and empower clinicians with secure AI tools that safeguard patient data.
Salt Productions and Pinch Creative, U.S.-based event production and creative partner agencies, are using Gemini Enterprise to turn fragmented files into a searchable, single source of truth, reducing administrative work and presentation bottlenecks so staff can focus on event production and design.
Sep 2, a digital security provider, uses Gemini Enterprise to deploy autonomous AI agents for 24/7 threat monitoring. It can accelerate incident detection and neutralize security threats.
Stratagem, a technology consulting company based in the U.S., uses Gemini Enterprise to automate administrative workflows and develop retail customer service agents, freeing consultants to spend more time with clients and accelerate service delivery.
Sunhouse, a strategic brand design agency, uses Gemini Enterprise to easily find archived design work stored on Google Drive, enabling its teams to spend less time hunting for files and more time growing its business with global brands.

Get trained on Google Cloud AI

To help SMB teams get the most out of AI, we’ve created easy, no-cost learning programs that you can use through our Google Skills platform. To get started today, there are two ways to engage on the platform for free:

Claim 35 no-cost credits to access hands-on labs and earn skill badges across all levels
Sign your team up for Google Skills for Organizations

Today, Google Skills offers more than 3,000 courses, labs, and credentials to help you build practical skills with Google AI and products like Google Workspace and Gemini Enterprise, earn skill badges, and achieve valuable certifications on Google Cloud. Courses on the new Gemini agent will be available soon; in the meantime, you can get started with recommended courses such as:

Exploring Data Transformation with Google Cloud: For SMBs looking to move past simple tools and actually start scaling their broader operations on Google Cloud, this course provides a great entry point for business leaders looking to modernize their infrastructure and security posturing.
Google Workspace with Gemini: For SMBs looking to quickly boost everyday efficiency, this course serves as an essential, highly practical guide to seamlessly integrating AI into daily business operations, from drafting emails to automating simple workflows directly inside Gmail, Docs, and Drive.
SMB Learning Path: For SMBs looking for practical ways to scale their business operations, this catalog of courses focuses on combining introductory gen AI technical foundations with Gemini-led automation and the GEAR framework for custom agent development.

Work with a partner

Our partner ecosystem plays an important role in helping businesses implement Google Cloud AI, including Gemini.

Today, many partners offer certified experts on Google Cloud AI to SMBs, and can help with everything from identifying use cases for AI, prototyping, building custom agents, and more.

Visit our Google Cloud Partner Network today to get started with a partner.

Innovation in Ireland: How Irish brands scale with Gemini Enterprise

Thu, 08 Oct 2026 12:00:00 +0000

In recent decades, Ireland has grown into a vibrant hub for global technology. As modernization accelerates, Irish organizations — spanning government agencies, established enterprise brands, and growing startups in AI and other areas — are all advancing their technical foundations. Today, leaders across the country are injecting agentic AI into every layer of their business and operational workflows.

We are seeing organizations move from early prototypes to production-ready deployments using Gemini Enterprise. This progress supports Ireland's competitive position within the European market. According to a recent Implement Consulting Group report assessing Google’s local contribution, public and private leaders estimate between 40 billion to 45 billion euros of economic potential in global AI markets.

This ambition aligns with broader government initiatives. Since assuming the presidency of the Council of the European Union for the second half of 2026, Ireland has steered higher-level technology and policy discussions, outlining a National Digital and AI Strategy to improve productivity and innovation across sectors.

To support this strategy, we are focused on making these capabilities more accessible. Our local teams are collaborating with industry leaders to implement Gemini Enterprise which offers capabilities that range from conversational AI agents and natural language data analysis, to optimizing operations with agentic ecosystems.

The economic impact is already taking shape. In 2025 alone, Google operations — including our AI and cloud services — added an estimated 10 billion euros to the Irish gross domestic product. And this momentum is shared: Organizations across Ireland are using Google Cloud to modernize infrastructure and put AI to work.

Here are several ways Irish organizations are modernizing with Google Cloud and Gemini Enterprise:

Ryanair supports millions of passengers through AI-backed operations

Ryanair, Europe’s largest airline, is modernizing its collaborative operations and infrastructure by deploying Gemini Enterprise and Google Workspace for some 35,000 employees. Flight crews, operations planners, and corporate staff will benefit from streamlined decision-making, optimized crew logistics, and enhanced corporate productivity. At the same time, a dual-cloud strategy strengthens infrastructure resilience as the airline targets growth to 300 million passengers by 2034.

Smyths Toys rethinks the customer experience with agentic support

As one the world’s largest specialty toy retailers, Smyths Toys is reshaping its customer support with Google Cloud and Gemini Enterprise. By introducing an AI agent named Codie, the retailer has resolved more than 60% of its web inquiries and reduced holiday email backlogs. Customer care teams that previously handled up to 1,000 peak-season emails a day can now focus on the roughly 50 cases that need more complex judgment.

The Irish Revenue Commissioners: Advancing public sector data

Public sector agencies are also making strides in Ireland. The Irish Revenue Commissioners — responsible for taxes, customs, and keeping Ireland’s economy running smoothly — is working with Google Cloud to migrate from legacy on-premise systems to a unified AI-lakehouse architecture designed to break down data silos and establish a single source of truth. Strict data sovereignty controls safeguard taxpayer information while Gemini Enterprise powers agentic data analysis, helping Revenue deliver secure, responsive public services.

Virgin Media Ireland: Accelerating workflows

Supported by Liberty Global's partnership with Google Cloud, Virgin Media Ireland is moving AI from exploration to production. Through modern infrastructure and automation, the team now deploys workloads three times faster, is in the process of rolling out multiple AI agents, and is launching new customer services designed to drive business growth.

Irish startups scaling with enterprise tools

While established enterprises are modernizing legacy systems, Irish startups are building on modern technology from day one. Backed by a strong local tech ecosystem, these agile teams are creating services that compete on an international stage.

Hexis: Scaling personalized sports nutrition

Hexis builds technology solutions that help athletes refine their performance and recovery. By storing and processing that real-time biometric and behavioral data in Google Cloud, Hexis uses Gemini Enterprise to power natural language conversations that explain fueling adjustments to users throughout the day. Today, the platform supports 40% of the Tour de France peloton and half of Premier League clubs during the 2026–2027 season.

Kitman Labs: Unifying performance analytics

Founded in Dublin in 2012, Kitman Labs provides sports science and performance intelligence to more than 4,000 sporting organizations across 26 countries. By combining its intelligence platform with advanced analytics and Gemini Enterprise, Kitman Labs helps teams make data-informed decisions that support athlete health, lower injury risk, and extend player career longevity.

Spryt: Automating healthcare coordination

Spryt streamlines healthcare coordination across the U.K. and U.S., using AI built on Google Cloud and Gemini Enterprise to optimize patients’ experiences, reduce missed appointments, and cut administrative burden. The company collaborates with the NHS, operates in more than 70 US clinics, and has analyzed 224 million appointment records as part of a multiyear initiative with Mayo Clinic.

IMPT: Automating carbon offsetting

By connecting AI, blockchain, and Google Cloud, IMPT embeds verified carbon offsetting into everyday hotel bookings and online purchases. Following a 15-week migration to Google Cloud, IMPT supports 8 million listings across 195 countries. With Gemini Enterprise, the company’s lean team handles about 400 daily customer chats with a six-second median response time, allowing staff to focus on growing its business rather than managing systems.

Transform ideas from planning to production

As Ireland continues to expand as a global technology hub, local organizations are demonstrating how to translate AI potential into practical results. Explore our customer stories to see how other organizations are building with Google Cloud, or reach out to our team in Ireland to accelerate your journey.

Cloud CISO Perspectives: How cybersecurity startups can win CISOs

Wed, 30 Sep 2026 16:00:00 +0000

Welcome to the second Cloud CISO Perspectives for September 2026. Today, Alicja Cade and Nick Godfrey, senior directors, Office of the CISO, share their guidance for cybersecurity startups on how to win over the CISOs who will become crucial business partners and customers.

As with all Cloud CISO Perspectives, the contents of this newsletter are posted to the Google Cloud blog. If you’re reading this on the website and you’d like to receive the email version, you can subscribe here.

aside_block: <ListValue: [StructValue([('title', 'Get vital board insights with Google Cloud'), ('body', <wagtail.rich_text.RichText object at 0x7f4c39f04090>), ('btn_text', 'Visit the hub'), ('href', 'https://cloud.google.com/solutions/security/board-of-directors?utm_source=cgc-site&utm_medium=et&utm_campaign=FY26-Q2-GLOBAL-GCP39634-email-dl-dgcsm-CISOP-NL-177159&utm_content=-&utm_term=-'), ('image', <GAEImage: GCAT-replacement-logo-A>)])]>

How cybersecurity startups can win CISOs

By Alicja Cade, Senior Director, Financial Services, Office of the CISO, and Nick Godfrey, Senior Director, Office of the CISO

Alicja Cade, Senior Director, Financial Services, Office of the CISO

Cybersecurity startups play a crucial role in technology as they build to solve both legacy, existential challenges and the latest problems on the cutting edge. A key part of winning and transforming the cybersecurity field is becoming a strategic partner to CISOs and their security teams.

Google has supported more than 50 cybersecurity founders over the past four years through our Google for Startups program, including Authologic, BforeAI, Build38, Cerby, Crowdsec, Risk Ledger, and Mokn.

Christian Torres, co-founder and CEO, Kriptos, and a Google for Startups participant, said that building connections between startups and CISOs is crucial to solving critical security challenges.

Nick Godfrey, Senior Director, Office of the CISO

"The Google for Startups program has been the most impactful initiative we've joined as a cybersecurity company. Unlike other accelerator programs, this one speaks our language — the challenges, the ecosystem, and the conversations are 100% aligned with what we do every day at Kriptos. The access to CISOs and security leaders has been invaluable, and the connections we've built through the program are ones we now see regularly across industry events. It's put us exactly where we need to be,” he said.

Cybersecurity startup founders face many competing taskmasters as they fight for survival, from demanding capital funders to the relentless pressure of growing their market and networks. CISOs should be key stakeholders for cybersecurity startups so that founders focus on solving thorny challenges in a way that works in the real world.

Listening to CISOs and understanding the businesses that they serve takes time and effort, and if done right can help deliver better value and create a lasting enterprise foundation and network of allies.

Here are three top tips from September’s Gemini Startup Forum for Cybersecurity, part of the Google for Startups program, where we offered vital guidance, addressed critical domains, and helped foster deep dialogue for the next generation of AI-native cybersecurity startups.

Tip 1: Listen then design and deliver for your customers

Avoid becoming a round peg in a square hole by combining your problem-solving startup with listening to CISOs who have to protect real systems, networks, and people. Listening to CISOs and understanding the businesses that they serve takes time and effort, and if done right can help deliver better value and create a lasting enterprise foundation and network of allies.

Here’s how to develop trusted CISO relationships:

Host diagnostics meetings. Your meetings with CISOs should focus on mapping their operational bottlenecks and co-authoring collaborative solutions while studying their pain points.
Create "unselling" spaces to build peer trust. Host intimate, pitch-free roundtable discussions on industry challenges or establish a critique-only advisory board to build genuine relationships with CISOs without the pressure of a sales environment.
Use neutral networks that don’t include venture capitalists. Engage with CISOs in low-friction environments by contributing to open-source security projects and participating in academic and geopolitical risk forums where security leaders gather to solve broad industry problems.
Avoid the bait-and-switch pitch. Never disguise a sales pitch as a research or feedback session, as tricking a CISO into a product demo will permanently destroy their trust.
Center their business context. Don’t limit your listening to the technical security stack, because the CISO’s primary job is to enable and protect the broader business strategy.

Tip 2: Evaluate AI security to filter out noise

Instead of just using AI to assemble the product, startups should critically evaluate what makes your approach unique and how you communicate that to potential customers.

Define your moat by investing in proprietary datasets, specialized fine-tuning, and unique orchestration layers that create a true technical moat. Don’t be a wrapper.
Secure the intelligence by proactively designing your models to resist adversarial attacks, prompt injection, and data poisoning. In cybersecurity, model robustness is your ultimate trust signal.
Deliver high-fidelity outcomes by clearly communicating how your AI product reduces cognitive load for defenders, minimizes false positives, and integrates safely into existing operations.

Avoid using generic marketing buzzwords like "cognitive," "autonomous," or "revolutionary" without the technical documentation, case studies, and whitepapers to back them up. In a skeptical market, transparency is your best sales tool.

Tip 3: Enthusiastically embrace your sector

Keep a sharp eye out for common due diligence pitfalls during investment and merger and acquisition cycles. These include ensuring that internal engineering and cybersecurity practices meet external claims, but also evaluating the regulatory context of your business sector as well as the security and reliability of your product and service.

You have to know whether you’re required to abide by data sovereignty, data residency, and other requirements. To avoid this pitfall, engage with broader stakeholders early who know the sector and its nuances well.

How to keep the conversation going

Even beyond the crowded field of aspiring cybersecurity companies, startups broadly can benefit immensely by making sure that they listen carefully, evaluate objectively, and take to their sector requirements enthusiastically.

"Google's Office of the CISO has been a bridge between LetsData and the security leaders we need to reach. Sometimes that bridge is advice on how our offering maps to a CISO's real priorities. Sometimes it is a direct introduction to a CISO who is looking for exactly what we build. For a startup, a warm introduction at that level is priceless,” said Ksenia Iliuk, founder and COO, LetsData.

To learn more about how Google Cloud’s Office of the CISO can help support your organization, check out our CISO Insights hub.

aside_block: <ListValue: [StructValue([('title', 'Learn something new'), ('body', <wagtail.rich_text.RichText object at 0x7f4c38d88050>), ('btn_text', 'Watch now'), ('href', 'https://www.youtube.com/watch?v=Wpo-5ke9uvQ'), ('image', <GAEImage: Cloud-CISO-Perspectives-logo-A>)])]>

In case you missed it

Here are the latest updates, products, services, and resources from our security teams so far this month:

Agentic hacks, real proofs: Inside Google's PageBreak project: Distinguishing a genuine, exploitable flaw from a convincing hallucination has become a major challenge, often increasing the burden on product teams. PageBreak is an internal AI agent of Google's Product Security team developed to test the security of our first-party web applications and address this challenge. Read more.
Investing together: Wiz Defend and Google Security Operations: Continuing to deepen the integration between Wiz Defend and Google Security Operations, helping teams work faster wherever they choose to investigate. Read more.
A unified view of Android security updates for enterprises and OEMs: We're introducing new libraries that give enterprise partners and OEMs a complete, real-time picture of a device’s security posture. Read more.
Delivering new partner security agents and AI defenses with Gemini Enterprise: We're expanding our catalog of partner-built security offerings in the Gemini Enterprise ecosystem to help you leverage your full security context. Read more.
Google named a Leader in the External Threat Intelligence Service Forrester Wave: We are proud to announce that Forrester has named Google a Leader in The Forrester Wave™: External Threat Intelligence Service Providers, Q3 2026. Read more.
Wiz named a Leader in the Proactive Security Platforms Forrester Wave: Forrester’s Proactive Security Platforms evaluation for Q3 2026 rated Wiz with top scores across eight areas, reflecting our commitment to securing the AI era. Read more.
Strengthen your CI/CD pipeline with new Secure Source Manager capabilities: To help you better address software supply chain threats, our Secure Source Manager lets you manage your source and CI/CD systems with unified authentication and authorization mechanisms. Read more.
Building an AI detection engine that understands agent intent: Analyzing model input and output logs in an AI-native detection pipeline to understand and uncover malicious AI agent behavior. Read more.
Using AI to discover and fix high-priority exposures across public services and critical infrastructure: New initiative partners with under-resourced organizations to uncover, remediate exploitable risk at scale. Read more.

Please visit the Google Cloud blog for more security stories published this month.

aside_block: <ListValue: [StructValue([('title', 'Join the Google Cloud CISO Community'), ('body', <wagtail.rich_text.RichText object at 0x7f4c38d8bb10>), ('btn_text', 'Learn more'), ('href', 'https://rsvp.withgoogle.com/events/google-cloud-ciso-community-interest-form-2026?utm_source=cgc-blog&utm_medium=blog&utm_campaign=FY25-Q1-global-GCP30328-physicalevent-er-dgcsm-parent-CISO-community-2025&utm_content=cisop_&utm_term=-'), ('image', <GAEImage: GCAT-replacement-logo-A>)])]>

Threat Intelligence news

ShinyHunters renewed mass exploitation campaign targeting Oracle PeopleSoft: Mandiant and Google Threat Intelligence Group (GTIG) have identified renewed mass exploitation of CVE-2026-35273 by UNC6240 (ShinyHunters), along with expanded global targeting across multiple sectors. Read more.
Proactively defend by hardening code pipelines and CI/CD infrastructure: Check out our actionable blueprint for software and platform architects designed to safeguard the software supply chain against threat vectors that are actively being exploited, third-party risks, and architectural vulnerabilities throughout the entire software development lifecycle. Read more.
Infostealer incursion: How stolen credentials breach cloud, code, and AI environments: Wiz Research analyzes NordStellar data to map the credentials targeted by infostealer families and assess their potential impact across cloud, code, and AI environments. Read more.

Please visit the Google Cloud blog for more threat intelligence stories published this month.

Now hear this: Podcasts from Google Cloud

Cloud Security Podcast: Patching browsers with AI, agents, Rust, and your tabs: Jasika Bawa and Doug Turner of Chrome Security explore how Google Chrome now uses AI agents to autonomously identify and patch security vulnerabilities at an unprecedented scale, significantly accelerating the browser's update cadence. Listen here.
Cloud Security Podcast: All about Project Atlas, Wiz's AI vulnerability research: Nir Orfeld, head of vulnerability research, Wiz, discusses how his team uses multi-agent AI systems for discovering high-impact zero-day vulnerabilities in cloud infrastructure. Listen here.
Cloud Security Podcast: How Google eliminates classes of vulnerabilities at scale: How do you build the foundations for a secure Google-scale enterprise that stays secure even if an AI is writing the code and nobody has time to review it? Christoph Kern, principal security engineer, Google, explores what secure-by-design really means in the AI era. Listen here.

To have our Cloud CISO Perspectives post delivered twice a month to your inbox, sign up for our newsletter. We’ll be back in a few weeks with more security-related updates from Google Cloud.

Why your startup needs open models alongside frontier APIs

Mon, 28 Sep 2026 16:00:00 +0000

Every week, I talk with founders who are building at an unbelievable pace. Teams are moving from inception to product-market fit faster than ever, with foundation models wired deeply into their core product workflows.

Yet as startup architectures mature, a clear divide has emerged between teams struggling with margins and those scaling sustainably. The most effective engineering teams have abandoned the one-size-fits-all model strategy.

In the early days of LLMs the default architecture was simple: send every interaction to the largest model available. But as applications move into production, serving millions of people and running autonomous multi-agent workflows, relying on a single frontier model starts to strain in three places:

Latency penalties: Relying entirely on cloud round trips makes it difficult to deliver the sub-second responsiveness that interactive mobile and desktop apps require.
Infrastructure overhead: Self-hosting large open models with more than 70 billion parameters forces early-stage teams to act like infrastructure providers, pulling senior engineers on cluster provisioning and multi-GPU orchestration.
Margin erosion: Sending high-frequency, structured tasks (like intent routing, JSON extraction, or status validation) to general-purpose frontier endpoints spends capital that could be funding product differentiation.

Great engineering teams pick the right tool for each job. Most production requests don’t require a frontier generalist, and routing every call to one can actually slow your product down. Instead, the winning pattern is a compound AI stack: pairing frontier models for complex synthesis with compact, open-weight models that you can tune, control, and run anywhere.

It’s for these reasons that an open model like Gemma belongs in your model lineup. With more than one billion downloads across the developer community, Gemma 4 is our most capable open model family to date, using the same foundational research and technology behind the Gemini models.

Built under one roof

Gemma is built by Google DeepMind using the same foundational research and architecture advances behind the Gemini models. Because they share common DNA and developer tooling, your team can prototype in Google AI Studio and design hybrid architectures where Gemini and Gemma work together.

Released under a commercially permissive Apache 2.0 license, Gemma 4 is engineered for parameter and token efficiency. Rather than forcing a single model architecture onto every hardware target, Gemma 4 spans five sizes across four specialized architectures: compact E2B and E4B models with native audio and vision for mobile and edge devices; an encoder-free 12B Unified multimodal model; a 26B A4B Mixture-of-Experts (MoE) model that activates only 4B parameters per token for high-throughput serving; and a dense 31B model that fits on a single GPU for maximum reasoning quality and fine-tuning. Every model includes configurable thinking modes, native function calling, up to 256K context, and built-in Multi-Token Prediction (MTP) draft models for speculative decoding.

Real proof: How startups are winning with Gemma

Founders are using Gemma to solve urgent problems around unit economics, output accuracy, and responsiveness.

Flipping the architecture: Cue is a voice-activated desktop assistant that runs natively on a user's machine to automate everyday tasks. They integrated Gemma 4 E4B via Ollama on local hardware to handle real-time transcript formatting. While they originally planned for Gemma to be a weak offline fallback, benchmarking proved it was so fast and precise that they made it their default engine—driving a 44% latency drop (from 876 ms to 488 ms).
True edge independence: Mobile development studio HubX built BetterSpeak, a voice-based interactive mobile English-learning tutor that simulates immersive, real-time voice conversations. To bypass cellular network lag and avoid charging users expensive subscription fees to cover cloud hosting, they packaged a 4-bit quantized Gemma 4 E2B model (~2.9 GB) natively on-device. The result is an offline, speech-to-speech mobile tutor that costs them $0 in server bills.
Scientific discovery and air-gapped security: K-Dense has built Faraday, an AI-powered scientific collaborator optimized end-to-end across hardware, software, and sensor suites, powered by Gemma 4 together with K-Dense's Scientific Agent Skills. Faraday runs fully air-gapped, making it suitable for secure, proprietary scientific work in pharma and biotech. Deployed on an NVIDIA DGX Spark, Gemma 4 can also be fine-tuned locally on a user's own proprietary datasets.
Unlocking infinite gameplay and retention: Gaming company Latitude integrated Gemma across their AI-native game products. By swapping in Gemma for AI Dungeon, they significantly improved player retention, while their new AI RPG platform Voyage leverages Gemma to deliver high intelligence at a cost that enables unlimited user gameplay with ultra-fast latency.

Four workloads where Gemma wins for startups

If you’re evaluating where Gemma fits into your stack today, start with these four jobs:

1. Edge and local execution (low latency, true privacy)

If you’re building mobile apps, developer desktop tools, robotics, or offline-first experiences, every cloud round-trip adds latency that people can feel. Gemma can run directly on laptops (including Apple silicon), smartphones, and local appliances. Your users get immediate feedback, and sensitive data never has to leave their device.

You can handle many local interactions on-device for zero incremental cost, and keep a bridge to frontier models in the cloud for the requests that need it. When a local workflow calls for large-scale multimodal reasoning, long-context data synthesis, or complex planning, your application can route that specific request to Gemini.

2. High-throughput triage and agent routing

In multi-agent architectures, agents spend a surprising amount of tokens on simple tasks like checking statuses, classifying intent, and routing tickets. With Gemma as your front-line gatekeeper, those high-volume background tasks run on a compact model and your team can save frontier reasoning for the requests where it creates product value.

3. Task-specific fine-tuning for real moats

Adapting a model to your proprietary data is one way to build a competitive moat. Because Gemma gives you full access to model weights and has a compact memory footprint, your team can run parameter-efficient fine-tuning (LoRA or QLoRA) on a single GPU in hours rather than days.

4. Turnkey vertical starting lines

DeepMind releases domain-specific variants of Gemma, so you don’t have to start from scratch. One example is MedGemma. MedGemma scores 87.7% on the MedQA benchmark, matching the clinical accuracy of frontier models at roughly one-tenth the inference cost. In a blind clinical study, board-certified radiologists judged that 81% of chest X-ray reports generated by the lightweight MedGemma 1.5 4B were accurate enough to result in equivalent patient management compared to reports written by human experts.

Beyond healthcare, biotech startups use C2S Scale to model virtual cellular responses and accelerate oncology research. Meanwhile, DataGemma cross-references more than 240 billion public data points to help reduce numerical hallucinations. If you’re operating in a specialized market, starting with a model that already speaks your industry's language can save you engineering time and compute budget.

Deploy wherever your business lives

Gemma is designed to fit into your existing engineering stack without lock-in:

Apache 2.0 licensing: Gemma 4 ships under the Apache 2.0 license, giving startups the freedom to fine-tune, quantize, redistribute, and deploy commercial products on-premises or at the edge with full ownership of their custom weights.
Day-zero open tooling: Run and fine-tune Gemma with the tools your engineers already use, including vLLM, Ollama, llama.cpp, LM Studio, MLX, Unsloth, Hugging Face, Kaggle, Keras, PyTorch, JAX, and LiteRT-LM.
Serverless and managed cloud deployment: Prototype immediately in Google AI Studio, scale to zero on serverless GPUs with Cloud Run, or deploy dedicated endpoints from Model Garden on Gemini Enterprise Agent Platform when traffic surges and you don’t want to manage GPU clusters.
Enterprise-ready safety: Gemma undergoes rigorous pre-release safety evaluations, data filtering, and red-teaming, and pairs with ShieldGemma 2 to help you meet enterprise compliance requirements.

Build with Gemma: What to do this week

Great technical architecture isn't about finding one model to do everything. It’s about assembling the right tool for each job so you can move faster, protect your runway, and ship a superior product.

Here’s my challenge to your engineering team this week:

Audit your model calls: Look at your logging dashboard and identify three high-volume, deterministic tasks (such as intent classification, JSON validation, or summarization) currently running on your most expensive models.
Benchmark Gemma: Run a quick test with a compact Gemma model locally or on a single endpoint. Measure the latency and calculate what happens to your gross margins when that workload runs with lower inference cost.
Redirect your runway: Take the capital and engineering hours you save on compute and invest them back into your core differentiators.

You can download the Gemma weights directly or deploy them through Model Garden. If you need compute credits and technical architecture reviews to get up and running, the Google for Startups team is ready to help you build - learn more.

The three things today's hottest startups are looking for in their AI stack

Thu, 24 Sep 2026 13:00:00 +0000

Google Cloud has become the platform of choice for startups building AI.

Our uniquely complete stack — including a choice of first- and third-party compute and models; our platform for building and managing agents; and our products for securing AI workloads — has emerged as the single most important driver of this growth and it is powering AI development for many of the most exciting and innovative startups in the world.

As a result, startups are choosing to build and run on Google Cloud at a higher rate than they were three years ago, at the start of the AI era.

Given how quickly the technology industry moves in the AI era, the choices startups make can be notable. As we’ve worked together and watch many of these leaders scale, we’ve observed a few important trends emerging over the past several months. We expect these decisions will continue to shape the choices startups make about the platforms and technology they use:

Gemini Enterprise, which includes our tools for managing TPU and GPU clusters, services for building and managing agents, and APIs to access both first- and third-party models, is growing significantly with startups. And when startups use Gemini Enterprise, they also tend to use our “core cloud” services like Storage, BigQuery, or GKE.
Gemini models — as well as several of the third-party models available through Gemini Enterprise — are providing very strong price-performance for startups. These customers are increasingly deploying both our frontier models and “workhorse” models as their AI to power workloads as diverse as scientific research, generative media creation, and financial analysis.
Access to compute on GPUs and TPUs is critical for AI and the ability to choose one — or both — is unique to Google Cloud. But importantly, startups almost always use additional products from our stack alongside these chips, like models, tools for building agents, or services like BigQuery or GKE. These additional technologies illustrate how the needs of startups are rarely singular, and just how much value they find in having ready access to a strong suite of second-, third-, and fourth-level technologies beyond just compute.

We can see the demand for these technologies first-hand in some of the recent deals we have struck in the past 60 days with a number of leading startups across sectors:

Artificial Agency, a startup focused on generative behavior in games, is running critical AI workloads and research on Google Cloud, where it is using NVIDIA GPUs for model training and inference, as well as Gemini models and Cloud Storage.
Arya Health is building the AI workforce for healthcare, deploying agentic AI to perform the non-clinical administrative work that limits providers’ ability to deliver and expand care. Arya’s AI agents work across scheduling, intake, recruiting, onboarding, compliance, after-hours operations, and other critical workflows, interacting through voice, text, email, and providers’ existing systems. Arya uses a range of Gemini models across its agentic infrastructure, including Gemini 2.5 Pro, 3.1 Flash, and 3.5 Flash Lite, selecting models based on the reasoning, speed, and cost requirements of each workflow.
Casco is a cybersecurity startup whose autonomous agent swarms execute sophisticated, multi-step attacks to uncover vulnerabilities across enterprise applications, cloud environments, and infrastructure. Its architecture combines advanced reasoning models for complex, long-running tasks with fast models such as Gemini 3.5 Flash for focused subagent work. Google Cloud’s model portfolio and infrastructure, including Provisioned Throughput, help Casco match each workload with the right combination of intelligence, speed, and capacity.
CodeRabbit has been a pioneer in independent AI code review and has expanded that layer into Agentic Change Management, the control plane for agentic software development. They use our Cloud Run and Storage products to underpin their application, and are now beginning to leverage Gemini 3.1 Pro and other Gemini models to power use cases like analyzing how a single code change impacts an entire project, writing clear and contextual review comments to explain logic bugs, and instantly generating one-click fixes for developers.
Comfy offers a platform for creatives to build brand-consistent content and media with generative AI. They are utilizing a mix of NVIDIA systems and Google media generation models, like Veo and Nano Banana, in their platform.
MicroAGI, a German AI robotics startup, recently announced they would access NVIDIA Blackwell systems for model training through Google Cloud. They will also use Gemini Enterprise Agent Platform and AI models on Google Cloud to help robotics process multimodal information like video.
Ineffable Intelligence, the London-based superintelligence startup, recently announced a partnership with Google Cloud to access NVIDIA Vera Rubin systems as well as high-efficiency AI networking and storage.
PEAR Health Labs built and runs its AI health and fitness companion and agentic health platform entirely on Google Cloud, using our Gemini 3.5 Flash model as well as our data cloud products.
Roboforce is a physical AI company building scalable robots for industrial environments. They recently signed a new agreement with Google Cloud to utilize our GPU-powered VMs, which will be used to train and serve their custom and post-trained physical AI models.
xFigura offers a platform that gives architecture teams a single, secure canvas that brings generative models into one place for AI-driven design. Built on Google Cloud, xFigura uses models like Nano Banana and Omni to help designers generate and refine concepts in plain language, with authorship staying with the architect.
SciFin is a startup that helps revenue teams uncover "revenue reality" by converging fragmented sales context across CRM systems, documents, emails, and other sources. They are utilizing Gemini models, infrastructure, and Gemini Enterprise to build and run their platform.

These new and expanding customers represent some of the most exciting names in AI and are demonstrative of the type of successes that startups are having with Google Cloud’s uniquely complete stack for building AI.

To learn more about Google Cloud’s work with leading AI startups, or to get started building with us, visit here.

How a solo founder runs a five-continent tender platform on AlloyDB and MCP

Thu, 17 Sep 2026 16:00:00 +0000

Editor's note: Lucius AI, a tender-intelligence startup covering markets across five continents, runs its entire data platform on AlloyDB for PostgreSQL with a single operator. By migrating semantic search to a ScaNN index and managing database operations through Model Context Protocol (MCP), query latency dropped by 47x while automating day-to-day administrative tasks via MCP.

Executive summary

Lucius AI runs a global tender platform spanning more than 210,000 tenders across the UK, EU, India, and Australia, requiring minimal operational overhead for a solo founder.
Lucius AI deployed AlloyDB for PostgreSQL to consolidate its relational catalog, audit logs, and vector embeddings into a single managed database engine.
Migrating semantic search to a ScaNN index lowered query latency from 1.14 seconds to 24 milliseconds — a 47x speedup on a representative production query.
Connecting an AI agent to AlloyDB using the Model Context Protocol (MCP) helps Lucius AI automate query analysis, data freshness checks, and incident forensics under strict least-privilege permissions.

Making tender intelligence work as a company of one

Lucius AI helps businesses bidding on public contracts evaluate opportunities across global markets. The platform ingests public procurement notices from the UK, the EU, the US and Canada, Australia and New Zealand, India and Singapore, alongside World Bank donor-funded notices across Africa and Asia. Lucius AI analyzes tender documents using Gemini to generate compliance matrices, bid recommendations, and draft responses citing original source pages. For small and mid-sized suppliers, this replaces days of manual document reviews and costly external consulting.

Running a platform of this scope requires extensive operational coordination:

Nightly ingestion from thirteen public procurement sources
A catalog of more than 210,000 tenders, including tens of thousands open for active bidding
Two production regions on Cloud Run: Europe, and an Australian deployment on its own AlloyDB cluster with customer-managed encryption keys (CMEK) for defense-adjacent customers
Ongoing analytics, performance tuning, data validation, and incident response

Managing these responsibilities without dedicated data engineering or database administration teams requires offloading operational maintenance. Lucius AI addressed this challenge on two fronts: using AlloyDB for PostgreSQL as the core system of record, and connecting an AI agent through the Model Context Protocol (MCP) to safely execute database operations.

Consolidating systems into AlloyDB

Rather than deploying separate relational databases, vector databases, and log stores, Lucius AI houses all core data in AlloyDB for PostgreSQL. The relational tender catalog, document metadata, audit logs, and vector embeddings reside in the same database engine. Storing vector embeddings alongside relational rows avoids managing separate vector stores, establishes a unified backup schedule, and centralizes identity management.

Authentication relies strictly on Cloud IAM. Services connect using dedicated Google Cloud service accounts mapped to database roles scoped to specific access requirements, without storing database passwords in application environments. Database reliability is managed natively by AlloyDB through automated backups and point-in-time recovery, avoiding custom disaster recovery procedures.

In production, this consolidated architecture supports:

More than 210,000 tenders in the catalog, with embeddings stored directly alongside them
Rebuilding the semantic index embedded 115,820 records in 10.6 minutes with the Gemini embedding model, for around three dollars in API spend; AlloyDB auto embeddings now keep those vectors current.
Retrieval reranking executed directly inside the database using the ai.rank function — with mean latency of 77-milliseconds - returning the most relevant results for search queries without requiring a standalone reranking microservice

Accelerating semantic search by 47x

Semantic search across the tender catalog initially relied on unindexed vector comparisons, where a representative query took 1.14 seconds. Migrating this workload to a ScaNN index in AlloyDB reduced query latency to 24 milliseconds — a 47x improvement.

The index recommendation originated from the AI agent during an automated performance audit, where it benchmarked the query plan before preparing the index migration.

Automating database operations with MCP

To delegate routine administrative tasks, Lucius AI configured the open-source MCP Toolbox for Databases using the prebuilt alloydb-postgres server.

Operational delegation requires strict access controls. The agent connects using a dedicated PostgreSQL role granted SELECT across the schema and UPDATE on a single operational table. Destructive commands (DROP, DELETE, TRUNCATE) are omitted, restricting agent actions to authorized operational boundaries.

Under this configuration, the AI agent performs regular database operations across four key areas:

On-demand analytics: Compiles retention cohorts, activation funnels, and catalog coverage by country via ad hoc SQL queries, removing the need to build and maintain manual dashboards or complex analytical pipelines.
Performance optimization: Performs query-plan inspections and index analysis, such as identifying the ScaNN indexing strategy.
Incident forensics: In response to an external security probe, the agent parsed audit logs to reconstruct the request timeline in minutes, verifying that tenant isolation remained intact.
Automated data-quality checks: Evaluates ingestion watermarks and freshness across all thirteen procurement sources every morning.

For teams adopting this architecture, establishing a progressive permission structure provides clear guardrails: start with read-only access, expand permissions as requirements dictate, and keep destructive operations restricted to human administrators.

Looking ahead

Lucius AI is planning three technical initiatives to further reduce operational overhead:

Automated vector embeddings in AlloyDB AI: After validating ai.initialize_embeddings across the full catalog, a weekly maintenance job uses ai.refresh_embeddings to update vectors.
Columnar engine acceleration: Having enabled AlloyDB’s columnar engine with auto-columnarization, the database identified and stored 40 frequently queried columns across four tables in memory within a day, accelerating reporting queries without a separate analytical store.
Managed Remote MCP Server: Transitioning from self-hosted Toolbox processes to Google Cloud's fully managed Remote MCP Server for AlloyDB will offload MCP server hosting and maintenance.

By anchoring core data in AlloyDB and managing routine operations through MCP, Lucius AI demonstrates how a single engineer can build and operate a resilient, multi-region procurement platform.

To explore Lucius AI, visit ailucius.com. To evaluate AlloyDB for PostgreSQL, deploy an AlloyDB cluster to test performance against your own workloads.

Reimagining work: How Pythian’s internal AI playbook delivers customer ROI

Thu, 27 Aug 2026 16:00:00 +0000

When Pythian rolled out Google Cloud’s Gemini Enterprise across our 500-person company in 27 countries, the goal was simple: use our own company as a proving ground to discover how enterprise AI actually delivers ROI.

What we found changed our strategy entirely.

Since the rollout of Gemini Enterprise and our previous enterprise AI deployments, Pythian observed firsthand why so many enterprise AI initiatives stall out or fail.

Most organizations trap themselves in a tool-centric mindset — buying licenses, making tools broadly available, and assuming value will naturally follow. They get stuck chasing "nickel and dime" micro-efficiencies (like saving 5 minutes per user) while missing structural, high-ROI workflow transformations. Compounding the problem, even when custom agents are built, they frequently stall in pilot mode or break down in production because teams lack the operational capability to manage AI model drift, agent lifecycles, and ongoing observability.

To solve this, we engineered the Pythian AI Operating Model — a multifaceted, end-to-end framework designed to take enterprise AI from high-level strategy all the way into sustained production. While our dual center of excellence (COE) serves as the core execution muscle, it is the application of the entire framework, from Field CTO strategy and tooling deployment to the dual COE and XOps, that consistently unlocks million-dollar outcomes.

By proving this complete model internally first, Pythian drove a 3x surge in active user engagement and cut our database incident resolution times by 80%.

The four pillars of the Pythian AI operating model

To move past the common failure points of enterprise AI, our framework consolidates strategy, execution, and operations into a single continuous loop:

Field CTO strategy ──> tooling deployment ──> dual COE execution ──> production XOps

Field CTO strategy and governance: Generative AI is arguably the most academically challenging architectural shift in IT history. Led by former C-suite tech leaders, our Field CTO practice provides executive advisory to establish steering committees and clear value metrics. The team audits operations using 16 horizontal agentic patterns (like automated document processing and runbook creation) to build a prioritized backlog of high-ROI use cases before development starts.
Tooling and platform deployment: The team establishes a secure, production-grade foundation on platforms like Gemini Enterprise and connects AI directly into CRMs, ERPs, and database estates to ground models in real corporate context.
The dualCOE: This execution muscle is split into two specialized engines:

People productivity COE: This group handles adoption and change management. Instead of expecting non-technical teams (like HR or Procurement) to build its own agents, this COE builds no-code agents for them, focusing entirely on enablement.
Process productivity COE: This team engineers deep, custom-coded AI agents and complex agentic workflows that integrate into core data platforms for autonomous operations.

XOps (AI production management): While deploying an agent is 20% of the journey, maintaining accuracy in production is 80%. Because AI models and prompt structures naturally drift over time, this XOps practice provides the continuous monitoring, prompt tuning, and model observability needed to keep agents performing without breaking core workflows.

The difference between chasing minor, scattered efficiencies and driving structural enterprise ROI comes down to how you align your operating strategy:

Alignment element	Tool-centric approach	Pythian AI operating model
Primary metric	Individual minutes saved per user	High-impact workflow reimagination and ROI
Operational focus	Broad, unguided tool availability	Prioritized backlog via 16 agentic patterns
Execution muscle	Ad-hoc user experimentation	Dual COE (people and process productivity)
Production lifecycle	Unmonitored static deployments	Active XOps (Continuous accuracy and drift management)

Real-world impact: from database ops to global supply chains

Whether managing 70 manufacturing plants or 30,000 enterprise databases, AI succeeds when tied to structural, high-value workflows:

Pythian “as a customer:” Across 15,000 monthly database tickets, our Process COE deployed an agentic workflow that reads tickets, searches knowledge bases, and auto-generates mini runbooks before an engineer touches them. The result was slashed mean time to resolution by 80% and tripled active user engagement.
Knowledge management customer: We deployed autonomous IT support agents across 10,000 consultants. As a result, we were able to automate 10% of 20,000 annual IT tickets into "no-touch" resolutions, saving 1,000,000+ operational hours.
Supply chain customer: By building custom agentic supply chain tools on Gemini Enterprise, we compressed forecast-matching cycles from weeks down to 2–3 days across 70 global manufacturing sites.
Retail customer: We combined Gemini Agentic AI and computer vision to automate store product onboarding. As a result, we transformed a 20-minute manual task into a multi-second flow.

Ready to build your AI operating model?

Scaling AI demands more than tool-level experimentation. It also requires an end-to-end AI operating model. Learn how Pythian pairs with Google Cloud to operationalize strategy, streamline XOps, and fast-track your Gemini Enterprise journey.

10 questions every startup should answer before moving to production with their AI prototype

Thu, 20 Aug 2026 16:00:00 +0000

It’s never been easier to start an AI-powered startup on Google Cloud.

You grab an API key from Google AI Studio at breakfast, paste it into Antigravity, and by lunch you’ll have a nascent prototype of your product.

But it’s not all one straight line to progress. It's common to bump into these three challenges as you build out your stack:

A leaked API key racks up a large bill in 48 hours.
A "quick" migration from AI Studio to Gemini Enterprise Agent Platform stalls the roadmap for weeks because nobody on the team owns Identity and Access Management (IAM).
The launch works, until the app starts returning HTTP 429 Too Many Requests because of default per-project quotas, and there's no clean path to more capacity without paying a premium.

None of these are unique edge cases. . They're default failure modes of moving fast without a plan, and we've all done it at least once.

Below are the 10 questions every startup should be ready to answer before they scale, grouped into the three phases where decisions can shape your future:

Onboard (setting up your own projects and identities right)
Scale (getting more throughput without breaking the bank)
Govern (keeping costs, keys, and agents from running away).

These ten are scoped to the prototype-to-production transition itself. Each question ends with a short, runnable snippet you can copy into your own project today. Adjacent decisions that matter just as much but aren't specific to that move, your data layer and RAG architecture, CI/CD, network design, are deliberately out of frame here.

Onboard: get the foundation right (in the first hour).

#1 Where should I start: Google AI Studio or Gemini Enterprise Agent Platform?

Both surfaces expose the same Gemini family of models, but they solve different problems.

Google AI Studio (with the Gemini Developer API) is the fastest path from an idea to working code. A browser IDE, an API key, a generous free tier, and no cloud project to configure. It's where most ideas should start, and Google's own guidance says as much.
Gemini Enterprise Agent Platform (formerly Vertex AI) has the same Gemini models (plus 3rd party and OSS ones) with enterprise controls around them: IAM and service-account auth instead of raw keys, VPC Service Controls, Cloud Logging and Monitoring, reserved capacity, regional endpoints, and the compliance surface your first enterprise customer's security review will ask about.

The right answer for most startups is both, sequenced deliberately: first prototype in AI Studio, then migrate before you have real users. The danger for startups is treating them as interchangeable solutions, AI Studio's simple key model does not translate to enterprise controls, and Agent Platform's IAM model might look like overkill until the day it saves you from a stolen-credential incident.

It's less work than it sounds like.

The unified google-genai SDK targets both:

code_block: <ListValue: [StructValue([('code', '# Prototype: Google AI Studio, raw API key\r\nfrom google import genai\r\nclient = genai.Client(api_key="YOUR_AI_STUDIO_KEY")\r\n\r\n# Production: GEAP, no key — uses Application Default Credentials (ADC)\r\nfrom google import genai\r\nclient = genai.Client(\r\n vertexai=True,\r\n project="my-startup-prod",\r\n location="us-central1",\r\n)\r\n\r\nresp = client.models.generate_content(\r\n model="gemini-2.5-pro",\r\n contents="Summarize this contract in three bullets.",\r\n)\r\nprint(resp.text)'), ('language', ''), ('caption', <wagtail.rich_text.RichText object at 0x7f4c39f95d10>)])]>

#2 How do I set up a Google Cloud project without becoming an IAM expert?

The biggest reason startups stall on the migration to Agent Platform isn't the code, it's the operational leap from "here's an API key" to a cloud project with folders, service accounts, org policies, logging, and IAM bindings. If your team doesn't have a dedicated cloud admin, that first project setup can eat a week of engineering time.

Three moves cut that dramatically:

Use an opinionated project template instead of clicking through the console. The Cloud Setup checklist and the Google Cloud Architecture Framework give you a production-grade folder hierarchy (prod / non-prod / dev), a central logging + monitoring project, Security Command Center turned on, and baseline org policies, without you having to design them from scratch.
Enable the APIs you'll actually use, once. Batch it so you're not doing it project-by-project when you need it. The billing-link step is not optional. Every paid API you're about to enable will refuse to activate on a project with no billing account attached, so we handle that first.
Let Gemini pick the roles, but ask it for the narrow ones. You don't have to memorize the roles reference. In the Grant access dialog, Help me choose roles lets you describe the task in plain language, "this service account needs to call Gemini models and read one Cloud Storage bucket", and get predefined roles back with the reasoning shown. One catch worth knowing on day one: by default it suggests roles that cover common journeys, which usually means a service's Admin, Editor, or Viewer. Those are broader than you want. Say "least privileged" or "narrowest access" in the prompt and it returns granular roles instead. Same amount of typing, considerably smaller blast radius when a credential leaks.

Sources: Get predefined role suggestions with Gemini assistance

code_block: <ListValue: [StructValue([('code', '# One-shot: create a Vertex-ready project and turn on the services a\r\n# typical AI startup uses.\r\ngcloud projects create my-startup-prod --name="My Startup (prod)"\r\ngcloud config set project my-startup-prod\r\n\r\n# REQUIRED before enabling billing-dependent APIs (aiplatform, run, etc.).\r\n# Use `gcloud billing accounts list` to find your billing account ID.\r\ngcloud billing projects link my-startup-prod --billing-account=012345-6789AB-CDEF01\r\n\r\ngcloud services enable \\\r\n aiplatform.googleapis.com \\\r\n run.googleapis.com \\\r\n artifactregistry.googleapis.com \\\r\n logging.googleapis.com \\\r\n monitoring.googleapis.com \\\r\n secretmanager.googleapis.com \\\r\n cloudbilling.googleapis.com'), ('language', ''), ('caption', <wagtail.rich_text.RichText object at 0x7f4c3968b750>)])]>

Sources: gcloud services enable reference, · gcloud billing projects link (GA), GE Agent Platform environment setup.

If you're a solo founder, resist the urge to build in your personal GCP account. Create a proper organization or self-owned org first, then create the project inside it. That single decision can make everything else, fromIAM to billing and audit, dramatically easier.

#3 I'm on Google Cloud, how should my code actually authenticate: API keys, service accounts, or user credentials?

There's a hierarchy of safety here, and the easiest option is rarely the right one in production.

Raw API keys are fine for local prototyping. They are dangerous in production because they are long-lived, easy to leak into a client bundle or a public repo, and grant unbounded access until you notice.
User credentials via OAuth (application default credentials) are best for interactive tools, CLIs, and any code that runs on a developer's laptop.
Service accounts with least-privilege IAM roles are the right answer for anything running on a server, in a container, or in a scheduled job.

The pattern you're aiming for is one where your code never sees a key at all. It just calls the Google Auth library, which quietly reads Application Default Credentials (ADC) from the environment, a short-lived token minted for whichever service account is attached to your Cloud Run service, GKE workload, or Compute Engine VM. You get enterprise-grade auth without writing any auth code.

code_block: <ListValue: [StructValue([('code', '# On a developer laptop\r\ngcloud auth application-default login\r\n\r\n# On a server (Cloud Run, GKE, etc.) — no login, no key file.\r\n# Attach a service account with just the roles the app needs.\r\ngcloud run deploy my-agent \\\r\n --image=us-docker.pkg.dev/my-startup-prod/agents/api:v1 \\\r\n --service-account=agent-runtime@my-startup-prod.iam.gserviceaccount.com \\\r\n --region=us-central1'), ('language', ''), ('caption', <wagtail.rich_text.RichText object at 0x7f4c39688a90>)])]>

code_block: <ListValue: [StructValue([('code', '# Application code — notice: no keys, no secrets.\r\nfrom google import genai\r\n\r\nclient = genai.Client(\r\n vertexai=True,\r\n project="my-startup-prod",\r\n location="us-central1",\r\n)'), ('language', ''), ('caption', <wagtail.rich_text.RichText object at 0x7f4c3968b590>)])]>

Do one last favor to your future self: give that service account the minimum IAM role your workload actually needs, usually roles/aiplatform.user for calling models, not the broader admin roles. It takes an extra 30 seconds and prevents the credential from becoming a master key if it leaks.

#4 When should I actually stop procrastinating and migrate from AI Studio's API key to Agent Platform's IAM model?

Sooner than you'd like, and the correct trigger is not when it breaks. It's when any of these is true:

Your key has left your laptop (checked into a repo, pasted into a Slack, shipped in a mobile app).
You have more than one person on the team who needs to call the API.
You're spending more than a few hundred dollars a month.
You're about to onboard paying customers.

A potential pitfall that can catch growing startups off guard is simple: a leaked Gemini API key on an account that normally spends $180 a month gets scraped from a public repo and used to run distillation attacks, accumulating tens of thousands of dollars in charges before the owner even sees the first billing alert. The Google Cloud Shared Responsibility Model is unambiguous: the customer is liable for charges incurred with their own valid credentials.

The migration itself is genuinely smaller than the anxiety around it. In google-genai it's the two-line change shown in #1. What takes real time is the project setup around it, which is exactly why #2 exists.

Practical checklist for cutover day:

code_block: <ListValue: [StructValue([('code', '# 1. Revoke every existing AI Studio key that has ever left a laptop.\r\n# (Go to https://aistudio.google.com/apikey and delete them.)\r\n\r\n# 2. Confirm your production code has no api_key= arguments.\r\ngrep -rn "api_key" src/\r\n\r\n# 3. Enable GEAP and confirm ADC works locally.\r\ngcloud services enable aiplatform.googleapis.com\r\ngcloud auth application-default login\r\npython -c "\r\nfrom google import genai\r\nc = genai.Client(vertexai=True, project=\'my-startup-prod\', location=\'us-central1\')\r\nprint(c.models.generate_content(model=\'gemini-2.5-flash\', contents=\'ping\').text)\r\n"'), ('language', ''), ('caption', <wagtail.rich_text.RichText object at 0x7f4c39689750>)])]>

If step 3 prints a response, you're on Agent Platform.

Scale: get more capacity without paying a premium.

#5 Now that I'm shipping, why on earth am I getting all these HTTP 429 errors, and how do I make them stop?

429 Too Many Requests from Agent Platform almost always means one of two things:

You've hit the Dynamic Shared Quota (DSQ) ceiling for your project's tier. DSQ is a shared pool sized against your project's history, new projects start with modest limits by design, to prevent abuse across the platform.
You're calling a global endpoint during a global demand spike, competing with worldwide traffic for shared capacity.

The instinctive reaction is to file a quota-increase ticket. You can do that if you must, but two architectural moves usually solve the problem faster and cheaper.

Pin to a regional endpoint. Over half of startup traffic on Agent Platform defaults to global routing. Pinning to a specific region (say us-central1) sidesteps global contention and typically improves latency at the same time. (One narrow exception, which we'll get to in the next question: if you specifically want Priority PayGo, that feature currently only ships on the `global` endpoint. For everything else, pin regionally.):

code_block: <ListValue: [StructValue([('code', 'from google import genai\r\n\r\n# Global (default): competes against worldwide demand.\r\n# Regional: routes only to the regional cluster, less contention.\r\nclient = genai.Client(\r\n vertexai=True,\r\n project="my-startup-prod",\r\n location="us-central1", # <-- this is the one-line fix\r\n)'), ('language', ''), ('caption', <wagtail.rich_text.RichText object at 0x7f4c3968ba10>)])]>

Add real retry and backoff. A 429 is a retryable signal, not a fatal error. Any production client should have exponential backoff with jitter. The modern google-genai SDK ships this behavior built in, but only if you actually enable it. This is easy to overlook. Don't reach for the classic `google.api_core.retry.if_transient_error` decorator you may have seen on older Vertex code. It's designed for the legacy exception classes and does not recognize the new `google.genai.errors.APIError, so it will silently pass 429s through without retrying. Use the SDK's built-in retry options instead:

code_block: <ListValue: [StructValue([('code', 'from google import genai\r\nfrom google.genai import types\r\n\r\nclient = genai.Client(\r\n vertexai=True, project="my-startup-prod", location="us-central1",\r\n http_options=types.HttpOptions(retry_options=types.HttpRetryOptions(\r\n attempts=5, initial_delay=1.0, max_delay=60.0, exp_base=2.0, jitter=1.0,\r\n http_status_codes=[408, 429, 500, 502, 503, 504],\r\n ))\r\n)'), ('language', ''), ('caption', <wagtail.rich_text.RichText object at 0x7f4c39688810>)])]>

How do you see this coming? Preferably not from a user telling you. Agent Platform publishes serving metrics to Cloud Monitoring, and there is a prebuilt dashboard you don't have to assemble: Console → Agent Platform → Dashboard → Model observability. It gives you requests per second, token throughput, first-token latency, and error rates out of the box.

The metric to actually alert on is aiplatform.googleapis.com/publisher/online_serving/model_invocation_count. It carries an error_category label with values of user, system, or capacity. Alerting on capacity isolates genuine throttling from your own bad requests, which a raw 429 count won't do.

One thing worth internalizing, because it trips people up: you cannot build a "warn me at 80% of my quota" alert for Standard PayGo. Under Dynamic Shared Quota there is no fixed per-project number to be at 80% of. A 429 means transient contention for shared capacity, not that you crossed a line. Percent-of-limit alerting only becomes meaningful once you're on Provisioned Throughput, which does expose real limit metrics.

code_block: <ListValue: [StructValue([('code', 'gcloud monitoring policies create --policy-from-file=capacity-alert.yaml'), ('language', ''), ('caption', <wagtail.rich_text.RichText object at 0x7f4c3968a650>)])]>

Sources: Agent Platform metrics list, Model observability dashboard, RetryOptions source, core retry_base.py, genai errors.py, reduce 429 errors, gcloud monitoring policies create, Dynamic Shared Quota.

Follow the Agent Platform rate limits documentation to understand what your project's current ceiling actually is before you assume you've outgrown it.

#6 Which consumption mode do I pay for: Standard PayGo, Priority PayGo, or Provisioned Throughput?

Three consumption models, three completely different workload shapes, and three completely different ways to proceed. Picking the right one can help startups see meaningful savings on AI bills. First let’s define them and then see when they are, or aren’t, a good fit:

Standard PayGo (DSQ): Pay per token from a shared pool; cheap, no guarantees.
Priority PayGo: Pay per token at a premium to jump the queue.
Provisioned Throughput (PT): Prepay for reserved capacity; predictable, use it or lose it.

Consumption type	Best for	Watch out for
Standard PayGo (DSQ)	Early-stage, low-QPS, spiky prototype traffic	429s during spikes; no reliability SLO
Priority PayGo	Bursty, revenue-critical traffic that can't tolerate 429s	Roughly 1.8x the standard token price
Provisioned Throughput (PT)	Steady, predictable, high-volume production traffic	Wasted spend if utilization is under ~40%; overflow to PayGo on spikes

The dominant startup mistake is buying PT too early. Usually this happens the week after a big launch when it feels like traffic will only ever go up. PT is reserved capacity. You pay whether you use it or not, and it only starts paying you back once your baseline is genuinely predictable, not just aspirational.

Here’s a pragmatic sequence:

Weeks one through four on Standard PayGo. Use it to measure your real request shape (tokens per minute at p50 and p99, request bursts, batchable vs. real-time split).
When you get your first bad 429 storm, flip on Priority PayGo for the traffic that actually matters. It's a config change, not a purchase order, nobody in procurement needs to be involved:

code_block: <ListValue: [StructValue([('code', '# Priority PayGo request: use the global endpoint + two extra headers.\r\nfrom google import genai\r\nfrom google.genai import types\r\n\r\nclient = genai.Client(vertexai=True, project="my-startup-prod", location="global")\r\nresp = client.models.generate_content(\r\n model="gemini-2.5-pro",\r\n contents="Rank these support tickets by urgency: ...",\r\n config=types.GenerateContentConfig(\r\n # Priority PayGo headers, per current GEAP docs.\r\n http_options=types.HttpOptions(headers={"X-Vertex-AI-LLM-Request-Type": "shared", "X-Vertex-AI-LLM-Shared-Request-Type": "priority"}),\r\n ),\r\n)'), ('language', ''), ('caption', <wagtail.rich_text.RichText object at 0x7f4c3968b010>)])]>

3. Once you can predict your baseline TPM, buy PT to cover the flat baseline and let anything above it overflow to PayGo. That's the combined pattern Google recommends for exactly this reason. Best of both worlds, not marketing spin.

Sources: Priority PayGo docs, google-genai HttpOptions source, GEAP REST reference.

#7 Which of my requests actually need to be live, and which should be batch jobs?

Most startup workloads are secretly batch jobs pretending to be real-time. Every one you move off the interactive path frees up DSQ headroom for the traffic that genuinely needs to be fast, the traffic where a user is actually watching a spinner.

Three questions to help you sort your traffic:

Does a human have to see the result within a second? That means: Live inference.
Can the user wait a few seconds and see a spinner? That means: Still live, but a candidate for streaming.
Would the user tolerate "we'll email you when it's ready" or "check back in a bit"? That means: Batch prediction.

Batch prediction on Agent Platform runs in a completely separate queue, does not consume your interactive DSQ, and is typically about half the price of on-demand inference. That's a rare double win: faster live traffic and a lower bill.

code_block: <ListValue: [StructValue([('code', '# Kick off a batch prediction job from a JSONL file in Cloud Storage.\r\n# Each line is one prompt; results land in another Cloud Storage prefix.\r\nfrom google import genai\r\nfrom google.genai import types\r\n\r\nclient = genai.Client(vertexai=True, project="my-startup-prod", location="us-central1")\r\n\r\njob = client.batches.create(\r\n model="gemini-2.5-flash",\r\n src="gs://my-startup-prod-batch/inputs/nightly-summaries.jsonl",\r\n config=types.CreateBatchJobConfig(\r\n dest="gs://my-startup-prod-batch/outputs/",\r\n ),\r\n)\r\nprint(job.name, job.state)'), ('language', ''), ('caption', <wagtail.rich_text.RichText object at 0x7f4c39689610>)])]>

Common candidates: nightly document summarization, background classification of new signups, bulk translation, embedding backfills, evaluation runs against your test set. If any of those are on your live path today, moving them is often the single highest-leverage change you can make this week.

Govern: Keep costs, keys, and agents under control.

#8 How do I set spend caps that actually reduce cost, and not just send me polite emails while my bill triples?

Until recently the honest answer was that budgets only notify, and you had to build your own brake pedal. That changed in July. There are now three mechanisms, and you should think of them as layers.

A spend cap budget (Preview). Cloud Billing budgets can now enforce rather than just email. Set a spend cap on a project and, when usage costs cross 100% of the budget, Google pauses the service until you manually lift it. Agent Platform is explicitly on the eligible list, alongside the Gemini API, Cloud Run, and Cloud Run functions. Alerts still fire at 50% and 80%, so the pause isn't a surprise.

Three things to know before you rely on it:

Each cap covers one project and one eligible service. It is not account-wide protection. If you want Agent Platform and Cloud Run both capped, that's two caps.

Enforcement is not instant and is based on estimated costs. Overages past the cap are billed as normal, so set the number below your real ceiling. Lifting it is manual, and service resumption can take up to an hour. It also pauses Provisioned Throughput usage, so if you've prepaid for capacity, a cap hit stops that too.

It's in Preview as of publication, and the eligible-service list is documented as growing. Check the current list before you design around it.

2. A billing budget with a Pub/Sub trigger that disables billing. Still the right tool when you need blast radius the spend cap can't give you: multiple services at once, an entire project, or a service that isn't eligible yet. When the budget hits a threshold, Pub/Sub fires a Cloud Function that detaches the billing account, which stops all billable activity within minutes. Blunter and more dangerous than the native cap — it can leave resources unrecoverable — so reach for it second, not first. Full walkthrough: Automatically respond to budget notifications.

code_block: <ListValue: [StructValue([('code', '# Sketch: create a budget SCOPED TO ONE PROJECT that publishes to Pub/Sub at 50%, 90%, 100%.\r\ngcloud billing budgets create \\\r\n --billing-account=012345-6789AB-CDEF01 \\\r\n --display-name="my-startup-prod hard stop" \\\r\n --budget-amount=2000USD \\\r\n --filter-projects=projects/my-startup-prod \\\r\n --threshold-rule=percent=0.5 \\\r\n --threshold-rule=percent=0.9 \\\r\n --threshold-rule=percent=1.0,basis=current-spend \\\r\n --notifications-rule-pubsub-topic=projects/my-startup-prod/topics/budget-alerts'), ('language', ''), ('caption', <wagtail.rich_text.RichText object at 0x7f4c3968ad90>)])]>

Sources: Manage spend cap budgets, Set up programmatic notifications gcloud billing budgets create reference, Cloud Billing budgets concepts, Disable billing with notifications walkthrough, Programmatic notification payload schema.

Two things to get ahead of for, as the defaults can cause unexpected issues:

Limit your budget scope: Without --filter-projects, your budget applies to your entire billing account. A spike in any project will trigger the kill switch for everything.
Deploy locally: The budget notification doesn't specify which project is affected. To ensure the kill switch only affects the intended project, deploy your Cloud Function in the same project you're protecting (e.g., my-startup-prod).

Then wire up a tiny Cloud Function to that topic that calls projects.updateBillingInfo to unlink the billing account when the 100% threshold fires. That is your circuit breaker.

Mechanical ceilings via quota overrides. Even if you never set up the above kill switch, you can cap the rate at which cost can accumulate by setting explicit per-model, per-region quotas below the platform default. If your app never legitimately needs more than 500 requests per minute for gemini-2.5-pro, cap it there in the Cloud Quotas console; a leaked key can't burn what the quota flatly refuses to serve.

#9 Where should I actually keep secrets? (Not in .env files!)

The short answer is: Secret Manager. Not in environment variables, not in .env files, and never in your repo. Grant read access via IAM only to the service account that needs it.

code_block: <ListValue: [StructValue([('code', '# Store a third-party API key (Stripe, OpenAI, whatever).\r\necho -n "sk_live_xxx" | gcloud secrets create stripe-live-key --data-file=-\r\n\r\n# Grant only the runtime service account access to read it.\r\ngcloud secrets add-iam-policy-binding stripe-live-key \\\r\n --member=serviceAccount:agent-runtime@my-startup-prod.iam.gserviceaccount.com \\\r\n --role=roles/secretmanager.secretAccessor'), ('language', ''), ('caption', <wagtail.rich_text.RichText object at 0x7f4c39688ed0>)])]>

code_block: <ListValue: [StructValue([('code', '# Application code fetches it at startup; nothing lives on disk.\r\nfrom google.cloud import secretmanager\r\nsm = secretmanager.SecretManagerServiceClient()\r\nresp = sm.access_secret_version(\r\n name="projects/my-startup-prod/secrets/stripe-live-key/versions/latest"\r\n)\r\nstripe_key = resp.payload.data.decode("utf-8")'), ('language', ''), ('caption', <wagtail.rich_text.RichText object at 0x7f4c39688c90>)])]>

Then two little disciplines that pay for themselves the first time you need them:

Rotation on a schedule and on suspicion. Secret Manager versions are cheap; treat them as immutable and roll forward.
Detection when a secret leaks. Secret Manager notifications and Google Cloud's Sensitive Data Protection can catch keys checked into a repo or pasted into a log stream, before an attacker does.

For any AI application that acts on a user's behalf, calls Gmail on their behalf, reads a Drive folder, hits a third-party SaaS with the user's credentials, do not store a long-lived token. Use OAuth 2.0 with short-lived access tokens and a refresh flow, so that when a user rage-quits or a compromised account gets revoked, the agent loses access at the same time.

#10 How do I stop my brand new AI agent from doing something it absolutely shouldn't?

An agent that can call tools, browse the web, or execute code needs the same defense-in-depth thinking as any other production service, arguably more, because it makes decisions that neither you nor the model can fully predict in advance.

Four layers, none optional once you have real users:

1. Identity for the agent itself. Give the agent its own service account, scoped only to the resources and tools it genuinely needs, the exact same least-privilege principle as any other workload. Agent Engine supports first-class agent identity so every action can be attributed to a specific agent instance in your audit logs.

2. Sandboxed code execution. If your agent runs generated code, a common pattern for data-analysis or "run this Python for me" flows, do not run it in your application process. Use an isolated sandbox so a bad combination can't touch your production data.

code_block: <ListValue: [StructValue([('code', '# Enable server-side code execution inside a sandbox for a request.\r\nfrom google import genai\r\nfrom google.genai import types\r\n\r\nclient = genai.Client(vertexai=True, project="my-startup-prod", location="us-central1")\r\nresp = client.models.generate_content(\r\n model="gemini-2.5-pro",\r\n contents="Compute the correlation between these two columns: ...",\r\n config=types.GenerateContentConfig(\r\n tools=[types.Tool(code_execution=types.ToolCodeExecution())],\r\n ),\r\n)'), ('language', ''), ('caption', <wagtail.rich_text.RichText object at 0x7f4c3968a750>)])]>

3. Prompt and response filtering. Model Armor sits in front of your model calls and screens for prompt injection, jailbreaks, sensitive-data exfiltration, and off-brand output, all of which are essentially guaranteed the moment you have real users being real users.

4. Behavioral monitoring. Security Command Center with threat detection flags anomalies in agent behavior, a service account suddenly calling an API it's never touched before, an agent reaching out to an unfamiliar external host, an unexpected spike in privileged operations. In near-real-time.

None of these are optional once your agent is acting on behalf of a real user or handling real money.

Your homework, so to speak:

Audit for raw API keys in your repo, your notebooks, and your production runtime. Rotate anything that shouldn't be there.
Move any workload that doesn't need a synchronous response to the Batch API.
Turn on the Model observability dashboard and put one alert on capacity errors, so the next 429 reaches you before it reaches a customer.
Set a spend cap on the project, and keep an eye out for 50% and 80% alerts. If usage crosses 100% of the budget, Google will pause the service until you manually lift it.

Do those four things this week and you're already ahead of most startups shipping AI features.

Have a scenario you'd like us to cover next? Reach us at Google Cloud for Startups.

Mirendil taps AI Hypercomputer TPUs and GPUs for pre- and post-training applications

Thu, 06 Aug 2026 13:00:00 +0000

Nearly every major AI lab uses Google Cloud infrastructure, including for training of models, inference for agents, and new frontier research. Google Cloud also continues to be the platform of choice for new, high-growth AI startups who are driving much of the industry’s research and innovation.

Today, we’re announcing that Mirendil, an exciting frontier AI lab focused on accelerating AI development, will also utilize Google Cloud’s AI Hypercomputer. This includes using a mix of Google’s TPU AI accelerators and full-stack NVIDIA AI infrastructure running on Google Cloud; this purpose-built AI infrastructure will support model pre-training and post-training applications for Mirendil.

The Mirendil team is building new AI systems that can help accelerate and democratize AI research and development. This means managing complex, end-to-end training workflows from initial model pre-training through post-training, and powering reinforcement learning on a massive scale. The ability to choose a mix of both TPU and NVIDIA’s full-stack accelerated computing platform through Google Cloud meant that Mirendil could access critical compute very quickly, and continue to match its workloads to the architecture best-suited to it over time.

We closely partnered with Mirendil on end-to-end design and deployment of combined TPU and NVIDIA AI infrastructure across compute, storage, networking, and control planes. We also collaborated on a system that uses managed training clusters running in Gemini Enterprise Agent Platform, which effectively streamlines the provisioning and management of both TPU and GPU environments for Mirendil. Mirendil is already live with a cluster of TPU v5P chips, with NVIDIA AI accelerated computing systems coming online soon.

"Progress in AI has been bounded by how fast humans can run the research loop - designing experiments, evaluating results, and iterating," said Behnam Neyshabur, cofounder and CEO of Mirendil. "We're building AI systems that can accelerate and improve that loop itself. Expanding on Google Cloud gives us the scale and flexibility to push those systems further and put frontier AI research capabilities in the hands of many more scientists and engineers to run that loop faster and at a greater scale."

You can read more about our partnership on Mirendil’s blog.

Meet the 33 cybersecurity startups joining the Gemini Startup Forum

Wed, 08 Jul 2026 13:30:00 +0000

Startups are at the forefront of tackling some of the world’s most complex challenges, especially in cybersecurity, where new ideas and adaptability are always needed. These companies are embracing AI as a powerful tool, enabling them to scale their impact.

Google believes that enabling the next generation of AI-native cybersecurity startups will have a positive impact across the globe. Today, we are thrilled to announce that our flagship Google for Startups program, Gemini Startup Forum: Cybersecurity, has selected its first 33 trailblazing startups. This exclusive forum is designed to address critical domains and foster deep dialogue on AI integration in cybersecurity.

Every startup will be able to work on their roadmap alongside AI and cybersecurity specialists from Google DeepMind, Google Cloud, and Wiz. This year’s cohort is organized into six specialized focus areas, from autonomous agent protection to post-quantum cryptography.

AI agent security and governance

Capsule Security (Israel): Delivers runtime protection and behavioral monitoring for autonomous AI agents. The platform tracks agent activities, API calls, and tool usage in real-time to prevent unauthorized actions and data exfiltration across software-as-a-service (SaaS) and endpoint agents.
Evoke Security (United States): Functions as an endpoint detection and response (EDR) platform designed for AI agents. It deploys an endpoint sensor to scan for risky agent configurations, monitor runtime behavior, and block unauthorized skill executions.
Manifold Security (United States): Provides agentless detection and response for AI agents using graph analysis and runtime monitoring. By tapping into open telemetry and API hooks, the platform maps agent behavior and connections to identify anomalies without installing software on local machines.
Mirror Security (Ireland): Specializes in data-in-use protection for AI workloads using fully homomorphic encryption. The platform enables enterprises to perform model inference, database searches, and agent communications directly on encrypted data without decrypting it.
Onyx Security (Israel): Offers an AI control plane to discover, govern, and monitor autonomous AI agents across enterprise environments. The platform analyzes agent posture, tracks session token streams, and uses a model mesh to enforce context-aware compliance policies.
Refractal (United Kingdom): Builds runtime security and governance, risk, and compliance (GRC) infrastructure to monitor and govern enterprise AI agents. It intercepts agent actions, evaluates them against organizational policies and European regulations, and generates cryptographic audit logs.
Unbound Security (United States): Acts as an agent access security broker (AASB) that provides visibility and governance for developer-focused coding agents. It deploys endpoint hooks to monitor terminal commands and restricts agent actions based on user identity and group permissions.
XOR (United States): Produces reinforcement learning training data and environments to help AI models autonomously find and fix their own security flaws. The company provides verified task trajectories and benchmarks that developers use to train secure coding agents.

Application security and vulnerability management

Aisy (United Kingdom): Models enterprise environments from an attacker's perspective to prioritize vulnerability remediation. The platform maps external attack surfaces, groups related assets into business-centered threat models, and identifies critical exploit chains.
Alt Security (Israel): Develops an agentic penetration testing platform that uses autonomous AI agents to perform security testing. The system automates reconnaissance, chains vulnerability findings to identify critical business risks, and validates exploits in sandboxed environments.
Arcjet (United States): Provides application security that executes natively inside developer codebases via an SDK. The platform handles bot detection, rate limiting, and prompt injection directly in the codebase, maintaining a low-latency decision loop.
Pixee (United States): Automates vulnerability triage and code remediation by converting scanner results into verified pull requests. The platform uses a context graph and a deterministic harness to generate codebase-compatible fixes that pass developer continuous integration (CI) pipelines.

Cloud, network, and infrastructure security

CloudFence (United States): Monitors a company's cloud network like an invisible overhead drone. It studies the normal patterns of how systems talk to each other and immediately alerts managers if a system starts talking to an unfamiliar location.
CyberSeQ (United Kingdom): Scans software code and digital pipelines to check if their encryption is outdated. It helps organizations plan their transition to modern, quantum-proof security keys.
Huskeys (Israel): Works as a smart tuner for a company's digital firewall. It automatically optimizes and adapts traffic filters in real-time, reducing false alarms so legitimate customers can visit websites without being blocked.
Native (Israel): Serves as a multicloud security control plane that manages built-in native provider controls across Google Cloud, AWS, Azure, and Oracle. The platform uses a simulation engine to preview the operational impact of security policy changes before deployment.
Prowler (United States): Automates cloud security posture management and compliance checks across multicloud environments. The open-source platform uses a database of community-driven security controls to identify and remediate misconfigurations.
QIZ Security (Israel): Manages enterprise cryptographic assets and compliance posture through a centralized dashboard. The platform maps certificates, databases, and source code into a knowledge graph to help organizations transition to post-quantum cryptography.
Tracebit (United Kingdom): Uses cloud-native decoys, or canaries, to detect infrastructure intrusions and unauthorized access. The platform deploys these decoy resources at scale via infrastructure-as-code to trigger alerts the moment an attacker moves.

Endpoint security and data protection

Bold Security (United States): Provides a user-space endpoint security agent that protects enterprise devices from data exfiltration and insider threats. The agent runs local classification models on-device, allowing for policy enforcement and data loss prevention without cloud latency.
Glow (Israel): Unifies endpoint, software, and AI agent monitoring into a single workspace protection layer. By using a team of collaborative AI agents, the platform maps organization-wide software footprints and blocks unauthorized browser extensions and plugins.
Jazz (United States): Modernizes data loss prevention (DLP) by analyzing user intent and business context rather than relying on static pattern rules. It uses a lightweight endpoint agent and an intelligent investigator to automatically triage and resolve data-flow alerts.
ORION Security (United States): Deploys an indicator-driven data loss prevention platform that tracks data lineage across endpoints, browsers, and SaaS tools. By mapping file movements in a graph database, the platform identifies exfiltration risks without requiring manual policies.
MokN (France): Combats credential theft through a proactive identity recovery platform. Its proprietary technology turns the tables on attackers by using ultra-realistic decoy access points to trick them into revealing the credentials they have stolen. This can help neutralize threats before the compromised credentials can be exploited.

SOC automation and offensive security

COGNNA (Saudi Arabia): Operates an agentic security operations center (SOC) platform that analyzes security alerts and automates threat response workflows. The platform ingests telemetry from existing endpoint and security information and event management (SIEM) tools, using AI agents to investigate incidents and reduce alert noise.
Latent Defense (United States): Builds foundational world models for cybersecurity to represent complex system architectures as dense, multi-dimensional graphs. The platform uses these graph representations to identify exploitable attack paths and test them with automated red-teaming agents.
Mate Security (Israel): Investigates security alerts at scale using an AI-native security operations platform. It integrates with existing SIEM and security orchestration, automation, and response (SOAR) tools to map asset relationships, automatically triage incoming alerts, and minimize false-positive rates.
Nrdsnipe (Sweden): Builds an AI-driven security testing platform, Hedgehog, that automates internal network penetration testing. Deployed directly in customer networks, it uses planner and terminal agents to map assets and construct exposure-based knowledge graphs.
RIFFSEC (Poland): Delivers an early-warning threat intelligence and attack surface management platform tailored for the central European market. It monitors the dark web, Telegram channels, and code repositories to identify leaks, exposed credentials, and phishing campaigns.
TandemTrace (Spain): Augments security operations teams with autonomous AI agents that analyze raw telemetry to investigate alerts and hunt threats. The platform connects directly to existing data lakes and SIEM APIs to build human-readable context around security incidents.

Security infrastructure and specialized services

Netsec (France): Manages the entire IT and cybersecurity lifecycle for mid-sized organizations through an all-in-one platform delivered directly inside collaboration channels. It automates user onboarding, device management, and SaaS posture remediation from a single control plane.
Revelum (United States): Detects and dismantles deepfake-driven fraud and impersonation campaigns at scale. The platform uses vision models and classification engines to analyze social media advertisements, verify biometric identities, and automate domain takedowns.
Synqly (United States): Provides a unified connectivity platform and API management control plane that simplifies integrations between IT and cybersecurity tools. It normalizes data schemas and acts as a deterministic middleware layer to facilitate bidirectional data sharing.

By collaborating with global specialists, these startups are well-positioned to build a safer, more resilient digital infrastructure. The Gemini Startup Forum is a benefit of the Google for Startups Gemini Kit, packed with APIs, tools, training and technical resources to help startups scale with AI. You can learn more about the kit here.

The threat landscape is always evolving. To meet this challenge, we will continue our commitment to Google for Startups. Over the past four years, we’ve supported more than 50 cybersecurity founders, including Authologic, BforeAI, Build38, Cerby, Crowdsec, and Risk Ledger.

This year's cohort reflects the industry's transition from perimeter defense to autonomous AI agent security — emphasizing enterprise governance and the safe deployment of self-governing AI. Looking ahead, the integration of these technologies will help define the next generation of proactive digital defense.

How growing UK midsize businesses are building in the AI era

Wed, 17 Jun 2026 08:00:00 +0000

The UK’s 5-million-plus small and midsize businesses and enterprises (SMBs) are the backbone of our economy. Today, we’re seeing these critical businesses begin to put AI to work, to operate more efficiently, move faster, and ultimately deliver better outcomes for their customers.

This shift is driven by tangible day-to-day results. According to recent research from Enterprise Nation published in partnership with Google, 71% of AI adopters surveyed in the UK say the technology helps them save time on routine tasks, and 64% report a direct boost in productivity. On top of this, AI-enabled productivity tools (like Google Workspace with Gemini) are delivering a 20% boost in productivity for SMBs, which effectively hands them back one full working day every single week.

At Google Cloud, we have a front row seat to this shift: SMBs have long utilized platforms like Google Workspace, and today they’re transforming with Google’s AI platform and models. In fact, we’ve seen the number of UK-based SMBs using Google Cloud AI nearly double year-over-year. This includes our Gemini models and products like Gemini Enterprise and AI Studio, which are helping SMBs do things like:

Roll out better customer support systems to help escalate and resolve customer support calls more quickly.
Automate repetitive actions in areas like payroll and accounting.
Help more employees understand and leverage data at work — even those not trained as data analysts.
Rapidly create and implement new designs for marketing collateral.
Help more people build their own AI agents to help them in their everyday jobs.
Conduct complex research projects at a speed and price point previously unavailable.

At today’s Google Cloud London Summit, we’re showcasing a number of innovative SMB customers who are actively using our AI tools to transform how they work, including companies who have recently expanded their work with us:

Neural Alpha, a sustainability fintech company, is using Gemini models to read unstructured environmental and corporate sustainability reports to automatically find and organize thousands of key facts, cutting months of slow, manual research down to a fraction of the time.
Sep 2, a digital security provider, uses Gemini Enterprise to deploy autonomous AI agents for 24/7 threat monitoring — accelerating incident detection and quickly neutralizing security threats for its customers.
Sunhouse, a strategic brand design agency, uses Gemini Enterprise to easily find archived design work stored on Google Drive, enabling its teams to spend less time hunting for files and more time growing its business with global brands.
Terrapinn, a global B2B events company, is transforming its operations by leveraging Gemini models, NotebookLM, Looker, and BigQuery to turn manual tasks into automated workflows, accelerating how its teams design, market, and deliver world-class conferences.
VoCoVo, a telecommunications provider, is integrating Google Cloud AI across its systems to turn isolated data into actionable intelligence and build autonomous workflows, streamlining routine operations so their team can focus on high-impact innovation.

Empowering Your Team: AI Upskilling Resources for Growing British Businesses

To help midsize teams maximize their impact and confidently navigate the modern AI landscape, we’ve developed a suite of dedicated, no-cost upskilling resources. Whether you want to train your existing teams or democratize data tools across your entire workforce, these programs will help you build an AI-ready organization:

SMB-Focused Programs: Explore our new SMB Learning Path or enroll in the Gemini Enterprise Agent Ready (GEAR) program for specialized training in agentic AI.
Google Skills for Organizations: Access our no-cost, on-demand learning platform featuring over 3,000 AI courses and hands-on labs created by experts at Google Cloud and Google DeepMind.
Get Certified: Ready to validate your team's expertise? This premium, cohort-based program offers instructor-led training, technical mentorship, and AI-infused skill badges designed to prepare your team for industry-recognized certifications.

By offering a full suite of SMB technology and training — from productivity in Workspace, to all our Ads services, and now powerful AI tools — Google is helping small and midsize firms thrive, no matter where the future takes us.

From AI potential to agentic reality: Driving the UK’s next chapter

Wed, 17 Jun 2026 08:00:00 +0000

The United Kingdom, and London in particular, continues to be one of the great hubs for AI development in Europe and the world. We’re home to Google DeepMind, of course, as well as significant AI unicorns — and Google Cloud customers — like Ineffable Intelligence, which is today announcing an important partnership with us.

A year ago, we joined you for the London Summit to showcase the vast potential of generative AI, including a major investment in upskilling the UK civil service. Today, as we welcome our partners once again to the historic vaults of Tobacco Dock, that potential has become an industrial-scale reality. In my conversations with leaders across both Whitehall and The City, the focus has moved from chatbots and media experiments to full-production execution. This is the moment of the agentic enterprise, where we shift from systems that simply chat with us to systems that can reason, plan, and execute multi-step workflows.

This transition is the cornerstone of the UK’s projected £400 billion economic boost from AI by 2030. At Google Cloud, we are the only provider offering the full integrated stack — custom silicon, frontier models, and planet-scale infrastructure — required to turn the Agentic Enterprise into a reality.

The new frontier of British enterprise and research

The banking sector is a key proving ground for this shift. And HSBC, one of the largest and most important financial institutions in the world, is showing the way. Today, we’re announcing a multi-year transformational partnership with HSBC to accelerate AI adoption across HSBC’s products and services globally. This new collaboration will further accelerate the shift towards AI-enabled ways of working across HSBC’s global operations.

HSBC will work with Google Cloud and Google DeepMind engineering teams to collaborate on new AI-powered tools and programmes, with access to Google’s latest agentic AI capabilities – including Gemini models and the Gemini Enterprise Agent Platform. The initial delivery focus on three areas: hyper‑personalised wealth management support, stronger financial crime risk management, and AI tools to enhance frontline/relationship manager client service

UK startups also continue to break new ground with technology, and AI in particular, as demonstrated by the work of frontier labs like Ineffable Intelligence. The company, which launched earlier this year, has chosen Google Cloud as its preferred cloud partner, utilizing Google’s full stack of AI-optimized hardware and tools to build and train Ineffable’s first generation of foundational models.

Led by David Silver, a former Google DeepMind researcher who was instrumental in the AlphaGo project, Ineffable Intelligence is taking a unique approach to AI development. The team are building systems that learn primarily through their own experience through reinforcement learning, instead of relying on the large-scale human-generated datasets behind language models. The ambition is to create a “superlearner” that develops knowledge through trial and error. This year, Ineffable Intelligence set a record for a European seed funding round of $1.1 billion, and now Ineffable Intelligence will support its training work by deploying one of the largest clusters of A5X, powered by the NVIDIA Vera Rubin NVL72 platform on Google Cloud, delivering massive computational scale.

To move from experimentation to true industrial production, businesses need more than just models; they need a roadmap. To help show them the way, we’re expanding our partnership with Deloitte, which will open a new AI Studio at its London campus. Developed in collaboration with Google Cloud, the studio will help British organisations move beyond AI experimentation to deploy autonomous, action-oriented AI systems at scale.

Deloitte is also committing to upskill 1,000 members of its UK AI and data workforce on Gemini Enterprise. This certification program will ensure that Deloitte’s AI and data engineers’ are equipped with the technical expertise to implement Google’s most advanced agentic architecture, providing UK clients with one of the largest pools of certified AI talent in the region.

Building a future-ready public sector

The blueprint for a modern digital government requires moving away from rigid legacy contracts toward agile, AI-driven public services. In collaboration with the Ministry of Housing, Communities and Local Government (MHCLG), the i.AI incubator, Google Deepmind, and Faculty, we are delivering tangible public sector reform and tools for reinvention that directly support the national goal to "get Britain building."

Agencies like MHCLG are already using a tool called Extract which was built using Google technology to help transform planning processes by reducing document processing times from two hours to just two minutes. Simultaneously, we are supporting trials of an AI planning tool — co-created with local planning authorities in Barnet, Dorset, and Camden — which aims to cut decision times for everyday applications by 50%. Furthermore, the Department for Transport (DfT) is utilizing Gemini to streamline public consultation analysis, a move projected to save £4 million annually.

Innovation on this scale also requires a secure, sovereign foundation. That is why Google Cloud is working to strengthen our UK data residency commitments, including measures like making Gemini 3.5 Flash, which features in-country AI processing, available by late June 2026 for sensitive sovereign use cases. We are giving British organizations the confidence to innovate within strict compliance boundaries.

To help keep businesses safe from the challenges posed by bad actors using AI and other digital threats, we also recently announced a comprehensive AI-powered cybersecurity platform — Google AI Threat Defense — which combines Wiz, Mandiant, Gemini & CodeMender to find, fix, and protect our customers from vulnerabilities.

Proven impact from the high street to public service

Autonomous agents are no longer a future prospect; they are delivering value across the UK economy today. Our work with THG Ingenuity, an ecommerce solutions provider, has delivered an 8x higher conversion rate via its AI Shopping Assistant. Starling is similarly empowering customers with "spending intelligence" tools for instant habit analysis around purchases and expenses. And Rightmove, has launched a beta version of an AI-powered conversational property search, built with Google’s Gemini models, enabling users to search for homes in their own words.

The breadth of this impact is visible across every sector: Kingfisher is pioneering retail-specific agentic applications; Openreach is driving field service optimization in telecommunications; andUnilever is using AI at scale across the entire value chain to drive growth and build desirable brands in the new era of consumer goods.

Meanwhile, VMO2 is streamlining complex data operations; Vodafone is executing a $1 billion partnership to redefine network performance; and WPP is integrating Gemini across creative workflows, whether that's generating high-fidelity campaign assets at speed and scale, powering AI agents, or training robotic camera operators.

Empowering the engine of growth for small to medium businesses and startups

The true measure of Britain’s AI success lies in its small and medium enterprises and startup ecosystem. Our AI Works research highlights a pivotal moment: AI has the potential to boost productivity for small and medium enterprises by 20% and unlock £198 billion in output for the UK economy. With 56% of smaller firms already seeking guidance, we have launched the AI Works for Britain upskilling initiative to ensure no business is left behind.

We also continue to foster the next generation of British unicorn startups through our ongoing partnership with Tech Nation at the London AI Hub. This sustained commitment ensures founders have the resources and community needed to scale, and this September, we will further this mission by hosting the Gemini Startup Forum: Cybersecurity in London to help startups build secure-by-design AI applications.

The Model Garden at Platform 37

Our belief in the UK’s potential is reflected in our physical footprint, too. We are continuing to invest in the UK's digital infrastructure to support growing demand: Our state-of-the-art data center in Waltham Cross launched in September 2025, a key part of our two-year, £5 billion investment to help power the UK's AI economy. And earlier this year, we opened our new office in London in Kings Cross, Platform 37, along with plans for The AI Exchange, a new public space dedicated to deepening understanding of AI.

Building on this momentum, we are excited to introduce The Model Garden at Platform 37, launching in the fourth quarter of 2026. This London-based hub is far more than a physical space; it serves as a strategic investment designed to fundamentally elevate how we engage with our most important customers. Blending the timeless aesthetics of a classic English garden with immersive, high-tech innovation — from living digital walls to a three-story atrium — The Model Garden acts as a physical marketplace for our best ideas.

The blueprint for the agentic enterprise

For UK businesses, civic leaders, and organizations to continue to lead in the AI moment, they must not only rethink the technology they use but also fundamental aspects of how we work. As we support thousands of organizations and millions of teams here and around the globe, we see three core strategies helping achieve success with AI:

Culture: We must reimagine our organizations for the future. True transformation means getting teams excited, enabled, and equipped to work with AI agents in completely new ways. It is about human-AI collaboration, not just automation.
Responsibility: We must build with safety and security in mind from day one. Protecting your users, your customers, and your brand is paramount. Our frontier models are built on a foundation of rigorous AI principles and secure-by-design infrastructure.
Sustainability: In an era of rising compute demands, we must scale in a way that is both financially viable and positive for our planet. At Google, we are committed to carbon-free energy 24/7, ensuring that the UK’s AI growth does not come at the cost of our climate goals.

Architecting the future together

Google Cloud is the primary partner for the UK’s agentic transition. We are moving beyond the hype of experimentation into the rigor of production. From the research labs of King's Cross to the diverse enterprises powering the high street, we are architecting a resilient, sovereign, and prosperous future for the United Kingdom.

Thank you to everyone who’s joining us in London — yesterday, today, and into the future. This year we’ve packaged up an exclusive on-demand experience, allowing you to stream the defining London Summit moments, available anywhere, anytime.

Announcing the newest cohort of the Google for Startups Accelerator: Middle East, North Africa & Turkey

Thu, 28 May 2026 07:00:00 +0000

Google’s mission is to organize the world’s information and make it universally accessible. In high-growth, technically ambitious markets like the Middle East, North Africa, and Türkiye (MENA-T), we fulfill this mission by supporting AI-First startups building the next generation of information-driven services on a global scale. In a region known for its resilience, we want to help founders flourish in any conditions.

The newest cohort of 15 companies in the Google for Startups Accelerator: MENA-T program starts on June 1. They follow on the success of our sixth group, which concluded in November 2025 and set a new benchmark for the region.

Over the course of the fall 2025 program, 14 AI-first startups from 8 different countries received more than 230 hours of specialized 1:1 mentorship from Google experts. This support allowed them to achieve measurable technical and business milestones, including refining their business strategies, accelerating AI/ML initiatives with Google Cloud, and enhancing overall product design.

We’re supplementing the 2026 program with additional resources, focus, and training to help these startups navigate the uncertain geopolitics that can affect the region and the world at any time.

Introducing the newest Google for Startups Accelerator: Middle East, North Africa, Turkey cohort

With a record breaking volume of applications, we are seeing more and more startups leveraging AI technology and addressing meaningful challenges with their business. Please join us in welcoming the 15 companies selected to participate in this cohort:

BioTwin creates virtual twins from health data to detect risks and recommend preventative actions.
Coral replaces manual sustainability processes with real-time enterprise overviews.
Each::labs builds the next generation of AI-native tools to streamline complex developer workflows.
Hakeem translates clinical studies into real-time, patient-specific guidance for clinicians.
inveon.ai deploys agentic AI to provide autonomous digital employees for e-commerce.
Jusoor Labs uses AI to analyze science experiment interactions and improve learning outcomes.
Openfarming automates distributor workflows to reduce waste and protect margins.
Plusfinity builds AI-native learning infrastructure for scalable, interactive education.
Promake empowers the manufacturing sector with AI-driven design and production optimization tools.
Qanooni transforms manual legal work into structured, searchable workflows.
Repzo uses AI to turn complex field data into natural language reports for field teams.
RFxAI streamlines procurement and sales through AI-driven response evaluation.
Tapper applies machine learning to detect anomalies and block invalid traffic.
TruBuild analyzes unstructured construction data for faster, objective tender evaluation.
Woliz uses voice AI to make digital ordering accessible for nanostore owners.

A curriculum designed for impact

Starting June 1st, founders will participate in a three-month program specifically tailored to help startups navigate their unique challenges. The curriculum provides intensive technical support, including comprehensive stack audits and one-on-one mentorship from global experts.

By balancing advanced technical training — focused on AI security and generative design — with strategic business modeling and go-to-market planning, we empower founders to scale their innovations securely. This holistic approach is designed to help startups maintain momentum and drive the region’s sustained digital growth and long-term resilience.

The program has already demonstrated significant impact for the fall cohort, with a number of startups accelerating their growth and development.

COGNNA, a provider of an agentic security operations center (SOC) suite, is among those seeing sustained growth. With improvements made during the accelerator, their platform now allows analysts to work 80% faster, and subsequently have closed a $9.2-million Series A funding round.

By using BigQuery to ingest petabytes of data and Google Kubernetes Engine to scale investigations, the startup has transformed its security operations and dramatically improved efficiency. "Google is shaping the future of COGNNA by enabling us to scale with global markets," said Ziyad Alshehri, co-founder and CTO of COGNNA.

Smart Bricks, a UAE-based startup for AI-powered real estate investing, recently closed a $5 million pre-seed round led by a16z Speedrun. Smart Bricks uses Google’s machine learning pipelines to automate 99% of manual real estate investment workflows across Dubai, London, and New York.

“The Google for Startups Accelerator played a key role in accelerating our technical development,” Mohamed Mohamed, founder and CEO of Smart Bricks, said. “Access to Google’s AI and cloud stack has been instrumental in building and scaling our agentic AI models, particularly given the scale and complexity of the data we’re working with. And infrastructure like Gemini Enterprise Agent Platform and BigQuery allowed us to significantly speed up our development cycles, improve model performance, and bring a much more robust, data-driven platform to market faster.”

Google’s commitment to MENA-T growth

We continue to support founders across the region, providing the specialized resources and cloud infrastructure needed to ensure that innovation continues to scale. Our goal is to ensure that the region’s digital economy continues its acceleration toward a more secure and innovative future.

We are excited to see how this new cohort will shape the future of the MENA-T ecosystem.

The Blueprint: How Movix fills a gap in dental skills with specialized agentic AI

Fri, 22 May 2026 16:00:00 +0000

Welcome to The Blueprint, a regular feature where we highlight how Google Cloud customers are tackling unique and common challenges across industries using the latest AI and cloud technologies. We hope to inspire others looking to innovate in their work.

The demand for dental appliances, like crowns and aligners, is booming, but it’s hard for manufacturers to keep up. At Movix, we’re building one of the first agentic AI solutions for dental appliance manufacturers and dental labs to help companies in the sector acquire digital technical expertise so they can scale clinical workflows cost-effectively and consistently.

The challenge:

Movix started in 2025 with a mission to solve a serious shortage of skilled dental technicians in aligner manufacturing through AI and agentic workflows. The need is significant: the global dental market is valued at nearly $400 billion and growing at double digits, yet many operations remain analog - creating enormous demand for co-pilot, agentic solutions.

Before founding Movix, we had previously started a vertically integrated dental aligner company that focused on very difficult dental situations, such as very crooked teeth. Yet even with highly skilled and trained technicians, there were often mistakes that would require remaking an aligner — a process that costs $300, roughly 25% of the retail price. Poor quality control took a real bite out of the company’s margins.

We saw an opportunity with Movix to address these mistakes by providing technicians with AI-powered quality control agents that automate aligner workflows and reduce errors. To achieve this, we needed to solve for a few key technical challenges:

Develop a custom AI model and end-to-end agentic workflow, since off-the-shelf solutions lacked domain expertise,
Ensure scalability would be built into the platform to prevent outages or production delays,
Achieve broad interoperability through a complex hybrid integration strategy since many dental practices are slow to adopt new technology and run on legacy systems.
Optimize security and compliance to comply with medical record regulatory requirements and keep patient data safe.

The solution:

In order to deliver AI agents that can provide expert-level accuracy, we needed to custom build a lot of the tooling ourselves. We started by developing our custom models for deep learning, computer vision, and 3D mesh analysis over a five-month period, using Google Cloud infrastructure. This intensive, methodical time helped ensure the right level of accuracy and quality control.

We use Google Cloud infrastructure across the full pipeline — from dataset storage and model training to evaluation — to build and refine our defect detection models for intraoral scans. Once defects are detected, we use Gemini Enterprise Agent Platform to generate client-facing feedback that reads as if it came directly from a human technician — acting as a digital team member in the quality control workflow.

Our 3D models use Cloud Run with L4 GPUs for the massive compute power we require; notably, performing the 3D segment scans and detecting defects across the entire fabrication process are highly compute-intensive processes. We use Compute Engine VMs to run experiments, along with various other GPUs to train our models, and perform the heavy lifting of model development in this environment.

Cloud Run and other tools like Cloud Storage support our scalability goals as we target large customers who handle high case volumes — some large labs might produce up to 200,000 appliances per year. Google Cloud's global network of data centers also simplifies regulatory compliance across regions and ensures fast delivery of large 3D datasets to clients worldwide.

The architecture:

The outcome:

Our agentic solutions automate data entry and quality control, which are traditionally manual, time-consuming, and error prone tasks. By automating the work of the best dental technicians, we’re ensuring a top quality product that will improve the fit of crowns, aligners, veneers, and implants for many, many patients. We estimate that our automation and the higher level of accuracy our QC agent delivers could save an aligner manufacturer $300 per remake, for example.

We also believe we’re helping to speed the appliance manufacturing process, leading to quicker turnaround times for dental appliances, which helps dental labs receive revenue faster and improve their cash flow. And we already know we’re meeting a critical need: After we launched the QC agent in October 2025, our first customer signed with us in December. That customer, Orthero, an aligner company serving more than 20 countries, has enjoyed significant results.

“Orthero benefits from this automation by making quality control faster, more consistent, and scalable,” Efer Turhan, a co-founder of Orthero, said. “With support from Movix’s QC AI Agent, we detect missing or inconsistent inputs early and flag unusual deviations before they cause delays.”

The details:

Even with the advantages of AI, our goals demand some serious work. Our architecture supports a solution that’s agentic and modular, integrates into existing on-premises dental systems, and ensures security and compliance.

Our agentic approach allows our system to run checks and balances, manage the complex, multi-step process of quality control for dental scans, and eliminate human errors that occur in data handling and quality review. Our goal is to develop five distinct AI agents by 2029 that cover the entire dental appliance workflow, from original patient dental scan to appliance manufacturing. While our first agents focus on data entry and dental scan quality control, our next agents will handle 3D file repair, clinical review, treatment planning, and manufacturing.

Our solution architecture also enables our system to integrate seamlessly with our customers’ existing lab management and manufacturing systems through API integrations. Because we are selling our solution into a conservative market, we decided to bear the burden of responsibility for successful adoption by doing as much of the integration work as possible.

Because we operate in the highly regulated healthcare industry, we built an environment that strictly follows compliance rules, anonymizing protected health information, or PHI, before it enters our machine learning pipeline to prevent health information from being exposed to the processing environment.

We plan to build hybrid solutions to capture a wider market as we move forward. We're designing an architecture that connects our cloud-based AI agents with older, on-premises software that many conservative labs still use — through lightweight local connectors and standardized APIs. This will allow us to access a large market segment that has not yet migrated to the cloud or begun to use new digital dental technologies.

Taken together, we are not just solving a skills gap, we are reimagining what is possible with co-pilot and agentic solutions across the entire dental industry.

The top announcements for startups from Google I/O ‘26

Thu, 21 May 2026 16:00:00 +0000

Many of the world’s fastest-growing AI startups are choosing to build their future — and the world’s — on Google Cloud because of our complete and open AI stack. We embed AI into every layer of our architecture so your team is fully equipped to build and scale for the agentic era.

At Google Cloud Next ‘26, we focused on providing updates to each layer of the stack to help lean teams move faster and operate more efficiently. We introduced a unified Agent Platform that moves beyond isolated AI tools to a complete lifecycle platform and a new generation of TPUs optimized for both training and inference. This was complemented by the world’s first Agentic Data Cloud, a system designed to take on the shift from human scale to agent scale and updates to our AI-powered cybersecurity platform, now combining Google’s Threat Intelligence and Security Operations with Wiz’s Cloud and AI Security Platform to prevent, detect, and respond to threats in the agentic era.

A diverse set of global startups are already putting this stack to work. For example, Photoroom relies on our highly optimized compute to process over 1,000 images per minute. In the enterprise customer engagement space, Satisfi Labs leverages Gemini, BigQuery, and AlloyDB to deliver intelligent interactions and real-time insights to 800 sports, entertainment, and tourism clients. Meanwhile, global collaboration platform Notion is at the forefront of integrating artificial intelligence into productivity tools, and Mantis AI operates high-throughput inference pipelines to deliver video-native intelligence to major broadcasters.

Now, at Google I/O ‘26, we are expanding on that foundation by introducing more cost-effective frontier models and bringing these agentic capabilities directly into your development workflow. We are bridging the gap between local prototyping and cloud scale, giving your team a straightforward path to build, test, and distribute your applications.

Let’s look at the key announcements for startups from Google I/O and how you can apply them to your business.

1. Smarter, faster models

Models continue to be the foundation of what many startups are building with Google Cloud, so we’re excited to have evolved them once again. This new generation of models deliver maximum intelligence with incredible efficiency.

Gemini 3.5 Flash delivers intelligence that rivals large flagship models at Flash speed. It’s our strongest agentic and coding model yet, ideal for tackling long-horizon agentic tasks, often at less than half the cost of comparable models.
Gemini 3.5 Pro (pre-announcement): We announced our flagship reasoning model officially rolls out next month.
Gemini Omni is a groundbreaking new model that produces dynamic video content by blending text, audio, image, and video inputs. Gemini Omni delivers a highly intuitive approach to video creation and editing — whether you are developing interactive virtual try-ons for e-commerce, streamlining complex post-production workflows, or generating tailored video narratives, Gemini Omni unlocks new ways to create content and drive deeper customer engagement.

What it means for startups: You now have direct access to build with new state-of-the-art models from DeepMind. We are continuing to push the boundaries of what AI can do, ensuring that as we pioneer new frontiers, we simultaneously deliver the speed and cost-efficiencies your startup needs to scale.

Landing in the top-right quadrant of the Artificial Analysis index, 3.5 Flash delivers frontier-level intelligence at exceptional speed — proving you no longer have to trade quality for latency.

2. AI that works for you

At Next ‘26, we introduced purpose-built agents to handle the heavy lifting at every layer of the stack — from threat-hunting SecOps agents and modular data agents, to self-driving cloud infrastructure that autonomously fixes misconfigurations.

But that massive, specialized agentic power needs a coordinator. Enter Google Antigravity — the ultimate control plane. These new updates transform how applications are built, deployed, and managed.

Antigravity 2.0: A dedicated, standalone desktop application for Mac, Windows, and Linux. This acts as an "agent-first" workspace to build, test, and orchestrate complex AI workflows without being tied down to a traditional code editor (IDE).
Antigravity CLI & SDK: For developers who want to stay at the keyboard, the blazing-fast and stack-agnostic command-line interface lets you run and monitor agents seamlessly. Meanwhile, the Python SDK opens up Google's internal agent infrastructure, allowing you to code and control stable agent loops programmatically.
Dynamic subagents: This allows a primary AI agent to automatically spawn smaller, specialized child agents to handle focused subtasks. It unlocks massive parallel engineering output — your main local agent can delegate a database query to a Cloud Data Agent, for example, or spin up a local code-review subagent, all without cluttering its main memory space.
Scheduled tasks: Remove manual maintenance chores by setting background timers and cron schedules. Startups can instruct Antigravity to trigger your cloud-based observability agents or run local repository sanity checks every night at midnight — completely autonomously.
Enterprise-grade security: Antigravity connects local desktop and terminal agent loops directly to your private Google Cloud projects. By inheriting Google Cloud’s standard data privacy protections and Terms of Service, this ensures your customer data is in your control and agent activity runs within your secure cloud boundary by default.

What it means for startups: You no longer just have a coding assistant; you have an entire fleet of specialized AI engineers at your fingertips. By combining the purpose-built cloud agents launched at Next with the Antigravity control plane, a lean team can orchestrate data pipelines, manage security, and execute massive parallel coding tasks — all from a single, secure environment.

Antigravity 2.0 can deploy simultaneous, agent-driven execution. This example shows automated code generation for your website, creation of on-brand assets, and personalized customer email development. Sequences shortened throughout

3. More ways to accelerate development

We are streamlining the developer workflow to help you move from a prompt to a production-ready application with far less friction. This year’s updates focus on removing the infrastructure setup that traditionally slows teams down.

Native Android support in AI Studio: You can now go directly from a natural language prompt to a fully native Android app within the browser. This includes support for the Google Play Console, allowing developers to publish apps directly to the test track without managing local SDK environments.
The seamless handoff: Through a new integration, you can export entire projects from Google AI Studio directly to your local Antigravity environment with a single click. This transfers your complete codebase, files, and conversation context so you can transition from web prototyping to local development without losing your place.
Managed agents: For lean startup teams, building a production-grade agent shouldn't require managing complex infrastructure. Available across both the Gemini API and Google Cloud's Agent Platform, the new Managed Agents API acts as an agent-as-a-service so you can "manage the mission, not the machine." Simply define your instructions and tools, and a single API call will spin up your agent within a secure, ephemeral Google Cloud sandbox. This allows your team to offload the heavy lifting of backend maintenance and focus entirely on building great agentic experiences

What it means for startups: We are providing a straightforward path from prototype to production. You can quickly test app concepts in the browser, move them to a local workspace for deep orchestration, and deploy user-facing agents using managed cloud infrastructure — saving your engineering team weeks of setup and maintenance.

4. A boost for your personal productivity

While your engineers are accelerating product development, we also want to help founders and operators manage the daily noise of running a company.

To help with this, we are introducing Gemini Spark — a new 24/7 personal AI agent that works in the background across Google Workspace and other daily tools. Instead of just answering questions, it can autonomously execute multi-step workflows on your behalf.

For example, Spark can identify a critical product delay, cross-reference your team's documents to recalculate the timeline, update internal tracking sheets, and draft an update email to your investors — all while waiting for your explicit approval before executing.

What it means for startups: While Antigravity builds your product, Spark acts as your digital chief of staff. It handles the routine, manual operational processes so you can stay focused on high-impact, strategic innovation.

Start building today: The Google for Startups AI Agents Challenge

Open globally to eligible startup founders and developers, this competition equips your team with $500 in cloud credits and access to our new Agent Platform, so you can build autonomous agents and compete for a share of a $90,000 prize pool.

We’re offering separate tracks, whether you want to build a net-new agent from scratch, optimize an existing prototype for production, or prep a business-ready agent for enterprise distribution, there is a track tailored to your exact stage. Submissions are open until June 5, 2026, and will be evaluated on technical implementation, business case, innovation, and your final demo. Learn more and sign up for the challenge here.

The Blueprint: Translating stream-of-conscious speech into responsive, actionable task lists

Wed, 06 May 2026 16:00:00 +0000

Welcome to The Blueprint, a new feature where we highlight how Google Cloud customers are tackling unique and common challenges across industries using the latest AI and cloud technologies. We hope to inspire others looking to innovate in their work.

Founded in 2007, Doist is a pioneer in async and remote-first work on a mission to simplify life’s complexities through apps like Todoist for task management and Twist for team communication.

The challenge:

We launched Ramble to take our popular Todoist application to the next level by capturing non-stop, stream-of-consciousness talking. Our inspiration was that scene from The Devil Wears Prada where Miranda Priestly rapid-fires a dozen tasks at her assistant. We asked: What if anyone could capture tasks that way? No typing, no careful formatting. Just talk and let Todoist do the organizing. That use case became our north star.

At the outset, we identified four big technical hurdles:

We needed fast and accurate real-time communication with tool-calling capabilities.
Multilingual support at scale but with great support for slang, accents, and more.
As traditional assertion-based testing would not work for our platform, we would have to find a way to achieve non-deterministic output testing and semantic validation.
Reliable, flawless handling of audio across browsers.

The solution:

We built Ramble using Gemini Enterprise Agent Platform and its previous iteration, Vertex AI; specifically, we’re using Agent Platform to access the Gemini Flash models. We chose these over other options primarily due to the quality of Google's state-of-the-art models and its clear terms and assurances about preserving privacy.

Gemini’s Live API (accessed via Agent Platform) powers Ramble’s core real-time interactions and key capabilities, including native audio streaming, proactive tool calling, session resumption, and multilingual understanding.

Ramble sends the raw PCM audio directly to the model without pre-transcription. Gemini handles language detection, speech recognition, and semantic understanding in a single pass, reducing latency. It then invokes our purposefully designed tools (addTask, editTask, deleteTask, etc) autonomously as the user speaks, without waiting for explicit commands.

The APIs in Agent Platform provide resumption tokens that let users pause and continue sessions, which is essential for mobile users who might switch apps or lose connectivity.

The end result is a clear, concise list of the tasks, regardless of how many, how inconsistently, or how confusingly they may have been rambled by the user.

The architecture:

The outcome:

Ramble has come to rely on the quality of Google’s AI models, particularly the reasoning and near-instant audio-processing capabilities of Gemini Flash. Other platforms and models offer similar capabilities, and we did bake in support for them, but none hit our internal quality bar as consistently as Gemini. When it came to a user's unstructured “rambling” and the need to fill in gaps, Gemini turned out to be the most intelligent of all the models we explored. The result was the clearest and most consistent breakdown of tasks, which was the exact magical user experience we wanted to create.

After an early rate-limit incident caused by unexpectedly high usage during alpha testing, we developed a deeper, more proactive partnership with Google, ensuring long-term sustainability and the support necessary for our high API usage. Since then, it's been easy for us to connect directly with Google Cloud staff, including engineers, when issues arise.

Here at Doist, Ramble took off both in a qualitative and quantitative sense. It’s become a hallmark experience that incentivizes us to explore tasteful applications of AI that can enhance our existing product experience, both in the B2C space as well as B2B. Beyond task creation, we’re considering several opportunities across the productivity journey, from capture to planning and even automation.

The details:

We structured our back-end to enable future voice-powered features. The architecture includes a provider-agnostic streaming layer; a dictation module for one-way audio; Ramble (our “brain dump” module); and a conversation module to support streaming bi-directional audio and future conversational features.

This layered design means we can ship new voice features with minimal additional infrastructure work. It also enables provider flexibility; although we’re using Gemini Enterprise Agent Platform in production, our abstraction layer also easily supports other solutions.

In addition to helping us tackle three of our four key technical challenges, Agent Platform delivered some nice surprises. First, session resumption was easier than we expected. We initially thought maintaining conversation state across reconnections would require complex server-side session management. But once we understood Agent Platform’s resumption token approach (the token is provided by the API and changes with each context update), implementation was straightforward across all platforms.

Second, context injection worked on the first try. We spent considerable time designing how to provide user context (projects, labels, preferences) to the model. We explored complex retrieval strategies and dynamic context windows. In the end, the simple "v1" approach—just passing most of the user's metadata in the system prompt—worked remarkably well.

For testing, we combined structural validation (task count, priority levels, date presence, etc) with semantic validation (did the model understand the user's intent?) following the LLM-as-judge approach. A second Gemini model evaluates whether the output semantically matches the expected outcome. Native speakers from our global team recorded real-world scenarios in their languages and local accents (15+ language variations and over 100 recordings total), with each scenario having expected semantic outcomes (e.g., "should create 3 tasks: one about calling family, one about shopping, one about exercise on Saturday at 11 AM").

We then created a defined pass-rate threshold for the test suite overall, while also monitoring per-language performance to catch regressions. This approach lets us evaluate new model versions systematically, understanding not just overall performance but also which specific languages might see improved or degraded experiences, and make data-informed decisions.

Ultimately, Ramble is a resounding success in helping our users handle the chaos of day-to-day life. It joins the ranks of Todoist’s Quick Add — our existing natural-language task input — in providing yet another way to capture tasks that is the best in its category.

The founder’s AI foundation: The top announcements for startups from Next ‘26

Wed, 29 Apr 2026 16:00:00 +0000

The momentum is undeniable: the world’s fastest-growing AI startups are building with Google Cloud. Instead of stitching together fragmented point solutions, founders are building their businesses here because we offer the entire AI stack in a single, open environment.

And, as we saw at Next ‘26 last week, we continue to advance the models, infrastructure, platforms, security, and governance that allow startups to build faster and dream bigger.

You can already see this scale and velocity in action every day. Platforms like Lovable are empowering developers to generate over 200,000 new projects daily. Thinking Machines Labs is leveraging the latest NVIDIA Blackwell chips through Google Cloud to double its training and serving speeds. When it comes to activating complex data, Parallel bypassed the multi-vendor headache, using our integrated ecosystem — bridging Gemini, BigQuery, and Spanner — to massively scale their high-accuracy search APIs.

Startups are moving at this pace because we’ve eliminated the operational plumbing. Aible is running complex enterprise agents directly where their data lives in BigQuery, while teams like Emergent AI are using Google Kubernetes Engine (GKE) to autonomously scale thousands of secure sandboxes for vibe coding natural language prompts into production-ready applications.

By putting everything in one place, we provide the ultimate foundation so you can focus 100% on shipping the AI-native products your customers are waiting for.

Let’s look at some of the biggest announcements out of Next ‘26 and what they mean for you.

1. Gemini Enterprise: comprehensive agent platform

The new Gemini Enterprise Agent Platform is how we’re moving beyond isolated AI tools to a complete lifecycle platform. It evolves the model building capabilities of Vertex AI with advanced features for:

Building: The upgraded Agent Development Kit (ADK) introduces a graph-based framework for complex multi-agent reasoning, while the low-code Agent Studio allows you to move seamlessly from prompts to deployed agents. Developers can also utilize Agent Garden to jumpstart development with pre-built agent templates.
Scaling: The re-engineered Agent Runtime delivers sub-second cold starts and supports long-running agents that maintain state for days at a time. The new Agent Memory Bank enables agents to recall high-accuracy details for personalized, long-term context.
Governing: To ensure security and compliance, the platform introduces Agent Identity for trackable auditing, Agent Gateway for unified connectivity and policy enforcement, and Agent Threat Detection to flag suspicious behavior in real time.
Optimizing: Teams can ensure quality using Agent Simulation to test against synthetic interactions, and rely on Agent Optimizer to automatically analyze real-world failures and suggest improvements.

What it means for startups: Startups no longer need to stitch together fragmented tools to build complex AI systems. The new Agent Platform provides every tool in one place — whether you are using the visual interface of Agent Studio to quickly prototype or the code-first logic of the ADK for advanced orchestration, a lean team can build and scale production-ready agents with security guardrails from day one. This approach helps eliminate technical debt and drastically accelerates time-to-market.

2. The Agentic Data Cloud: Data that actually does the work

We are moving your data from rows and columns to autonomous action. This is where you build AI agents that understand your business logic and execute the next best action—closing the gap between simply analyzing data and actually resolving real tasks on your behalf. Because agents generate orders of magnitude more workloads, our Agentic Infrastructure scales to handle bursty startup demands, improving query speeds and reducing infrastructure costs.

Query anywhere, leave data where it lives: The Cross-Cloud Lakehouse lets you use Google's advanced AI on data living in AWS or Azure. Build smarter agents without requiring migrations, experiencing vendor lock-in, or incurring additional costs to move your data.
The zero-ETL startup: Features like One-Click Reverse ETL and Lakehouse Federation collapse the walls between operational databases and analytics, allowing data to flow seamlessly without manual engineering.
Give your agents business context: Before an agent acts, it needs to understand your specific business rules. The new Knowledge Catalog turns passive data into active semantics, ensuring your agents know the exact difference between "booked revenue" and "projected revenue" without guessing or hallucinating.
Eliminate "glue code" and manual connectors: Fully managed, remote Model Context Protocol (MCP) servers are now embedded directly across the database portfolio (Firestore, Spanner, BigQuery, and AlloyDB).
Production-grade guardrails: Connecting AI to your database usually means risking dangerous SQL hallucinations. New Tools for Data Agents solve this by providing pre-built, modular building blocks that deliver near-100% text-to-SQL accuracy, letting your custom agents interact with live data safely and reliably.
Instant app generation: A new integration with Google AI Studio allows developers to use natural language for vibe coding to turn a simple text prompt into a live, deployed application connected to Firestore in seconds.

What it means for startups: Your proprietary data can be a competitive moat, but activating it usually requires a massive integration tax — breaking down cloud silos, writing brittle glue code, and managing complex ETL pipelines. Our latest data advances eliminate the friction. Startups can now leave their data right where it lives in AWS or Azure, "vibe code" an app into existence, and instantly plug agents into secure databases using standard MCPs. It turns data integration from a multi-week engineering sprint into a single afternoon's work.

3. Next-Gen Compute: Performance and Cost-Efficiency

Google Cloud is drastically expanding its AI Hypercomputer portfolio to offer increased performance-per-dollar, while introducing autonomous features that handle the heavy lifting of infrastructure management for lean teams:

TPU 8t for training: Optimized for training massive models, the TPU 8t achieves nearly 3x higher compute performance than the previous generation and delivers up to 2x better performance-per-watt.
TPU 8i for inference: Designed to make running agents more affordable, the TPU 8i delivers 80% better performance-per-dollar for inference than the prior generation.
Axion N4A for general compute: Google Cloud Axion instances deliver 100% better price-performance than comparable x86 instances, ensuring highly efficient, sustained operations.
Multi-agent networking: New C4N and M4N machine series offer a nearly 4x increase in network bandwidth, purposefully built to handle high-volume communication between AI agents.
GKE sub-second agent scaling: Google Kubernetes Engine Agent Sandbox now deploys 300 sandboxes per second, per cluster. This delivers sub-second cold starts for AI workloads, ensuring instant response times and secure multi-tenant isolation.
The self-driving cloud: Google Cloud integrated Gemini reasoning directly into its telemetry. The system now performs autonomous root-cause analysis, identifying and fixing infrastructure misconfigurations before a human even realizes there’s a problem.
NVIDIA integration: Google remains at the bleeding edge of partner hardware, becoming one of the first to offer the NVIDIA Vera Rubin NVL72 alongside Blackwell and Hopper.

What it means for startups: Managing that infrastructure can drain precious engineering cycles. Google Cloud’s expanded portfolio solves both problems. You can train models faster with the TPU 8t, scale millions of concurrent agents cost-effectively using the TPU 8i, and deliver zero-latency experiences using GKE sandboxes.

More importantly, with our autonomous root-cause analysis handling the operational plumbing, your engineers are freed up to focus on what actually matters: building your product.

4. Agentic Defense: Security as Your Enterprise Go-To-Market Engine

Google Cloud is combining its global threat intelligence with Wiz's Cloud and AI Security Platform to provide startups with a fully automated, unified security posture from code to cloud:

API economics protection: The new Google Cloud Fraud Defense acts as an intelligent bouncer, preventing malicious bots from scraping IP or running up massive compute bills on unauthorized agent interactions.
Full-stack AI protection: The Wiz AI Application Protection Platform (AI-APP) secures every layer of the AI stack and natively supports the "outer layer" of the cloud, now securing tools startups actively use like Vercel, Databricks, and Cloudflare.
Google-scale threat intelligence: New agentic SecOps tools, like the Threat Hunting Agent and dark web intelligence, automate detection rules and elevate only the threats that matter with 98% accuracy. Meanwhile, the Triage and Investigation agent uses Gemini to shrink 30-minute security investigations down to just 60 seconds.
SecOps agents: We introduced three new agents in Google Security Operations — Threat Hunting (proactive pursuit of stealthy threats), Detection Engineering (automated detection creation), and Third-Party Context (workflow data enrichment) — that empower teams to defend at the speed of AI.
Drag-and-drop security automation: Wiz Workflows introduces a new hub with a customizable drag-and-drop interface, allowing lean teams to easily orchestrate how and when these AI agents act without writing complex security scripts.

What it means for startups For B2B startups, the enterprise information-security review is where deals often go to die. By building on our unified security foundation, you can demonstrate that your application has continuous, automated red-teaming and runtime protection.

It turns the dreaded vendor risk assessment into a competitive advantage, allowing you to bypass procurement roadblocks and close enterprise deals faster without needing to hire a massive SecOps team.

5. The Go-To-Market Engine & Ecosystem Capital

Google is solving the ultimate startup hurdles — distribution and funding — by opening up its enterprise channels and deploying massive capital to partner ecosystems:

Agent Gallery in Gemini Enterprise: Startups can now monetize their customized agents directly inside the Gemini Enterprise app via Google Cloud Marketplace. This reaches millions of users right in their daily workflows and turns everyday user discovery into an automated procurement flow for IT, accelerating purchasing cycles by up to 50%.
$750-million fund: A capital injection dedicated to partner agent development and co-marketing, designed to fuel the next generation of AI builders.

What it means for startups Building the product is only half the battle; surviving enterprise procurement is the other. By integrating into the Agent Gallery, your startup can reach millions of Gemini Enterprise users directly in their daily workflow, where they can easily click to request IT procurement. Plus, their purchases can draw down on existing Google Cloud commitment, allowing you to tap into a $240 billion backlog. Paired with the $750-million partner innovation fund, Google is putting capital and distribution directly behind your growth.

Start building with the AI Agents Challenge

We’ve given you the architecture, the security, and the distribution channels — now it’s time to get hands-on. To help you accelerate your development, we are launching the Google for Startups AI Agents Challenge. Open globally to eligible startup founders and developers, this six-week competition equips your team with $500 in cloud credits and access to our latest AI tools, including Gemini Enterprise, so you can build autonomous systems and compete for a share of a $90,000 prize pool.

Startups are building the next big thing with Google Cloud AI

Wed, 22 Apr 2026 12:00:00 +0000

The future is taking shape in Las Vegas this week, where the world’s leading startups are showcasing their groundbreaking AI work at Google Cloud Next. Whether they need the top AI models, infrastructure purpose-built for the most demanding workloads, intelligent multi-model databases, or security and sovereignty solutions to grow with confidence — or all of the above — startups of all sizes and in every industry are choosing Google Cloud.

To take one example of thousands, Lovable, the popular software creation platform, utilizes Google Cloud infrastructure and builders are now creating over 200,000 new projects on Lovable every day. OpenEvidence, a medical AI platform widely used by U.S. physicians, uses Google's AI tools to provide evidence-based answers at the point of care; the company recently announced major partnerships with Wiley and Mount Sinai, expanding access to trusted medical research and to its AI assistant.

This week at Next we’re announcing even more startups who are starting or expanding their work with Google Cloud. These include leading names in areas like vibe coding, model training and inference; healthcare and life sciences; image generation; financial services; gaming and much more. It also includes an exciting new group of startups who have recently begun accessing GPU-based AI compute through Google Cloud’s AI Hypercomputer, thanks to our deep technical partnership with NVIDIA.

Google Cloud has always offered startups the entire AI stack in one unified place, giving founders maximum velocity. With new exciting launches at Next ‘26, the options and opportunities are only growing.

We’re helping you build complex agents faster with Gemini Enterprise Agent Platform, optimize compute needs with two differentiated TPU offerings for inference and training, reduce data friction with our next-gen cross-cloud Lakehouse, and better govern everything with the help of Wiz AI agents and Google-scale threat intelligence, to name just a few advances. We’ve also unveiled the ability to sell your AI solutions directly to enterprise users through marketplace integrations in Gemini Enterprise.

Supporting startup innovation across categories

If there was any question of just how dramatically AI is transforming every facet of work and life, look no further than the innovations being brought to market by our startup customers.

Coding and developer tools

Emergent AI uses Google Kubernetes Engine (GKE) as its core infrastructure to orchestrate thousands of on-demand, isolated development sandboxes for vibe coding. By leveraging GKE Autopilot for automated scaling and Gemini models for code generation, the platform transforms natural language prompts into live, production-ready applications with instant previews.
Lovable is expanding its work with Google Cloud and launching a new coding agent, now available to purchase through Google Cloud Marketplace and Gemini Enterprise.
Parallel builds high accuracy web search and research APIs, purpose-built for AI agents. Leveraging Google Cloud offerings including Gemini, BigQuery, BigTable, and Spanner has enabled them to build, launch and massively scale their platform since launching in August 2025.
Temporal partnered with Google Cloud to build a unique Application Development Kit (ADK) integration. This enables organizations to deploy agentic systems for mission-critical use cases that require strong execution guarantees, uptime, auditability, and operational resilience.

Customer engagement

Satisfi Labs is an agentic AI platform designed to transform customer engagement for destinations and experiences, serving 800 clients across sports, entertainment, and tourism. They are built on the Google Cloud ecosystem and leverage technologies including Gemini, BigQuery, AlloyDB, and Datastream to enable intelligent interactions, scalable data processing, and real-time insights.
Vapi, a platform for developers creating conversational voice AI, leverages Gemini as its intelligence core, allowing developers to maximize their compute-per-dollar while delivering a premium voice experience with incredible scale.
Vurvey Labs leverages Gemini Enterprise Agent Platform to build agent populations and simulated worlds that predict human behavior and future needs.
ZenBusiness utilizes Gemini Enterprise Agent Platform to simplify business ownership, including the launch of 'Velo,' an AI assistant for formation, compliance, and website building.

Document intelligence

Reducto leverages Gemini Enterprise Agent Platform and AI Studio alongside its core models to automate high-stakes workflows and power production AI pipelines at scale.

Finops

Mavvrik is a FinOps platform using Gemini and BigQuery to power their unified cost management platform for FinOps, IT, and finance leaders to gain visibility, automate allocation, and govern spending across public clouds, private clouds, AI workloads, and SaaS applications.

Generative media

ComfyUI is streamlining complex node-based gen media production by integrating Nano Banana Pro while improving the consistency of their customer experience using Provisioned Throughput.
Gamma used by over 70 million people to create presentations, documents, websites, and social posts, uses Nano Banana 2 to instantly generate localized visuals and transform ideas into ready-to-share content in seconds.

Industry AI

Chorus, spun out of Alphabet’s X innovation arm, is building AI orchestration tools to transform how physical assets are made, moved, and managed across industries. Leveraging Gemini Enterprise Agent Platform, advanced machine learning, and proprietary algorithms, Chorus provides granular asset-level visibility, temperature intelligence, and dynamic delivery estimates across mission-critical supply chains.
ExaCare AI, an AI platform for post-acute care operations, is using Gemini Enterprise Agent Platform and Gemini for inference and to process clinical and operational data across admissions and reimbursement workflows.
Industrility, an AI-powered aftersales platform for machine manufacturers, built a maintenance intelligence engine on Google Gemini’s advanced language models, extracting critical maintenance tasks and reducing a two-month manual project to three hours.
Insilica uses Gemini Enterprise Agent Platform to process millions of toxicology regulatory documents and thousands of databases into a unified, agentic risk assessment platform.
Proximal Health utilizes DocAI to automate the insurance claims adjudication process.
Stord, which offers fulfillment and software for omnichannel retail, leverages Gemini Enterprise Agent Platform to build ML models that predict estimated delivery dates and optimize demand planning across millions of orders, all grounded in real data behind over $10B in annual Gross Merchandise Value.
Stylitics uses Gemini Enterprise Agent Platform to power more than 1,000 AI programs for world-class retailers, including AI image generation, outfitting and styling, and AI analytics.
Wand uses Gemini Enterprise Agent Platform to power Game Guide, a contextual companion built into their single-player PC gaming platform, that helps players get unstuck, uncover hidden content, and stay immersed without ever leaving the game window.

Productivity

Notion is at the forefront of integrating artificial intelligence into productivity tools, leveraging Gemini models to power its text and image generation capabilities.

Sustainability

Watershed, a leading enterprise climate platform, is leveraging Google Cloud to transform sustainability data into actionable climate intelligence, accelerating global decarbonization and enabling organizations to move from measurement to action at an unprecedented scale.

Supporting model training and inference with NVIDIA AI infrastructure

Today, when the world’s most innovative startups need to access NVIDIA’s cutting-edge GPU-powered compute, many of them rely on Google Cloud. Our longstanding partnership with NVIDIA allows these startups to train and serve AI models, support fine-tuning, and scale new agentic workloads.

This week at Google Cloud Next, we’re highlighting the successes these customers are having and announcing exciting new partnerships with more startups choosing to utilize NVIDIA AI infrastructure and Google Cloud’s AI stack.

These new and expanding customers include:

Thinking Machines Labs will utilize the latest NVIDIA Blackwell chips through Google Cloud, including being among the first Google Cloud customers to utilize GB300s. In early testing, Thinking Machines has seen training and serving speed increase 2x on GBs compared to prior generation chips.
GMI Cloud leverages Google Cloud’s GPU infrastructure to run its inference-optimized platform, powering GMI Studio workflows and large-scale model serving across text, image, and video. This infrastructure supports high utilization driven by growing demand for real-time, production-grade AI applications.
Inferact is developing vLLM, a highly performant and easy to use inference engine for open source LLMs on data center hardwares, utilizing a cluster of NVIDIA GB200’s on Google Cloud, as well as services like Managed Lustre and Cloud Filestore.
Neuraldefend is partnering with Google Cloud and NVIDIA to utilize H100 and L4 GPUs for model training and production inference for their real-time deepfake detection engine and digital trust solutions.
Nucleus AI is utilizing a cluster of NVIDIA H100 GPUs through Google Cloud for model training.
Reflection has expanded its usage of G4 VMs through Google Cloud, as well as services like Google Cloud Storage to support training and inference for its open-weight models.
Reve uses A3 Mega (NVIDIA H100 GPUs) and A3 Ultra (NVIDIA H200 GPUs) instances on Google Cloud to handle their intensive multimodal foundation model training and inference. They also use GKE to manage their high performance training needs.

Startups most commonly access NVIDIA GPUs through Google Cloud G4 VMs, powered by NVIDIA’s latest Blackwell infrastructure. We’re also excited to offer NVIDIA’s Vera Rubin architecture to startups later this year. Our deep partnership with NVIDIA means that the Google Cloud AI Hypercomputer is highly optimized for GPU customers and can deliver ultra-low latency, high-throughput, and cost-effective performance across multiple use cases.

Startups are choosing to utilize GPUs through Google Cloud in part because of this high level of optimization and because they can easily access highly differentiated technologies in our AI stack like Gemini Enterprise, Agent Builder, BigQuery, Google Kubernetes Engine, and much more.

We also continue to partner closely with NVIDIA to support early-stage startups with AI compute and cloud services through NVIDIA’s Inception program and our Google for Startups Cloud Program. This includes exciting startups like:

Aible, which is building long‑running enterprise AI agents. They are accessing GPUs on Google Cloud to help them run Nemotron 3 and NemoClaw agents directly on BigQuery, orchestrate multi‑agent blueprints like the Intelligent Warehouse, and deploy serverless RTX‑accelerated inference on Cloud Run.
Baseten provides a managed inference platform that optimizes model serving for large language and vision models. They are accessing GPUs on Google Cloud to help them deploy NVIDIA‑accelerated inference with improved tokens‑per‑second, cost‑per‑1K requests, and GPU utilization for customers running production gen AI workloads.
CodeRabbit builds AI-driven workflows for code review, supported by Nemotron models via Google Enterprise Agent Platform, delivering low-latency PR summarization, embedding-powered context enrichment, and scalable concurrent review pipelines.
Factory AI develops autonomous code-editing agents that can traverse complex code graphs, resolve dependencies, and generate safe patches against real production repositories. Nemotron models in Gemini Enterprise Agent Platform allow Factory AI to power Nemotron Super-driven "droids" that perform semantic code search, tool-augmented reasoning such as grep and static analysis, and multi-file refactoring plans at scale.
Mantis AI is creating a video‑native intelligence layer for media and entertainment companies across Latin America. Through NVIDIA’s Inception program and its collaboration with Google Cloud, Mantis AI operates high‑throughput inference pipelines, rigorously benchmarks fine‑tuned model variants, and experiments with NVIDIA NIM, Blueprints, and vector search services to deliver production‑ready stacks that help broadcasters and streamers move from analyzing video to orchestrating it at scale.
Photoroom specializes in diffusion‑ and transformer‑based vision models that generate and edit product imagery. Photoroom processes over 1,000 images per minute. They use TensorRT for a 4x speedup and xformers to make their Stable Diffusion pipelines 100% faster. They’ve consumed over 500,000 GPU hours on Google Cloud without issue.

Sell agents directly in Gemini Enterprise

For AI startups, navigating enterprise procurement is often a bottleneck to growth. By integrating the Google Cloud Marketplace into Gemini Enterprise, we're providing a global commercialization engine to accelerate procurement.

Today at Next, we’re announcing an exciting group of more than 70 startups who are bringing their agents to our Marketplace.

And, to help even more startups build agents and bring them to market, we are launching a new $750 million dollar fund to support agent development and marketing for our partners. You can read more about this new funding here.

Presenting the Gemini Startup Forum

In June, we’re hosting our Global Gemini Startup Forum, a two-day, in-depth gathering of founders from seed to Series A startups. At the event, attendees will engage with Google leaders and engineers, build and scale products, and share feedback with teams across our AI portfolio. Take a deeper look at the program and register your interest for upcoming editions.

Building on that momentum, we’re also excited to share that applications are open for our next program: the Gemini Startup Forum: Cybersecurity, taking place in London later this year. If your startup is utilizing AI to build the next generation of security solutions, we invite you to apply to join us!

Join the Google for Startups AI Agents Challenge

Startups, it's your chance to get hands-on right away. We’re excited to launch the Google for Startups AI Agents Challenge at Next 2026.

Open to anyone (not just Next attendees), this global competition runs for six weeks and gives teams $500 in credits and the tools — like the new Gemini Enterprise Agent Platform — to build autonomous systems and compete for a share of a $90,000 prize pool.

The challenge, which opens for submissions today, and runs until June 5, 2026, offers three tracks tailored to where you are in your development journey. You can build a net-new agent from scratch, optimize an existing prototype for production reliability, or refactor a business-ready agent for potential enterprise distribution on Google Cloud Marketplace and the Gemini Enterprise app.

Submissions will be evaluated based on the following weighted criteria: Technical Implementation (30%), Business Case (30%), Innovation and Creativity (20%), and Demo and Presentation (20%). Learn more here and sign-up now.

Read the new Future of AI: Perspectives on generative media for Startups report

Our Future of AI: Perspectives on generative media for Startups report compiles exclusive interviews with eight startup founders, investors, and Googlers to capture their candid advice, strategic priorities, and predictions for the industry's future.

Understanding these firsthand founder perspectives will be instrumental in helping you navigate the next wave of AI-driven content creation and build the future of generative media on Google Cloud — and maybe find yourself featured on a list like this at a future Next.

Small and midsize businesses jumpstart their AI transformations with Gemini Enterprise

Wed, 22 Apr 2026 12:00:00 +0000

Small businesses are the backbone of the global economy. With 400 million SMBs worldwide and 36 million in the U.S. alone, they provide 50% of global employment. Now, with Google Cloud AI, they’re scaling faster, operating more efficiently, and delivering better results for their customers.

For years SMBs have been working with cutting edge products like Google Workspace, Google Ads, Search, YouTube, Maps, Google Wallet, and more to grow their business. Now, more SMBs are taking the first big steps in their AI journeys with our leading models and our agentic platform, Gemini Enterprise.

By making AI accessible to every employee, Gemini Enterprise serves as a launchpad for SMBs ready to develop an AI strategy and build new solutions to work smarter than ever before.Businesses are using AI to automate manual tasks and build custom solutions that extend their reach and help them better support their own customers.

We are proud to support these early milestones, and at Next we’re excited to showcase dozens of new customer partnerships with organizations taking their first steps toward an AI-driven future. Read on to explore their success stories and find our guide to the best SMB resources at Google Cloud Next ’26.

How Gemini Enterprise helps SMBs solve complex challenges

Across the globe, small and medium businesses are using AI to help grow their business. Here’s just a few of the many exciting examples of how SMBs are already benefiting from the secure, scalable, enterprise-grade tools in Gemini Enterprise:

Apex Leaders, a US-based business services company, is using Gemini Enterprise to power an internal search engine, providing easy access to internal data sources and automated information summarization and content drafting for consultant teams.
ARCHINERGY, a US based architecture firm, utilized Gemini Enterprise to create an AI agent that answers questions about architectural design and internal documentation, generating faster insights in just a fraction of the time compared to traditional search methods.
Constructora Las Galias, a Colombia-based construction firm, is using Gemini Enterprise to power a centralized internal knowledge platform, providing streamlined sales and legal workflows and immediate employee access to critical business information.
Eficode, which operates as a software development and DevOps firm within Finland, is using Gemini Enterprise to create a multi-platform support engine that quickly indexes complex internal wikis, so employees can provide customers with faster technical support
Grant Thornton Chile, a leading audit and consulting firm, is using Gemini Enterprise to enhance its agile management by powering secure financial data validation, accelerating HR and sales questions, and enabling unified search across disparate document types.
Grupo Ruiz, a leading Spanish company in sustainable mobility, uses Gemini Enterprise to improve team productivity and process efficiency, achieving an optimized workflow and greater data visibility
Huge, an American business services agency, utilized Gemini Enterprise to create AI agents that automate market research and contract analysis, generating new business intake in just minutes compared to several days.
Incrementa, a Chilean marketing company, is using Gemini Enterprise to automate client reporting and data analysis, helping it scale personalized service and boost campaign reach by 40%.
Koufu, one of Singapore's largest food and beverage companies, which operates and manages a large network of food courts and coffee shops, is using Gemini Enterprise to automate sales reporting and BigQuery data analysis. This reduces spreadsheet bottlenecks so shop managers can focus on front-of-house service quality.
La Maison du Whisky (whisky.fr), a specialized spirits and beverages retailer in France and internationally, chose Gemini Enterprise to create a 'Digital Sommelier' that transforms technical product data into high-end storytelling and marketing copy in seconds
SAI360, a US-based software company, is using Gemini Enterprise to power the design, production, and localization of ethics training, achieving 99% faster course creation and a 95% reduction in training media production costs.
Tauá Resorts group is using Gemini Enterprise to create an intelligent agent that generates real-time insights and dialogue suggestions. This technology allows the sales team to answer complex questions instantly, leveraging data from previous events to craft the perfect proposal for every client.
Tirol, a leading dairy producer in Brazil, is using Gemini Enterprise to create an interactive knowledge base, providing democratized data access and enhanced visibility across its entire value chain.
Wimberly Allison Tong & Goo leverages Gemini Enterprise to streamline digital planning, landscape, architecture and interior design workflows and visual queries, enabling the International Hospitality Design firm to accelerate the transitions from idea generation to design concepts and high-fidelity visualizations.

Practical Resources for Lean Teams at Next ‘26

At Google Cloud Next 2026, we are showcasing how SMBs can enhance their impact by integrating AI into their daily workflows. For attendees focused on maximizing impact with focused resources, we have curated a dedicated track to help you navigate the AI era.

When browsing the Session Library, use the Small IT Team filter to find deep dives into:

Cost Optimization: How to deploy AI effectively and manage resources on a budget that’s right for your goals.
Low-Code AI: Training on how to accelerate your business goals by building custom apps and agents.
Adaptive Security: Proactive defense strategies designed to provide enterprise-grade protection regardless of your security team's size.

Get Started Today

If you can’t make every session, we’ve made it easy to keep the momentum going:

The AI Starter Kit: A streamlined, four-step guide to deploying Cloud AI across your organization.
Monthly AI Newsletter: Sign up to receive the latest industry highlights and Cloud AI updates all year long.

Check out the full SMB guide here and start planning your itinerary. See you there!

Technical guide: Four steps for startups to build multi-agent systems

Thu, 13 Nov 2025 17:00:00 +0000

AI agents are transforming the nature of work by automating complex workflows with speed, scale, and accuracy. At the same time, startups are constantly moving, growing, and evolving – which means they need clear ways to implement agentic workflows, not piles of documentation that send precious resources into a tailspin.

Today, we’ll share a simple four-step framework to help startups build multi-agent systems. Multi-agentic workflows can be complicated, but there are easy ways to get started and see real gains without spending weeks in production.

In this post, we’ll show you a systematic, operations-driven roadmap for navigating this new landscape, using one of our projects to provide concrete examples for the concepts laid out in the official Startups technical guide: AI agents.

Step #1: Build your foundation

The startups technical guide outlines three primary paths for leveraging agents:

Pre-built Google agents
Partner agents
Custom-built agents (agents you build on your own).

To build our Sales Intelligence Agent, we needed to automate a highly specific, multi-step workflow that involved our own proprietary logic and would eventually connect to our own data sources. This required comprehensive orchestration control and tool definition that only a "code-first" approach could provide.

That’s why we chose Google's Agent Development Kit (ADK) as our framework. It offered the balance of power and flexibility necessary to build a truly custom, defensible system, combined with high-level abstractions for agent composition and orchestration that accelerated our development.

Logical flow on how our agent works

Step #2: Build out the engine

We took a hybrid approach when building our agent architecture, which is managed by a top-level root_agent in orchestrator.py. Its primary role is to act as an intelligent controller using an LLM Agent for flexible user interaction, while delegating the core processing loop to more deterministic ADK components like LoopAgent and custom BaseAgent classes.

Conversational onboarding: The LLM Agent starts by acting as a conversational "front-door," interacting with the user to collect their name and email.
Workflow delegation: Once it has the user's information, it delegates the main workflow to a powerful LoopAgent defined in its sub_agents list.
Data loading: The first step inside the LoopAgent is a custom agent called the CompanyLoopController. On the very first iteration of the loop, its job is to call our crm_tool to fetch the list of companies from the Google Sheet and load them into the session state.
Tool-based execution in a loop: The loop processes each company by calling two key tools: the research_pipeline tool that encapsulates our complex company_researcher_agent and the sales_briefing_agent tool that encapsulates the sales_briefing_agent. This "Agent-as-a-Tool" pattern is crucial for state isolation (more in Step 3).

This hybrid pattern gives us the best of both worlds: the flexibility of an LLM for user interaction and the structured, reliable control of a workflow agent with isolated, tool-based execution.

Step #3: Tools, state, and reliability

An agent is only as powerful as the tools it can wield. To be truly useful, our system needed to connect to live data, not just a static local file. To achieve this, we built a custom tool, crm_tool.py, to allow our agent to read its list of target companies directly from a Google Sheet.

To build our read_companies_from_sheet function, we focused on two key areas:

Secure authentication: We used a Google Cloud Service Account for authentication, a best practice for production systems. Our code includes a helper function, get_sheets_service(), that centralizes all the logic for securely loading the service account credentials and initializing the API client.
Configuration management: All configuration, including the SPREADSHEET_ID, is managed via our .env file. This decouples the tool's logic from its configuration, making it portable and secure.

This approach transformed our agent from one that could only work with local data to one that could securely interact with a live, cloud-based source of truth.

The "Agent-as-a-Tool" Pattern for managing state in loops: A critical challenge in looping workflows is ensuring state isolation between iterations. ADK's session.state persists, which can cause 'context rot' if not managed. Our solution was the "Agent-as-a-Tool" pattern. Instead of running the complex company_researcher_agent directly in the loop, we encapsulated its entire SequentialAgent pipeline into a single, isolated AgentTool (company_researcher_agent_tool).

Every time the loop calls this tool, the ADK provides a clean, temporary context for its execution. All internal steps (planning, QA loop, compiling) happen within this isolated context. When the tool returns the final compiled_report, the temporary context is discarded, guaranteeing a fresh start for the next company. This pattern provides perfect state isolation by design, making the loop robust without manual cleanup logic.

Step #4: Go from Localhost to a scalable deployed product

Here is our recommended three-step blueprint for moving from a local prototype to a production-ready agent on Google Cloud.

1. Adopt a production-grade project template
Our most critical lesson was that a simple, local-first project structure is not built for the rigors of the cloud. The turning point for our team was adopting Google's official Agent Starter Pack. This professional template is not just a suggestion; for any serious project, we now consider it a requirement. It provides three non-negotiable foundations for success out of the box:

Robust dependency management: It replaces the simplicity of local tools like Poetry with the production-grade power of PDM and uv, ensuring that every dependency is locked and every deployment is built from a fast, deterministic, and repeatable environment.
A pre-configured CI/CD pipeline: It comes with a ready-to-use continuous integration and deployment pipeline for Google Cloud Build, which automates the entire process of testing, building, and deploying your agent.
Multi-environment support: The template is pre-configured for separate staging and production environments, a best practice that allows you to safely test changes in an isolated staging environment before promoting them to your live users.

The process begins by using the official command-line tool to generate your project's local file structure. This prompts you to choose a base template; we used the "ADK Base Template" and then moved our agent logic into the newly created source code files (App) .

code_block: <ListValue: [StructValue([('code', '# Ensure pipx is installed\r\npip install --user pipx\r\n\r\n# Run the project generator to create the local file structure\r\npipx run agent-starter-pack create your-new-agent-project'), ('language', ''), ('caption', <wagtail.rich_text.RichText object at 0x7f4c39f8f250>)])]>

The final professional project structure:

code_block: <ListValue: [StructValue([('code', 'final-agent-project/\r\n├── .github/ # Contains the automated CI/CD workflow configuration\r\n│ └── workflows/\r\n├── app/ # Core application source code for the agent\r\n│ ├── __init__.py\r\n│ ├── agent_engine_app.py\r\n│ ├── orchestrator.py # The main agent that directs the workflow\r\n│ ├── company_researcher/ # Sub-agent for performing research\r\n│ ├── briefing_agent/ # Sub-agent for drafting emails\r\n│ └── tools/ # Custom tools the agents can use\r\n├── tests/ # Automated tests for your agent\r\n├── .env # Local environment variables (excluded from git)\r\n├── pyproject.toml # Project definition and dependencies\r\n└── uv.lock # Locked dependency versions for speed and consistency'), ('language', ''), ('caption', <wagtail.rich_text.RichText object at 0x7f4c39842f10>)])]>

With the local files created, the next step is to provision the cloud infrastructure. From inside the new project directory, you run the setup-cicd command. This interactive wizard connects to your Google Cloud and GitHub accounts, then uses Terraform under the hood to automatically build your entire cloud environment, including the CI/CD pipeline.

code_block: <ListValue: [StructValue([('code', '# Navigate into your new project directory\r\ncd your-new-agent-project\r\n\r\n# Run the interactive CI/CD setup wizard\r\npipx run agent-starter-pack setup-cicd'), ('language', ''), ('caption', <wagtail.rich_text.RichText object at 0x7f4c39e1c850>)])]>

2. Cloud Build
Once the setup is complete with the starter pack, your development workflow becomes incredibly simple. Every time a developer pushes a new commit to the main branch of your GitHub repository:

Google Cloud Build fetches your latest code.
It builds your agent into a secure, portable container image. This process includes installing all the dependencies from your uv.lock file, guaranteeing a perfect, repeatable build every single time.
It deploys this new version to your staging environment. Within minutes, your latest code is live and ready for testing in a real cloud environment.
It waits for your approval. The pipeline is configured to require a manual "Approve" click in the Cloud Build console before it will deploy that exact same, tested version to your production environment. This gives you the perfect balance of automation and control.

3. Deploy on Agent Engine and Cloud Run
The final piece of the puzzle is where the agent actually runs. Cloud Build deploys your agent to Vertex AI Agent Engine, which provides the secure, public endpoint and management layer for your agent.

Crucially, Agent Engine is built on top of Google Cloud Run, a powerful serverless platform. This means you don't have to manage any servers yourself. Your agent automatically scales up to handle thousands of users, and scales down to zero when not in use, meaning you only pay for the compute you actually consume.

Get started

Ready to build your own?

Explore the code for our Sales Intelligence Agent on GitHub.
Dive deeper with the Startups technical guide: AI agents.
Get started with production deployments using the Agent Starter Pack.
No matter where you are with AI adoption, we are here to help. Contact our Startup team today and you could get up to $350,000 USD in cloud credits with the Google for Startups Cloud Program,

^{The technical journey and insights detailed in this blog post were the result of a true team effort. I want to extend my sincere appreciation to the core collaborators whose work provided the foundation for this article: Luis Sala, Isaac Attuah, Ishana Shinde, Andrew Thankson, and Kristin Kim. Their hands-on contributions to architecting and building the agent were essential to the lessons shared here.}