Financial Services

How SAP Concur automates expense reporting with agentic AI

Fri, 10 Apr 2026 16:00:00 +0000

For decades, expense automation relied on a simple premise: If the machine can read the text, it can do the work. But anyone who has ever tried to scan a crumpled, smudged, or sun-bleached receipt from their pocket knows that reading isn't enough. When key data is missing, such as a city name or a clear date, the machine halts and the burden falls back onto the user for manual entry.

To close this gap, where traditional Optical Character Recognition (OCR) fails, SAP Concur’s engineering team set out to break new ground. While much of the industry was still focused on the design of conversational interfaces, SAP Concur foresaw a bigger shift. They recognized early on that the next leap in efficiency wouldn't come from better scanning, but from intelligent reasoning.

The result is an agentic AI upgrade for ExpenseIt, moving automation beyond simply reading text to solving messy logic puzzles, significantly reducing the need for manual intervention. Now, travelers can simply snap photos of their receipts as they receive them, upload digital scans, or forward receipts as emails, and ExpenseIt instantly transforms them into accurate expense entries with no date entry or itemization required.

Bringing this next-generation system called for a partner who could push the boundaries of innovation while matching the ambition to execute at startup speeds. SAP Concur fused its visionary roadmap with Google Cloud’s full-stack AI power, partnering with the only provider that co-designs every layer, from custom silicon and data platforms to world-class models and agents. Together, the teams engineered a true breakthrough in cost management — an AI agent that not only captures the receipt but intuitively understands the business traveler’s reality.

Speed, scale, and ingenuity

Standard expense automation is great at seeing what is on receipts but can’t see what is not there. SAP Concur saw the emergence of AI agents as an opportunity to create systems that could reason, decide, and act.

Suppose you upload a lunch receipt from “The Main St. Café,” which doesn’t include the address. In the past, this missing information would completely derail the automation and require you to manually enter this data to continue.

Agentic capabilities enable analyzing contextual clues, such as a vendor’s name, expense types, and trip itinerary data, to fill in the gaps. SAP Concur wanted to create an AI agent that could think like a human assistant: "I see 'Main St. Café.' I also see this transaction coincides with a business trip, where the user has a flight to Dallas and a hotel in Greenville, Texas. Therefore, this vendor is probably the restaurant located near the hotel in Paris, Texas — not Paris, France."

To solve this challenge, the teams approached the problem with a dynamic, startup-style mindset. Instead of a lengthy development cycle, the collaboration was defined by rapid prototyping and bold problem-solving.

Utilizing Google’s Gemini models, they built the Receipt Analysis Agent, underpinned by a cognitive architecture.

Here’s how it works:

Ingestion: The user snaps a photo in the SAP Concur mobile app, uploads a digital scan, or forwards a digital receipt as an email.
Deterministic core: SAP’s foundational technology, refined over decades of processing global expenses, applies finely tuned logic to lift the visible text on receipts with high precision.
Intelligent rRouting layer: If the scanned receipt data is clear, there’s no need to trigger additional actions. If the data is ambiguous (e.g., "Missing location"), the routing logic dynamically directs the task to the Receipt Analysis Agent.
Contextual reasoning: Built with Gemini models, the AI agent doesn’t just guess — it uses tools and grounding to infer missing information. ExpenseIt feeds the partial receipt data to the agent, alongside grounding data like the user’s travel itinerary and business calendar.
ReAct (Reason and Act framework): The Receipt Analysis Agent connects the dots, validating the vendor against the location history, and then completes the expense entry.

ExpenseIt with agentic AI (Receipt Analysis Agent)

Based on the example above, ExpenseIt identifies the receipt image as missing the location, and the intelligent routing layer triggers the Receipt Analysis Agent. Using Gemini, the agent will then identify what’s missing, analyze surrounding contextual clues and user-specific data, and make decisions based on information like travel bookings and calendar events.

Key design patterns for successful AI agents

The Receipt Analysis Agent was designed based on the core principles from Agentic Design Patterns, a hands-on guide written by senior Google engineer Antonio Gulli. This critical guidance helped SAP Concur successfully transform ExpenseIt into a system that can reason on data both inside and outside of receipts to accurately create expense entries.

First, the teams implemented the Routing Pattern to avoid running every receipt through the AI agent, helping to optimize for both cost and intelligence. A routing architecture classifies incoming tasks: Receipts with a high OCR confidence score are routed to the standard deterministic path, while those with low scores (e.g., “Missing location) are dynamically routed to the Receipt Analysis Agent.

Next, the Reflection Pattern is applied to solve issues like the Paris Paradox, ensuring the agent doesn’t just generate an answer like a basic chatbot. This pattern involves an internal generator-critic loop, where the model generates a hypothesis (“I think this is Paris, France”) and then acts a critic, checking it against established facts (“The itinerary says Dallas, Texas. This hypothesis is likely false.”).

Finally, the agent follows the Tool Use Pattern, providing explicit API access to grounding sources like trip itineraries from Concur Travel. This approach allows the agent to fetch the truth rather than hallucinating it, turning the system from a text generator to a factual researcher.

Architecting for ambiguity: Google Cloud’s ecosystem advantage

This project highlights a pivotal shift in intelligent system design. By combining a deterministic core with an agentic reasoning layer, SAP Concur demonstrated that AI’s highest value often isn't in processing the data we have, but in reasoning to find the data we are missing. A defining moment in this engineering journey was the shift in how the model was utilized. The teams moved beyond treating Gemini as a generative interface and instead deployed it as a logic engine.

Why did SAP Concur choose to build this future with Google Cloud? Because an agent is only as good as its understanding of the world — and no one understands the digital world like Google.

While this current release relies on the reasoning power of Gemini, the partnership opens the door to a future of multimodal, full-stack intelligence that’s unique in the market, including:

Real-world grounding: Imagine an agent that cross-references a receipt with Google Maps data to ensure the business actually exists at that location.
Frictionless flow: Future integrations could use Google Wallet to match transaction timestamps instantly, or Gmail to surface hotel folio receipts automatically.
Edge intelligence: With mobile advancements like Gemini Nano and the service system Android AICore, sensitive processing could eventually happen right on devices, giving users speed and privacy without the data ever leaving their phone.

SAP Concur has the deep domain expertise that powers the world’s financial transactions. Google Cloud brings the full AI stack from the custom-designed chips (TPUs) optimized for training, to the mobile OS in the user’s pocket.

Ready to build your next-generation agent?

You don't need to reinvent the wheel to build a reasoning engine like ExpenseIt. The architectural patterns discussed here — Routing, Reflection, and Tool Use — are codified directly in the Google Agent Development Kit (ADK). The ADK provides the frameworks and best practices to help you move from "prompt engineering" to "system engineering," serving as a blueprint for building agents that are reliable, scalable, and ready for the enterprise.

PayPal's historically large data migration is the foundation for its gen AI innovation

Thu, 26 Feb 2026 17:00:00 +0000

With the dawn of the gen AI era, businesses are facing unprecedented opportunities for transformative products, demanding a strategic shift in their technology infrastructure. A few years ago, PayPal, a digital-native company serving hundreds of millions of customers, faced a significant challenge. After 25 years of success in expanding services and capabilities, we’d created complexity in our data analytics infrastructure. Some 400 petabytes of data was spread across a dozen siloed systems due to limitations of scale and acquisitions of companies like Venmo, Braintree, and others.

Our very success in growth and innovation had created complexity that threatened our next evolution.

To continue leading the next wave of innovation in financial services, we knew we had to modernize our data foundation. Today, we’re proud to share how PayPal successfully completed what’s arguably one of the largest data migrations in history, culminating with the move of our analytics to BigQuery, Google Cloud’s enterprise data warehouse. This effort marks a significant leap in creating the robust data framework we’ll need to expand and advance our business priorities and meet the ever-evolving financial needs of our customers.

This migration was essential, but the scale was daunting. In fact, by some measures, such as our now sunset Teradata system, we believe this was one of the biggest data migrations in history. Befitting of such history, we wanted to offer some insights into how we tackled this migration and what others might consider when undertaking a significant migration of their own.

Untapped potential of data

As one of the original digital payment pioneers, PayPal processes billions of transactions, and houses decades of valuable customer insights. We have a mountain of data — really a mountain range — that had developed over decades without being fully leveraged in the service of our customers and merchants.

Each acquisition and new service added valuable capabilities but also introduced new data challenges. For example, a small business owner might use PayPal for online sales and Venmo for local transactions. However, providing a unified view of their business required complex processes that were costly and slow.

The fragmentation of data limited our ability to offer personalized experiences to consumers, thereby reducing the potential to maximize the value of their money and hindering our ability to gain deeper insights from the data.

As the gen AI era dawned, our digital fragmentation was becoming more than just a technical inconvenience. With AI becoming a transformative force in financial services with huge potential ROI, we knew fragmented data would severely limit our ability to create the intelligent experiences customers have come to expect. These could run from further strengthening our industry-leading fraud detection models to providing a best-in-class commerce platform for merchants to help them succeed in the competitive global economy.

To get there, we had to get our disparate data platforms in order, first.

Legacy systems, modern ambitions

The scope was massive. We needed to consolidate multiple data platforms, including what’s believed to be the world’s largest Teradata deployment, along with Hadoop clusters, Redshift, Snowflake, and various other systems processing petabytes of transaction data. This migration also had to be executed while maintaining the uninterrupted security and reliability our customers depend on.

As a technology company, PayPal has considerable internal resources, so we first had to decide whether to tackle this challenge ourselves. We weighed the costs and benefits and decided that if we were to unify and scale our on-premise infrastructure to meet our future needs, the cost and time-to-complete would have been prohibitive. Plus, the innovations in AI were happening at a rapid pace in the cloud. To truly leverage the power of our data, we needed to be where that innovation is happening.

We assessed various data warehousing solutions and chose BigQuery due to its numerous advantages. It is a fully managed, cloud native platform with disaggregated compute and storage that can scale independently. It has powerful capabilities at the scale and performance we needed, and a familiar SQL interface meant a gentler learning curve for our developer community.

Most importantly, BigQuery’s native integrations with AI enable seamless and efficient data analytics.

The journey to unified data

After choosing Google Cloud as our data partner, we embarked on our historic data migration. This may sound hyperbolic, but when you consider the scale of PayPal’s business, the geographies across which we operate, the regulations within each, the sensitive and quite literally valuable nature of this data, the scope of the challenge starts to be clear.

With the help of partners and experts from Google Cloud Consulting, we migrated more than 300 petabytes of data and streamlined operations, decommissioning around 25% of workloads. And we managed this all while maintaining zero downtime of our business operations and with no impact to customers. Here are some key factors that contributed to our success.

Alignment: The first hurdle in achieving transformations at scale is aligning stakeholders on a shared goal. So, we made it an enterprise-wide priority.

Discovery and analysis: Detailed inventories of data, workloads and inbound/outbound data streams is crucial for defining scope, effort and forecasting budget. Establishing lineage allowed us to trace the origins and relationships of various components, thereby providing a clear and comprehensive view of the dependency graphs.

Strategy: It is crucial to establish fundamental principles for the migration process, such as deciding between lift-and-shift versus modernization, defining security principles, setting governance guardrails, and determining how consumption will be tracked.

Execution: We automated every possible task and developed live dashboards to continuously monitor the progress of migrations. FinOps was integrated through the migration process with clear visibility of consumption and performance.

Benefits from BigQuery and beyond

We’ve achieve faster insights. Queries are 2.5x to 10x faster, including complex queries used by data scientists. This unlocks real-time insights, enabling PayPal to personalize product recommendations, offers, and customer support.

We’ve built new AI foundations. Data accessible for model training is 16x fresher. Feature engineering, a crucial step in AI development, is improved by instant access to clean, governed data. This accelerates the development personalized financial guidance, and predictive analytics for both consumers and businesses.

We’ve optimized operations. By migrating to BigQuery Data infrastructure vendors were reduced from four to one, streamlining operations and reducing complexity. Data duplication between platforms was entirely eliminated.

Our new unified data platform in BigQuery has become the source for PayPal's next wave of innovation, enabling us to create more intuitive, personalized experiences across our entire ecosystem and to leverage the power of gen AI.

AI-powered innovation unleashed

Looking ahead, we're exploring how this unified data platform will enable us to deliver AI-powered experiences that weren't possible before, including:

Predictive fraud prevention that spots potential issues before they affect our customers.
Personalized financial insights that help merchants optimize their businesses.
Seamless payment experiences that adapt to each customer's preferences and patterns.
More intelligent risk assessment that could help expand financial access to underserved communities.

Agentic commerce and future possibilities we are now able to imagine.

Lessons for the AI era

While our migration may be extraordinary in its scale, we are not alone in our needs or ambitions. There are ample considerations for companies within and well beyond financial services who may be pondering their own data foundations at this time.

First off, do not underestimate how under-utilized your data may be, and how unorganized.

Making sure your data is centralized, accurate, and consistent paves the way for AI experimentation and deployment. Organizations that spend time cleaning up their data fabric will be able to bring machine learning and generative AI applications to market more quickly, and do so at scale.

Second, ensuring data is accessible to everyone within your organization, with the proper controls, unlocks so much potential. Data orchestration and enterprise search, coupled with generative AI, has the potential to break down longstanding organizational silos and speed up decision-making across your organization. It’s one of the most promising applications of AI.

The financial world will continue to evolve, driven by new technologies and changing customer expectations. PayPal’s data transformation shows how even established companies can reinvent themselves to stay ahead of this change — provided they're willing to tackle the fundamental challenges that stand in their way.

In doing so, we've not only preserved our position as a digital payments pioneer but set ourselves up to continue leading the next wave of innovation in digital commerce.

Build financial resilience with AI-powered tabletop exercises on Google Cloud

Wed, 11 Feb 2026 17:00:00 +0000

In the financial sector, resilience isn't optional. Recent cloud outages have shown us exactly how fast critical data can disappear.

The risk is amplified by major regulatory drivers like the Digital Operational Resilience Act (DORA), which mandates that financial institutions are ready for any disruption. The recent designation of Google Cloud as a Critical Third-Party Service Provider (CTPP) under DORA further underscores this strong commitment to enabling secure and resilient financial operations for our customers.

Consider a major bank, with many critical apps processing thousands of transactions daily. For them, a critical incident means more than just downtime; it means regulatory fines and an erosion of client trust.

The problem: Traditional tabletop exercises fail in two key ways:

They rely on generic scenarios that do not reflect the complexity and unique weaknesses of the institution's actual production environment.
They often involve only IT or Compliance teams, failing to capture the cross-functional collaboration essential for real-world incident response.

In this blog, we’ll share a solution using context-aware scenario modeling on Gemini Enterprise.

Our solution? Context-aware scenario modeling powered by Google AI

Google Cloud’s Technical Account Management (TAM) team has pioneered a new approach to operational resilience testing that moves beyond textbook scenarios.

Our innovation is making these exercises dynamic and truly context-aware. Our team dives deep, ingesting and analyzing the customer’s actual operational information from different sources to build deeply customized, realistic scenarios. We look at everything from past support cases and meeting minutes to the application’s architecture and even billing metrics and SLAs.

Then, we use Gemini Enterprise to design a bespoke scenario, complete with a step-by-step timeline of exactly what should fail and when, plus the mitigation checks needed. Context-aware AI preparation can help FSI customers test their resilience against situations rooted directly in their own production environment, supporting their response strategy is fit for purpose.

High-level sample scenario timeline

To illustrate the realistic, evolving scenario designed by Gemini Enterprise, here is a high-level snapshot of a simulated incident. Note - the metrics are for illustrative purposes only.

Time (hh:mm)	Phase / Action	Customized context-aware event
T + 0:00	Initial anomaly	A spike in latency is detected on a critical transactional processing service. Logs show unusual API calls.
T + 0:15	Escalation / Discovery	Application operations reports that the primary customer database is showing a 150% increase in read errors.
T + 0:45	Critical impact	The App team confirms the errors are due to data corruption in a critical region. Regulatory alert issued.
T + 1:15	Remediation attempt	An Incident Manager attempts to execute the standard failover runbook, but the attempt fails due to a known, outdated configuration issue.
T + 2:00	Crisis point	Internal communication systems become slow due to load, forcing the team to use alternative means of communication (Testing communications protocol).

The simulated incident progression, from a specific service anomaly to a regulatory crisis point, is deeply informed by the customer’s actual environment and documented weaknesses, making the exercise intensely relevant.

The execution: A cross-functional emergency drill

In a recent simulation with a major FSI customer, this approach uncovered a dual critical incident involving massive latency and data corruption — a perfect stress test for their core systems.

The result was a real-time, interactive emergency drill that brought together the full spectrum of business decision makers. The diversity of participants was key to exposing gaps across technology, process, and communication.

The cross-functional simulation strategy led to a high-fidelity discussion, helping the customer uncover blind spots and refine its emergency response strategy in a safe, yet realistic, setting.

Impact & key results

We’ve successfully executed this AI-driven approach with large FSI customers across the DACH (Germany, Austria, Switzerland) region, and the impact has been immediate and measurable:

Practical steps: The exercise highlighted organizational strengths (e.g., cross-functional communication) and led directly to high-priority initiatives (e.g., implementing specific automated failover runbooks). Crucially, nearly all suggestions were quickly implemented because they were grounded in a real production-like scenario.
A shift in strategy: The realism was so impactful that many customers are now actively looking into integrating AI-based threat modeling into their existing compliance processes.
A lasting partnership: The AI-powered tabletop program, high-value service is now scheduled as a regular exercise with the Technical Account Managers, solidifying Google Cloud Consulting as a strategic collaborator in the customer's operational resilience journey.

Get started

Interested in moving beyond generic disaster drills and truly validating your organization's resilience?

Reach out to your Google TAM today to learn how our AI-powered tabletop exercises can validate your readiness.

A smart investment: FINRA builds a culture of improvement with DORA

Thu, 08 Jan 2026 17:00:00 +0000

FINRA, the Financial Industry Regulatory Authority, consistently seeks to achieve the highest standards in its technology practices. To elevate its software development lifecycle, FINRA — which oversees member broker-dealers — engaged Google consultants to help apply a metrics-driven methodology to its engineering practices.

DORA is a popular framework for helping organization improve software delivery performance through capabilities that can be measured by key metrics. These include deployment frequency, change lead time, change failure rate, failed deployment recovery time, and rework.

While FINRA had begun laying the groundwork to adopt DORA internally, the organization recognized an opportunity to accelerate implementation by tapping Google's firsthand experience.

Google conducted a discovery effort alongside technology leaders to identify opportunities for improvement. The recommendation that followed included increasing the existing focus on continuous improvement, adopting a user-centric approach to developing software and further enabling a generative culture within the department.

The implementation itself was deliberately flexible. Rather than recommending a one-size-fits-all approach, Google helped FINRA tailor its actions to individual team objectives. Teams prioritizing product value concentrated on lead time and deployment frequency metrics, while teams focused on stability concentrated on change failure rates and failed deployment recovery time.

Over the first year of implementation, engineering teams demonstrated continuous improvement across DORA capabilities, achieving a 9% per-developer productivity gain and reporting directionally positive developer experience feedback.

Sprint velocities also improved by 5%, enabling smaller engineering teams to deliver greater incremental product value to the business. Beyond raw metrics, teams also reported heightened transparency around delivery performance and appreciation for a standardized methodology.

Looking ahead, FINRA is maturing its DORA practice by providing more granular metrics tied to high-level DORA measurements, increasing emphasis on developer experience and correlating product metrics with software delivery performance indicators.

Want to discover what AI can do for governments, nonprofits, and other public sector organizations? Register to attend our upcoming Gemini for Government webinar on February 5, where we will dive deeper into the transformative technology powering the next wave of innovation across the public sector.

Using MCP with Web3: How to secure agents making blockchain transactions

Fri, 05 Dec 2025 17:00:00 +0000

At Google Cloud, we sit at a unique intersection of two transformative technologies: AI and Web3. The rise of AI agents capable of interacting with blockchains opens up a world of automated financial strategies, fast payments, and more complex scenarios like executing complex DeFi operations and bridging assets across multiple chains.

However, the practical viability of this new paradigm hinges on who hosts the agent, and who holds the private key to the operations.

The core issue is simple. Since most cryptocurrency users are not going to run their own secure servers to manage agent keys, providers are likely to turn to one of two primary architectures: A custodial model where users delegate funds to a third-party agent that controls a private key, and a non-custodial model where the agent only crafts transactions for the user to sign with its own private key.

Most of today’s examples showcase an agent directly holding a private key, and most cryptocurrency model context protocol (MCP) servers can only be used if you configure them with a private key. However, that may not be the only option.

The agent-controlled model

This model is designed for a world where users interact with agents hosted by a third party — a realistic assumption for mainstream adoption. In this scenario, you don’t give the agent your private key. Instead, the agent has its own key, and you give it an allowance to spend on your behalf.

How it works

Agent-controlled model sequence diagram.

Agent gets a wallet: The agent possesses one or multiple private keys. The private keys for these wallets are managed securely by the host of the agent, never by you.
User delegates funds: From your personal wallet (such as MetaMask, or a hardware wallet), you send a specific amount to the agent’s public address.
Agent gains autonomy: The agent now has full, autonomous control over the funds in the wallet it controls. The agent can use its key to sign and execute transactions — swapping tokens, buying NFTs, and paying another agent for data — until the pre-paid balance runs out.

The inherent risks

While this model provides automation, it introduces significant risks that shift from you to the agent and its host.

Performance risk: The agent could be bad at its job: A trading agent might execute a flawed strategy and lose the funds you delegated.
Malice risk: A poorly-designed or intentionally-malicious agent could misuse the funds. For example, the agent could send its balance to an unauthorized address. To prevent that scenario, the hosting platform should have robust safeguards, audits, and rules to constrain agent behavior. Another option is to ensure the agent funds are secure in a smart contract that guarantees how the funds are going to be used.
Security risk: The third-party host is now a custodian of your delegated funds. If their platform is hacked and the agents’ private keys are compromised, your pre-paid balance would be a primary target.

The self-hosted variant

A small minority of technically advanced users will want to run this model on a personal server. Because we’re in the nascent stages of AI agent development, this small group of developers and early adopters represents the current primary user base.

Consequently, this self-hosted model is the one most encountered in today's landscape, and it's what most crypto MCP servers are being built to cater for. In this case, it’s technically viable to give the agent a private key, because the key never leaves the user's own controlled environment.

Self-hosted agent model sequence diagram.

However, it also comes with a very high risk level. The private key can get hacked if your machine is compromised, and erratic, unauthorized agent behavior can lead to significant losses.

For example, if you say “I want to swap 500 USD for UNI”, the agent could decide to sell the UNI, or buy the UNI, mess up the slippage percentage, or buy the wrong UNI token. We recommend using this approach only for tests.

The transaction-crafter model

This is the non-custodial and fundamentally more secure alternative for most user interactions. Here, the agent never holds any of your funds. Its purpose is to do exactly the same thing as the agent-controlled model, but instead of signing and sending the key, the transaction is returned for the user to sign and send the key to the blockchain network.

How it works

Transaction crafter model sequence diagram.

User instructs agent: You ask the agent to perform a task, such as “swap my ETH for USDC.”
Agent crafts the transaction: The agent analyzes your query, and constructs the raw transaction, such as the swap transaction.
User signs the transaction: The agent passes this data back to you. Your wallet displays a pop-up showing exactly what you are about to do. Only you can approve and sign it with your private key.

How to build the agent with Google Cloud tools

To demonstrate this model, I built a sample agent using a suite of Google Cloud tools. The agent's reasoning is powered by the Gemini 2.0 Flash model and orchestrated using the Google Agent Development Kit (ADK). For testing, I acquired funds from the public Google Cloud Ethereum Faucet, a key resource for developers.

Developing agents using ADK is quite simple, and comes with useful features such as a web UI for simple testing and development environment, a powerful integration to Agent Engine and Google Cloud Run to easily deploy the agent in production, a simple way to run the agent as an API server for easy connection with a custom front end, and a toolbox to easily connect to MCP Servers, Agent-to-Agent (A2A) protocol, and tools such as Google search.

If you want to learn more about how to build agents using the Google Cloud stack, you can read this article.

The two main parts of this app are the agent that crafts transactions, and the front end that gets the crafted transaction from the agent and sends it to MetaMask for signature and sending.

The agent using Google ADK is quite simple to develop, the craft_eth_transaction function though can be quite complicated depending on the type of operations supported. These can include chains, assets, and swaps:

code_block: <ListValue: [StructValue([('code', 'from google.adk.agents import Agent\r\nfrom web3 import Web3\r\n\r\nETH_RPC_URL = "RPC URL"\r\n\r\n# (This is the tool function defined in the next section)\r\ndef craft_eth_transaction(to_address: str, amount: float, from_address: str, chain_id: int):\r\n # Step 1: Fetch the sender\'s next transaction count (nonce)\r\n # Step 2: Determine transaction type (ETH transfer or smart contract call)\r\n # Step 3: Construct the \'data\' field using ABI\r\n # Step 4: Assemble and return the final, unsigned transaction\r\n\r\n# The Agent is defined with a simple, non-custodial instruction\r\nroot_agent = Agent(\r\n name="transaction_crafter_agent",\r\n model="gemini-2.0-flash",\r\n description="An agent that crafts Ethereum transactions for a front-end to send via MetaMask.",\r\n instruction=(\r\n "You are an agent that crafts ETH transactions. "\r\n "Your only job is to collect the information from the user to craft Ethereum transactions.. "\r\n "The sender\'s address will be provided to you as context, along with the chain ID."\r\n "Use the `craft_eth_transaction` tool to generate the transaction object. "\r\n "The tool will return a JSON object that is ready to be sent to MetaMask. "\r\n "Leave gas and gasPrice fields empty; MetaMask will set them."\r\n "**IMPORTANT:** After using the tool, you must present the final transaction JSON in the response, formatted exactly like this:\\n"\r\n " ```json\\n"\r\n " {\\n"\r\n " \\"to\\": \\"0x...\\",\\n"\r\n " \\"from\\": \\"0x...\\",\\n"\r\n " \\"value\\": \\"0x...\\",\\n"\r\n " \\"nonce\\": \\"0x...\\",\\n"\r\n " \\"chainId\\": \\"0xaa36a7\\"\\n"\r\n " }\\n"\r\n " ```\\n"\r\n ),\r\n tools=[craft_eth_transaction],\r\n)'), ('language', ''), ('caption', <wagtail.rich_text.RichText object at 0x7f8a793e3490>)])]>

On the client-side, the logic is clean and focused on Web3 interactions. The front-end doesn't need to know anything about large-language models (LLMs) or agent orchestration. It calls the agent's API endpoint (hosted on Google Cloud Run), gets back a standard JSON transaction object, and passes it to MetaMask.

ADK's ability to easily run the agent as an API server provides this clean separation of concerns. The two main functions on the frontend are to extract the transaction from the agent’s response and how to send it to MetaMask. Here’s an example of those functions:

code_block: <ListValue: [StructValue([('code', '/**\r\n* Step 1: Process the agent\'s response to extract the crafted transaction data.\r\n* The agent\'s only output is a standard, unsigned transaction object.\r\n*/\r\nfunction extractTransactionFromAgentResponse(agentEvents) {\r\n const functionResponse = agentEvents.find(\r\n e => e.content?.parts?.[0]?.functionResponse?.name === \'craft_eth_transaction\'\r\n )?.content.parts[0].functionResponse;\r\n\r\n if (functionResponse?.response?.success) {\r\n // The raw transaction object, ready for the user\'s wallet\r\n return functionResponse.response.transaction;\r\n }\r\n return null;\r\n}\r\n\r\n/**\r\n* Step 2: Pass the crafted transaction to the user\'s wallet for execution.\r\n* This function triggers a MetaMask pop-up, putting the user in full control.\r\n*/\r\nasync function executeMetaMaskTransaction(txData) {\r\n if (!txData || typeof window.ethereum === \'undefined\') {\r\n console.error("Invalid transaction data or MetaMask not found.");\r\n return;\r\n }\r\n\r\n try {\r\n // The \'eth_sendTransaction\' call asks the wallet to sign and send.\r\n // The private key is never exposed to our web application.\r\n const txHash = await window.ethereum.request({\r\n method: \'eth_sendTransaction\',\r\n params: [txData], // txData is the JSON object from the agent\r\n });\r\n\r\n console.log(`Transaction sent successfully! Hash: ${txHash}`);\r\n return txHash;\r\n\r\n } catch (error) {\r\n // This error typically means the user rejected the transaction in MetaMask.\r\n console.error("Transaction failed or was rejected by user:", error);\r\n }\r\n}'), ('language', ''), ('caption', <wagtail.rich_text.RichText object at 0x7f8a793e3b50>)])]>

How the agent confirms intent

While the agent operation is the dialogue with the agent to reach a decision, the MetaMask pop-up is the conclusion of that conversation.

Think of it as the digital equivalent of your financial advisor explaining a strategy and then handing you the final document to sign. The signature is the deliberate, necessary confirmation that you understand and consent to the action. It can turn the agent’s recommendation into a reality with your explicit approval, providing crucial peace of mind. Especially given that an agent’s interpretation can vary wildly depending on the underlying LLM, the conversational context, and the data it has access to, it’s always good to check a wallet transaction twice before approval.

MCP servers should serve both realities

The vision of a future where agents autonomously pay other agents for services necessitates the agent-controlled model. Agents in this economy will need their own capital to operate.

However, the transaction-crafter model provides a secure bridge to that future. It can be used to safely fund an agent, or to simply execute one-off transactions for simpler operations. This flexibility is key.

From a developer’s perspective, adding this capability shouldn’t be a heavy lift. If a MCP server can already prepare and sign a transaction with a key it holds, it should be able to perform the same logic without the final signing step, returning the unsigned transaction instead. This minor change unlocks a much safer and flexible paradigm for users and can even enable more complex designs, like a dedicated “signer agent” in a multi-agent system.

Therefore, any robust MCP server designed for broad adoption should provide developers with the flexibility to build applications that can:

Advise and craft for secure, user-centric financial decisions.
Execute with delegated funds for specialized, automated, and clearly defined tasks.

We recommend pursuing this dual support to foster real innovation while protecting users.

You can learn more about building Web3 agents using Google Cloud here.

How CME Group builds a faster, smarter exchange on Cloud SQL

Wed, 03 Dec 2025 17:00:00 +0000

Editor’s note: The Chicago Mercantile Exchange (CME Group) has evolved from a nineteenth-century commodities exchange into one of the most advanced financial market infrastructures in the world. To support real-time trading and risk management at a global scale, the company launched a strategic partnership with Google Cloud. By migrating to Cloud SQL and adopting AI-powered insights, CME Group empowered developers, paid down technical debt, and unlocked new opportunities for data-driven innovation across financial markets.

From butter and eggs to bandwidth

CME Group is where risk meets opportunity. Every transaction that happens in our exchange — every order placed, trade executed, or risk calculated — relies on data moving flawlessly and instantly. The integrity of our markets depends on it.

Behind each of those trades is a database storing valuations, ownership, and so much more information, all of which can shift from millisecond to millisecond throughout the day. At our scale, those databases have to store and retrieve that information under relentless demand. We’re processing millions of messages a day with no margin for latency or error. That level of precision doesn’t come easily, especially in a highly regulated industry where performance has to coexist with security and reporting. Every change we make must align with strict compliance standards and global regulatory frameworks.

Speed has always been our currency, but scale became a challenge. CME Group's legacy database estate required significant engineering effort to maintain performance and meet regulatory demands. We needed to reduce operational overhead while improving our security posture. This required a managed database solution that offered transparent observability and clear compliance controls.

When Cloud SQL meets the trading floor

Our 10-year strategic partnership with Google Cloud aims to address this by migrating all our technology to the cloud, enabling us to innovate and collaborate on pushing the boundaries of what cloud infrastructure can support. Together, we’re experimenting with new ways to achieve ultra-low-latency performance in the cloud. As data volumes surge and AI becomes increasingly central to risk management, the ability to move and interpret information in milliseconds is a technical requirement. We’re building systems with Google Cloud that let us keep the market running, even as we lead it into the future.

With Cloud SQL, we’ve found a way to keep our data layer as fast and dependable as the markets we serve. Cloud SQL gives our teams real-time visibility into what’s happening inside the database. When an application slows, we can identify the root cause in minutes instead of hours. Those insights are built into the platform, which means we don’t need custom tooling or manual analysis to keep operations steady.

But for us, the value of Cloud SQL goes beyond performance tuning. It’s about confidence. Our database administrators can focus on strategic improvements, and our developers can validate and optimize queries without waiting for escalation. Taken together, we have faster troubleshooting and a data foundation ready for the always-on demands of global trading.

aside_block: <ListValue: [StructValue([('title', 'Build smarter with Google Cloud databases!'), ('body', <wagtail.rich_text.RichText object at 0x7f8a7935dd60>), ('btn_text', ''), ('href', ''), ('image', None)])]>

Cloud SQL is our new favorite teammate

The more we use Cloud SQL, the more it feels like we’ve added a new member to the team. AI-assisted insights from Cloud SQL have changed how the CME Group team works. When an application slows, Cloud SQL tells us why. It surfaces anomalies, walks us through guided analysis, and even suggests query optimizations that restore performance in minutes. Developers can see those recommendations right in their workflows, test fixes, then move on. No waiting, no hand-offs, no firefights.

In other words, AI-assisted troubleshooting has made performance management into a shared responsibility. And because Cloud SQL delivers a consistent experience, our teams can move seamlessly between environments. There’s less training – and a lot more collaboration. The end result is a smarter, more unified data culture at CME Group.

Performance is our competitive advantage

The work we’re doing with Google Cloud is about more than modernization. Every improvement in speed, reliability, and visibility translates directly into business confidence. CME Group can now deploy new features faster while maintaining the continuity our clients depend on.

Cloud SQL has given us a foundation for that agility. Fewer performance issues mean more time focused on innovation: expanding our analytics capabilities, accelerating AI initiatives, and exploring new ways to commercialize data responsibly. When you stop chasing outages, it turns out you have more time to take bigger bets and build the future.

For us at CME Group, performance has always been the product. Now, it’s also the platform. We’re building the infrastructure with Google Cloud that keeps global markets moving and the intelligence that will shape what comes next.

Learn more:

Sign up for the new Cloud SQL free trial, a dedicated 30-day program designed to give both new and existing Google Cloud users hands-on access to premium, enterprise-grade features of Cloud SQL (PostgreSQL and MySQL).
Download this IDC report to learn how migrating to Cloud SQL can lower costs, boost agility, and speed up deployments.
Learn how Ford and Yahoo gained high performance and cut costs by modernizing with Cloud SQL.

How Global Payments built a resilient architecture for scale with Cloud SQL

Thu, 30 Oct 2025 16:00:00 +0000

Editor’s note: When payments are your product, downtime isn’t an option. To meet the critical demands of global availability, Global Payments partnered with the Google Cloud product team to architect a resilient, multi-region solution leveraging Cloud SQL Enterprise Plus. This collaboration ensures enhanced uptime and streamlined disaster recovery. This strategic adoption of Google Cloud's advanced database capabilities empowers Global Payments to deliver always-on performance. In this blog, Principal Architect Govindaraj Palanisamy shares how his team is using Cloud SQL to deliver always-on performance.

At Global Payments, we power payment services around the world, across every kind of industry—from schools and hospitals to stadiums and gas stations. Whether it's a small daily purchase or a critical invoice, every transaction matters. When people go to pay their bills, buy a snack, or get through a turnstile, they expect things to just work—without delay. This commitment means powering seamless transactions, and our collaboration with Google Cloud on resilient architecture is key to delivering a superior customer experience and achieving scalability. That expectation becomes even more critical for our Tier 1 systems, like the portals that handle invoice and utility payments.These apps need to be up 24/7, with near-zero tolerance for downtime or data loss.

To meet those demands, we chose Cloud SQL Enterprise Plus edition, which provides the kind of performance, scalability, and business continuity we need for global operations. Today, we use Cloud SQL for multiple workloads, including managed Cloud SQL for SQL Server for high-priority transactions and Cloud SQL for PostgreSQL and Cloud SQL for MySQL for value-added services. This comprehensive adoption across all three key database engines underscores our deep confidence in Cloud SQL's capabilities and, by extension, in Google Cloud's robust platform.

aside_block: <ListValue: [StructValue([('title', 'Build smarter with Google Cloud databases!'), ('body', <wagtail.rich_text.RichText object at 0x7f8a79310ac0>), ('btn_text', ''), ('href', ''), ('image', None)])]>

High expectations, high availability
For our Tier 1 workloads, we require 99.99% uptime, rapid failover, and recovery point objectives (RPO) of under a minute. These are systems that can’t afford to go down, even for maintenance. With Cloud SQL Enterprise Plus, we get:

Near-zero planned downtime (often under 1 second)
Less than a minute RTO and zero RPO using multi-zone HA configurations with synchronous replication
Easy disaster recovery orchestration and testing using advanced disaster recovery switchover and write endpoints

For less critical but still important workloads, like merchant-facing services, we use similar configurations with slightly more flexible recovery windows. But in all cases, we ensure that data protection, performance, and compliance requirements—including the Payment Card Industry Data Security Standard (PCI DSS), GDPR, and the NIST Cybersecurity Framework— remain paramount.

Architecture built for resilience
Our current deployment spans three Google Cloud regions, with each region running web and application tiers in Google Kubernetes Engine (GKE) and connecting to Cloud SQL through a Cloud SQL Auth Proxy. Every Cloud SQL database is replicated across zones and regions. Read replicas support low-latency reads, and cascading replication helps route read traffic away from write-heavy nodes to balance the load.

Fig. 1 - Global Payments’ Architecture Diagram featuring Cloud SQL & GKE

That distributed topology gives us options. With a three-region deployment, we’re able to keep a failover-ready region available—even in the rare event of a dual-zone failure—so we can shift traffic without disruption. If we anticipate traffic spikes—for example, during a seasonal billing surge—we can scale our read replicas to maintain consistent performance. And with point-in-time recovery and backup retention decoupled from the database instance, we’re covered even if an instance is deleted.

Uptime well spent
The results of our Cloud SQL deployment have been substantial:

Near-zero downtime during maintenance
Consistent performance with read/write separation
Streamlined disaster recovery testing (including quarterly failover drills driven by compliance requirements)
Up to 60% reduction in operational overhead

We also use query insights for automated alerting and performance monitoring, giving us better visibility into key metrics as we scale.

Cloud SQL’s managed services free up our team to focus on innovation, not on managing database infrastructure and with Cloud SQL Enterprise Plus, we have the performance and availability guarantees our clients expect. We’re already exploring new capabilities like managed connection pooling and read pools, which promise to simplify scaling and reduce latency even further. These are the kinds of enhancements that let us keep growing without outgrowing our infrastructure.

At Global Payments, reliability is business critical. Cloud SQL helps us deliver it.

Editor’s note: Govindaraj Palanisamy originally shared Global Payments' story onstage at Google Cloud Next. During this session, Google Cloud also announced support for C4A virtual machines, powered by Google Axion, in Cloud SQL and AlloyDB. These VMs offer up to 65% better price-performance than current-generation x86 instances and up to 2x higher throughput than comparable Arm-based offerings. C4A is built to handle the scale, speed, and efficiency today’s enterprise databases demand.

Watch the full session or read the Axion announcement blog to learn more. And check out our web page to dive into the power of Cloud SQL.

Introducing an agentic commerce solution for merchants from PayPal and Google Cloud

Mon, 27 Oct 2025 16:30:00 +0000

Modern consumers demand a seamless, personalized shopping journey, from initial product discovery all the way to final purchase. With the rise of agentic AI, merchants now have an opportunity to deliver a truly assistive and cohesive experience across every touchpoint.

That’s why today, building on our goal of transforming commerce, PayPal and Google Cloud are thrilled to announce that we’re bringing agentic shopping experiences to life with a new offering that combines Google Cloud’s Conversational Commerce agent with payments powered by PayPal.

This combination will allow merchants to rapidly deploy agentic commerce experiences directly on their own digital surfaces to drive more consumer engagement, personalization, and conversion. Merchants are able to maintain full control over the agent’s tone, look, and the customer relationship.

How it works

The PayPal Agent will communicate securely with the merchant’s agent over the open Agent2Agent (A2A) Protocol, as well as being integrated with the Agent Payments Protocol (AP2) — a payments layer built on top of A2A and the Model Context Protocol (MCP) that provides trust, accountability, and fraud controls.

A2A Protocol is an open standard designed to enable AI agents to communicate, collaborate and delegate tasks to one another across organizations. AP2 provides a set of requirements, including Verifiable Digital Credentials, which secure agentic transactions.

Smooth, simple shopping journeys: The power of agent collaboration

With this new offering, merchants will have the option to adopt Google Cloud’s Conversational Commerce Agent or build their own agents using Google’s Agent Development Kit (ADK). Fully brand-compliant and acting as an intelligent sales associate for the merchant, the Conversational Commerce Agent is designed to engage shoppers in natural, human-like conversations, guiding them all the way from initial intent and product discovery to a completed purchase.

Once deployed, the merchant’s commerce agent can understand complex requests, suggest relevant products, answer questions, and personally assist the user through their shopping journeys. During product discovery and selection, the merchant’s commerce agent engages the PayPal Agent through A2A to provide context on the user's shopping history, based on permissioned data, to help improve product recommendations.

Once a customer is ready to check out, the PayPal Agent, in line with AP2, will provide a seamless and secure checkout experience within the conversational interface. The PayPal Agent can also surface payment method recommendations and check "buy now, pay later" eligibility. With the shopper’s consent, merchant agents will then connect to the PayPal Agent in an authenticated manner, and authorize the transaction on a trusted surface.

Consumer trust at the core

Agentic commerce holds massive opportunity, but also exposes potential challenges around control, risk, and fraud, which Google Cloud and PayPal are proactively addressing.

AP2 is an open protocol that’s payment-method agnostic, thanks to its development by Google in collaboration with more than 100 industry partners. AP2 provides a common, secure language for AI agents to transact on behalf of users, extending the core constructs of the A2A Protocol and MCP to establish the essential foundation for secure, accountable, and authorized commerce.

AP2 uses mandates — tamper-proof, cryptographically-signed digital contracts that provide verifiable proof of user intent. These mandates are signed by Verifiable Digital Credentials (VDCs), creating a non-repudiable audit trail.

For example:

Cart Mandate: The foundational credential used when the user is present to authorize a purchase. Cart Mandates are generated by the merchant and cryptographically signed by the user (typically via their device), binding authorization to a specific transaction.
Payment Mandate: A separate VDC shared with the payment network and issuer to provide visibility into the agentic nature of the transaction, helping the network and issuer build trust and assess risk. This credential contains signals for AI agent presence and the transaction modality (e.g. Human Present vs. Not Present).

Essentially, AP2 provides the critical foundation for trusted, agent-led payments, providing verifiable intent and establishing clear transaction accountability. Instead of inferring action, trust is anchored to deterministic, non-repudiable proof of intent from the user, which directly addresses the risk of agent error. Payment mandates act as the foundational evidence for every transaction, creating a secure, unchangeable audit trail that helps payment networks to establish clear and fair principles for accountability and dispute resolution.

For example, with PayPal’s AP2-compliant agent, merchants will be able to have the assurance that a user was present to authorize the payment. Instead of using APIs, it will connect agents using AP2, helping ensure users, merchants, and payment providers can confidently initiate and transact with agent-driven payments.

With today’s announcement, Google Cloud and PayPal are proud to work together to provide a largely out-of-the-box solution for merchants who want to deploy agentic commerce experiences without building the complex framework from scratch, all while owning the experience and relationship with the consumer. Building the solution using A2A and AP2 protocols ensures safety and security throughout the process.

To learn more, contact your Google Cloud sales representative or reach us here.

_{Disclaimer: The video shown in this post is for informational purposes only and contains forward-looking statements, projections, and assumptions. These are not guarantees of future performance, and actual results and experiences may vary.}

Google Cloud and AMD at STAC Summit NYC: H4D VMs for Finance

Wed, 22 Oct 2025 17:00:00 +0000

In capital markets, the race for low latency and high performance is relentless. That’s why Google Cloud is partnering with AMD at the premier STAC Summit NYC on Tuesday, October 28th! We’re joining forces to demonstrate how our combined innovations are tackling the most demanding workloads in the financial services industry, from real-time risk analysis to algorithmic trading.

H4D VMs for financial services

At the core of our offerings are the Google Cloud H4D VMs, now in Preview, powered by 5th Gen AMD EPYC processors (codenamed Turin).

The financial world operates at lightning speed, where every millisecond counts. The H4D VM series is purpose built to deliver the extreme performance required for high-frequency trading (HFT), backtesting, market risk simulations (e.g. Monte Carlo), and derivatives pricing. With its exceptional speed and efficiency of communication between cores, massive memory capacity, and optimized network throughput, the H4D series is designed to execute complex computations faster, reduce simulation times, and ultimately deliver a competitive edge.

H4D: Superior performance for financial workloads

To quantify the generational performance leap, we commissioned performance testing by AMD. They compared the new H4D VM directly against the previous generation C3D VM (powered by 4th Gen AMD EPYC processors), using the KX Nano open-source benchmark. This benchmark utility is designed to test the raw CPU, memory, and I/O performance of systems running data operations for kdb+ databases. These high-performance, column-based time series databases are widely used by major financial institutions, including investment banks and hedge funds, to handle large volumes of time-series data like stock market trades and quotes.

The results demonstrated a significant, out-of-the-box performance gain for the H4D series. With no additional system tuning, the H4D VM outperformed the C3D VM by an average of ~34% across all KX Nano test scenarios.

Figure 1: Per-core, cache-sensitive operations (Scenario 1) showed H4D's generational lead with a ~1.36x uplift in performance across all test types, confirming superior speed and efficiency of communication between cores and memory latency for key financial modeling functions. *1

Figure 2: Multi-core scalability with the number of processors set to the max core count and 1 kdb worker per thread (Scenario 2) delivered a ~1.33x performance uplift across all test types, demonstrating H4D's strong capability for parallel processing across all available cores. *2

Figure 3: For heavy, concurrent multi-threaded workloads with 8 threads per kdb+ instance and 1 thread per core (Scenario 3), H4D sustained substantial leadership, delivering relative gains of ~1.33x uplift across all test types. *3

These benchmark results demonstrate the H4D VMs are built to accelerate your most demanding, low-latency workloads, providing the performance required for high-frequency trading, risk simulations, and quantitative analysis.

A full spectrum of financial services solutions

The H4D VMs will be a major highlight for Google Cloud and AMD at the STAC Summit next Tuesday. Our booths will also showcase our full spectrum of solutions for financial institutions. Stop by to discuss how we can help optimize your entire technology stack, from data storage to advanced computation:

IBM Symphony GCE and GKE Connectors: Discover how to extend and manage your existing Platform Symphony grid compute environments by bursting jobs to Compute Engine or Google Kubernetes Engine (GKE).
Managed Lustre: Get extreme performance file storage for your most demanding HPC and quantitative workloads without the operational overhead.
GPUs and TPUs: Learn how our powerful accelerators can dramatically speed up machine learning, AI, and risk analysis tasks.
Cluster Director with Managed Slurm: Easily deploy and manage your HPC cluster workloads with our integration for the popular Slurm workload manager.

Come talk to experts!

We know that performance, security, and compliance are non-negotiable in financial services. Our team will be on site to discuss your specific challenges and demonstrate how Google Cloud, in partnership with AMD, provides the robust, high-performance foundation your firm needs to innovate and thrive.

We look forward to connecting with you at the Google Cloud and AMD booths at STAC Summit NYC on October 28th!

How AI can scale customer experience — online and IRL

Tue, 14 Oct 2025 21:00:00 +0000

Customer service teams at fast-growing companies face a challenging reality: customer inquiries are growing exponentially, but scaling human teams at the same pace isn’t always sustainable.

Intelligent AI tools offer a new path forward. They handle routine questions automatically so employees can focus on more complex customer service tasks that require empathy, judgment, and creative problem-solving.

LiveX AI enables businesses to build and deploy advanced AI systems that deliver natural conversational experiences at scale. These can show up as chat bots, call center agents — even 3D holographic personas in live settings.

To handle thousands of concurrent, real-time interactions with low latency requires infrastructure that is both powerful and elastic, especially when seamlessly escalating complex issues to human agents.

In this joint technical post, we'll share the technical blueprint LiveX AI uses to build and scale its intelligent customer experience systems on Google Cloud, demonstrating how the right combination of services makes this transformation possible.

Why this architecture matters: Proven ROI

This architecture delivers measurable business impact.

90%+ self-service rate for Wyze: Smart home leader Wyze deployed LiveX AI to achieve a 90%+ self-service rate, enabling their support team to focus on complex cases that require human expertise while improving the overall customer experience.
3x conversion for Pictory: The video creation platform Pictory saw a 3x increase in conversions by using LiveX AI to proactively engage and qualify website visitors.

These results are only possible through a sophisticated, scalable, and secure architecture built on Google Cloud.

Platform capabilities designed for scale

The LiveX AI platform is designed to be production-ready, enabling companies to easily deploy intelligent customer experience systems. This is possible through key capabilities, all running on and scaling with Google Cloud's Cloud Run and Google Kubernetes Engine (GKE):

AgentFlow orchestration: The coordination layer that manages conversation flow, knowledge retrieval, and task execution. It routes routine queries automatically and escalates complex issues to human agents with full context.
Multilingual by design: Built to deliver native-quality responses in over 100 languages, leveraging powerful AI models and Google's global-scale infrastructure.
Seamless integration: Connects securely to internal and external APIs, enabling the system to access account information, process returns, or manage subscriptions, giving human agents complete context when they step in.
Customizable knowledge grounding: Trained on specific business knowledge to ensure accurate and consistent responses aligned with team expertise.
Natural interface: Deployed via chat, voice, or avatar interfaces across web, mobile, and phone channels.

Figure 1: LiveX real-world 3D assistants

The technical blueprint: Building intelligent customer experience systems on Google Cloud

LiveX AI's architecture is intelligently layered to optimize for performance, scalability, and cost-efficiency. Here's how specific Google Cloud services power each layer.

Figure 2: LiveX AI customer service agent architecture on Google Cloud

The front-end layer

Managing real-time communication across web, mobile, and voice channels requires lightweight microservices that handle session management, channel integration, and API gateway services.

Cloud Run is the ideal platform for this workload. As a fully managed, serverless solution, it automatically scales from zero to thousands of instances during traffic spikes, then scales back down, so LiveX AI only pays for the computation they actually use.

The orchestration and AI engine

The platform's core, AgentFlow, manages the conversational state, interprets customer intent, and coordinates responses. When issues require human expertise, it routes them to agents with complete context. The system processes natural language input to determine customer intent, breaks down requests into multi-step plans, and connects to databases (like Cloud SQL) and external platforms (Stripe, Zendesk, Intercom, Salesforce, Shopify) so both AI and human agents have complete customer context.

Cloud Run for orchestration automatically scales based on request traffic, perfectly handling fluctuating conversational loads with pay-per-use billing.

GKE for AI inference provides the specialized capabilities needed for real-time AI:

GPU management: GKE's cluster autoscaler dynamically provisions GPU node pools only when needed, preventing costly idle time. Spot VMs significantly reduce training costs.
Hardware acceleration: Seamless integration with NVIDIA GPUs and Google TPUs, with Multi-Instance GPU (MIG) support to maximize utilization of expensive accelerators.
Low latency: Fine-grained control over specialized hardware and the Inference Gateway enable intelligent load balancing for real-time responses.

With this foundation, LiveX AI can serve millions of concurrent users during peak demand while maintaining sub-second response times.

The knowledge and integration layer

From public FAQs to secure account details, the knowledge layer provides all the information the system needs to deliver helpful responses.

The Doc Processor (on Cloud Run) builds and maintains the knowledge base in the vector database for the Retrieval-Augmented Generation (RAG) system, while the API Gateway manages configuration and authentication. For long-term storage, LiveX AI relies on Cloud SQL as the management database, while short-term context is kept in Google Cloud Memorystore.

Putting it all together

Three key advantages emerge from this design: elastic scaling that matches actual demand, cost efficiency through serverless and managed GKE services, and the performance needed for real-time conversational AI at scale.

Looking ahead: Empowering customer experience teams at scale

The future of customer service centers on intelligent systems that amplify what human agents do best: empathy, judgment, and creative problem-solving. Businesses that adopt this approach empower their teams to deliver the personalized attention that builds lasting customer relationships, freed from the burden of repetitive queries.

For teams evaluating AI-powered customer experience systems, this architecture offers a proven blueprint: start with Cloud Run for elastic front-end scaling, leverage GKE for AI inference workloads, and ensure seamless integration with existing platforms.

The LiveX AI and Google Cloud partnership demonstrates how the right platform and infrastructure can transform customer service operations. By combining intelligent automation with elastic, cost-effective infrastructure, businesses can handle exponential inquiry growth while enabling their teams to focus on building lasting customer relationships.

To explore how LiveX AI can help your team scale efficiently, visit the LiveX AI Platform.
To build your own generative AI applications with the infrastructure powering this solution, get started with GKE and Cloud Run.

The oracles of DeFi: How to build trustworthy data feeds for decentralized applications

Thu, 02 Oct 2025 05:00:00 +0000

Distributed ledger technology (DLT) emerged with Bitcoin as a censorship-resistant way to conduct payments between distrusting peers. After a period, traditional financial institutions began to explore the technology, recognizing the potential of its immutability, decentralization, and programmability to redesign financial instruments and workflows.

However, a foundational issue has stalled many enterprise blockchain projects at the pilot stage: the data integrity problem. Moving from controlled test environments to production systems introduces attack vectors and failure modes that don’t exist in traditional centralized systems – a challenge that is particularly pressing for institutions like DZ BANK that are pioneering enterprise DLT adoption.

To solve these challenges, DZ BANK and Google Cloud built an architectural solution for trustworthy data delivery to blockchain applications. This post describes how market data can be securely fed into DZ BANK’s Smart Derivative Contracts (SDCs) using Google Cloud technology.

2025 is the year of strategic engagement

Distributed ledger technology (DLT) has reached a maturity inflection point. Regulatory frameworks are stabilizing, scalability limitations are being addressed, and the Eurosystem's validation work demonstrates that smart contract protocols enable efficient settlement across separate DLT infrastructures. The exploratory phase is ending — institutions that haven't moved beyond pilot projects risk being left behind as competitors deploy production blockchain systems.

The technology's core value proposition — immutable, programmable, decentralized execution — enables new digital financial products that weren't previously feasible. Smart contracts can eliminate counterparty risk, automate complex settlement procedures, and enable peer-to-peer financial interactions without traditional intermediaries. But realizing these benefits requires solving a fundamental challenge that early blockchain systems were designed to avoid.

Eurosystem research shows that smart contract protocols enable efficient settlement for financial instruments and functional interoperability across separate DLT infrastructures. Based on these results, the Eurosystem recently announced plans to strengthen activities in this field, aiming to create a harmonized digital European financial market infrastructure.

aside_block: <ListValue: [StructValue([('title', '2025 ROI of AI in Financial Services'), ('body', <wagtail.rich_text.RichText object at 0x7f8a7a643280>), ('btn_text', 'Read now'), ('href', 'https://cloud.google.com/resources/content/roi-of-ai-financial-services'), ('image', None)])]>

The need for trustworthy data

While DLT was conceptualized as a technology where code is law, banks, asset managers, and clearinghouses require off-ledger data from external sources. This includes price feeds, interest rate feeds, KYC/AML attestations, legal event triggers, proofs of reserves, IoT sensor data, and payment confirmations.

Data trustworthiness is paramount in DLT systems. Wrong data can have unintended consequences in any system, but DLT transactions are often irreversible. A payment using an incorrect interest rate can be cancelled in traditional systems, while DLT transactions are typically final – especially when participants are pseudonymous and unreachable by legal interventions. This creates new attack vectors: attackers who manipulate off-ledger data can cause significant financial harm, including theft of on-chain assets.

Building trustworthy oracle architecture

Off-chain data is supplied to DLT systems via oracles –- interfaces that deliver external information to smart contracts. Trustworthy oracle services require addressing three key aspects:

Data must be correct at the source – underlying systems must produce accurate information.
Data must remain untampered during transit and processing.
Data must be delivered timely and reliably.

The combination of Google Cloud’s secure, highly available infrastructure with DZ BANK’s vision for standardized, deterministic financial protocols meets these requirements. Google’s global technical infrastructure provides security throughout the entire information processing lifecycle, ensuring data integrity and reliability at source and in transit. DZ BANK’s focus on technology-agnostic, deterministic patterns for financial instruments enables trustworthy, automatable financial innovations. Together this approach provides the foundation for delivering timely, untampered data to any DLT system and establishes scalable protocol standards for secure digital financial services.

This architectural pattern provides a blueprint for other institutions facing similar challenges, creating reusable components for trustworthy data delivery across different financial smart contracts and blockchain networks.

The Smart Derivative Contract: a DLT-based financial instrument relying on trustworthy data

The Smart Derivative Contract (SDC) use case covers a fully algorithmic financial product lifecycle that relies heavily on external data, serving as an archetype for oracle-based financial data feeds produced on Google Cloud and consumed by smart contracts. The deterministic settlement cycle requires robust oracle services to determine settlement values. Key functionalities include deterministic valuation, automated margining, netting, and trade termination handling to remove counterparty credit risk in OTC transactions. These processes depend on reliable real-time market data from oracles to determine net present value (NPV) and settlement amounts.

The underlying protocol is published as Ethereum Request for Comments (ERC) 6123. Based on this open protocol, DZ BANK has conducted several pilot transactions with binding legal effect and successfully validated the use case on Bundesbank’s DLT infrastructure (the Trigger Solution).

Why are derivatives the hardest test case? They require precise mathematical calculations using current market data to determine NPV for settlement. Interest rate swaps, for example, need current swap quotes to bootstrap discount and forward rate curves before calculating settlement amounts. The entire process must be deterministic and tamper-proof, with cryptographic evidence of data integrity throughout the pipeline. A secure oracle service is essential for the SDC lifecycle and automated settlement.

Technical foundation: security layers

The oracle system architecture addresses distinct threat models, each requiring different technical countermeasures:

Mitigating software supply chain issues

For SDCs, an attacker might tamper with code that determines NPV calculations – for example, modifying functionality to return artificially low values by default. This would cause settlement at incorrect amounts. To mitigate these issues, we follow Secure Software Supply Chain practices, leveraging Binary Authorization to enforce policies that only permit container images with valid attestation to be deployed. This attestation certifies that container images have passed required checks, such as build verification by trusted CI pipelines, generating build provenance. For maximum security, SLSA Level 3 assurance is desired. By verifying this attestation at deploy time, Binary Authorization blocks unauthorized or tampered images, preventing the malicious code from running.

Secure connections to data sources

Oracle functions calculating NPV need to access relevant market data, which could reside in cloud databases like Cloud SQL. The workload can connect via Private Service Connect, enabling access to the Cloud SQL instance using a private internal IP address within its own VPC network. This keeps all traffic within the Google Cloud network, allowing secure data access without traversing the public internet.

TEE attestation

Ensuring oracle data correctness requires proving that data was generated by specific untampered software versions. For SDCs, this applies to software responsible for value calculations. Using Confidential Space, a trusted execution environment, ensures that only authorized, unmodified software workloads can process data.

This is achieved through remote attestation, where data owners set access conditions based on verification of authorized workload attributes, including the specific digest of containerized software images. Code identity verification complements the inherent trust participants place in the oracle’s business logic. A verifiable attestation token provides strong assurance and can be packaged with the oracle data output to prove origin and correctness.

Transport Layer Security

Transport Layer Security (TLS) adds an additional layer by encrypting oracle data output during transmission to the blockchain. While TEE attestation proves data was generated by authorized, unmodified software within a secure environment, TLS protects data from interception or tampering during network transit.

Applications beyond derivatives

The architectural patterns developed for SDCs apply to other enterprise blockchain use cases that require trustworthy external data. Cross-chain asset transfers, for example, need reliable payment confirmation data to avoid double-spending attacks. Supply chain applications need verified sensor readings and logistics confirmations.

Secure cross-chain protocols are essential for financial interoperability, especially for settling asset and cash legs across separate networks. However, current protocols like HTLC rely on time-outs, creating security vulnerabilities. A more secure approach, proposed in ERC-7573, uses a stateless oracle that releases cryptographic keys contingent on payment success to either complete asset swaps or return funds. By decrypting keys as instructed by a smart contract, the oracle enhances both security and efficiency – an example of how: trustworthy off-chain oracles enable smart contracts.

Building production-ready blockchain infrastructure

The collaboration between DZ BANK and Google Cloud demonstrates that enterprise blockchain adoption is no longer limited by technology. Success depends on integrating decentralized applications with existing business systems while maintaining security and reliability standards.

For developers and architects working on enterprise blockchain projects, this collaboration provides both technical patterns to emulate and infrastructure components to leverage. The challenge isn't building blockchain applications — it's building blockchain applications that enterprises can trust with critical business processes.

Ready to explore how these architectural approaches can support your blockchain infrastructure requirements? The frameworks and patterns developed through this collaboration offer practical starting points for building trustworthy oracle systems that meet enterprise security and reliability standards.

The team would also like to thank Christian Fries at DZ BANK and Googlers Chris Diya, Yuriy Babenko, and Latif Ajouaoui for their contributions.

Deutsche Bank delivers AI-powered financial research with DB Lumina

Tue, 23 Sep 2025 16:00:00 +0000

At Deutsche Bank Research, the core mission of our analysts is delivering original, independent economic and financial analysis. However, creating research reports and notes relies heavily on a foundation of painstaking manual work. Or at least that was the case until generative AI came along.

Historically, analysts would sift through and gather data from financial statements, regulatory filings, and industry reports. Then, the true challenge begins — synthesizing this vast amount of information to uncover insights and findings. To do this, they have to build financial models, identify patterns and trends, and draw connections between diverse sources, past research, and the broader global context.

As analysts need to work as quickly as possible to bring valuable insights to market, this time-consuming process can limit the depth of analysis and the range of topics they can cover.

Our goal was to enhance the research analyst experience and reduce the reliance on manual processes and outsourcing. We created DB Lumina — an AI-powered research agent that helps automate data analysis, streamline workflows, and deliver more accurate and timely insights – all while maintaining the stringent data privacy requirements for the highly regulated financial sector.

“The adoption of the DB Lumina digital assistant by hundreds of research analysts is the culmination of more than 12 months of intense collaboration between dbResearch, our internal development team, and many others. This is just the start of our journey, and we are looking forward to building on this foundation as we continue to push the boundaries of how we responsibly use AI in research production to unlock exciting new innovations across our expansive coverage areas.” - Pam Finelli, Global COO for Investment Research at Deutsche Bank

Creating AI-powered research experiences

DB Lumina has three key features that transform the research experience for analysts and enhance productivity through advanced technologies.

1. Gen AI-powered chat

DB Lumina’s core conversational interface enables analysts to interact with Google’s state-of-the-art AI foundation models , including the multimodal Gemini models. They can ask questions, brainstorm ideas, refine writing, and even generate content in real time. Additionally, the chat capability supports uploading and querying documents conversationally, leveraging prior chat history to revisit and continue previous sessions. DB Lumina can help with tasks like summarization, proofreading, translation, and content drafting with precision and speed. In addition, we implemented guardrailing techniques to ensure the generation of compliant and reliable outputs.

2. Prompt templates

Prompt Templates offer pre-configured instructions tailored for document processing with consistent, high-quality outcomes. These templates enable analysts to facilitate the summarization of large documents, extraction of key data points, and the creation of reusable workflows for repetitive tasks. They can be customized for specific roles or business needs, and standardized across teams. Analysts can also save and share templates, ensuring more streamlined operations and enhanced collaboration. This functionality is made possible by Google’s long context window combined with advanced prompting techniques, which also provide citations for verification.

3. Knowledge

DB Lumina integrates a Retrieval-Augmented Generation (RAG) architecture that grounds responses in enterprise knowledge sources, such as internal research, external unstructured data (such as SEC filings), and other document repositories. The agent enhances transparency and accuracy by providing inline citations and source viewers for fact-checking. It also implements controlled access to confidential data with audit logging and explainability features, ensuring secure and trustworthy operations. Using advanced RAG architecture, supported by Google Cloud technologies, enables us to bring generative capabilities to enterprise knowledge resources to give analysts access to the latest, most relevant information when creating research reports and notes.

DB Lumina architecture

DB Lumina was designed to enhance Deutsche Bank Research’s productivity by enabling document ingestion, content summarization, Q&A, and editing.

Built on Google Cloud, the architecture leverages the following services:

Google Kubernetes Engine (GKE) for microservice orchestration
Cloud SQL with the pgvector extension for vector support
Cloud Storage for managing and storing unstructured data
Dataflow for document ingestion and embedding
Vertex AI for powering multimodal AI capabilities with Gemini
Discovery Engine API to enable RAG
Cloud Natural Language APIs for text and content moderation

All of DB Lumina’s AI capabilities are implemented with guardrails to ensure safe and compliant interactions. We also handle logging and monitoring with Google Cloud’s Observability suite, with prompt interactions stored in Cloud Storage and queried through BigQuery. To manage authentication, we use Identity as a Service integrated with Azure AD, and centralize authorization through dbEntitlements.

RAG and document ingestion

When DB Lumina processes and indexes documents, it splits them into chunks and creates embeddings using APIs like Gemini Embeddings API. It then stores these embeddings in a vector database like Vertex AI Vector Search or the pgvector extension on Cloud SQL. Raw text chunks are stored separately, for example, in Datastore or Cloud Storage.

These diagrams below show the typical RAG and ingestion patterns:

Overview of the agent.

When an analyst submits a query, the system then routes it through a query engine. A Python application leverages an LLM API (Gemini 2.0 and 2.5) and retrieves relevant document snippets based on the query, providing context that is then used by the model to generate a relevant response. The sources indicate experimentation with different retrievers, including one using the pgvector extension on Cloud SQL for PostgreSQL, and one based on Vertex AI Search.

User interface

Using sliders in DB Lumina’s interface, users can easily adjust various parameters for summarization, including verbosity, data density, factuality, structure, reader perspective, flow, and individuality. The interface also includes functionality for providing feedback on summaries.

An evaluation framework for gen AI

Evaluating gen AI applications and agents like DB Lumina requires a custom framework due to the complexity and variability of model outputs. Traditional metrics and generic benchmarks often fail to capture the needs for gen AI features, the nuanced expectations of domain-specific users, and the operational constraints of enterprise environments. This necessitates a new set of gen AI metrics to accurately measure performance.

The DB Lumina evaluation framework employs a rich and extensible set of both industry-standard and custom-developed metrics, which are mapped to defined categories and documented in a central metric dictionary to ensure consistency across teams and features. Standard metrics like accuracy, completeness, and latency are foundational, but they are augmented with custom metrics, such as citation precision and recall, false rejection rates, and verbosity control — each tailored to the specific demands and regulatory requirements of financial research and document-grounded generation. Popular frameworks like Ragas also provide a solid foundation for assessing how well our RAG system grounds its responses in retrieved documents and avoids hallucinations.

In addition, test datasets are carefully curated to reflect a wide range of real-world scenarios, edge cases, and potential biases across DB Lumina’s core features like chat, document Q&A, templates, and RAG-based knowledge retrieval. These datasets are version-controlled and regularly updated to maintain relevance as the tool evolves. Their purpose is to provide a stable benchmark for evaluating model behavior under controlled conditions, enabling consistent comparisons across optimization cycles.

Evaluation is both quantitative and qualitative, combining automated scoring with human review for aspects like tone, structure, and content fidelity. Importantly, the framework ensures each feature is assessed for correctness, usability, efficiency, and compliance while enabling the rapid feedback and robust risk management needed to support iterative optimization and ongoing performance monitoring. We compare current metric outputs against historical baselines, leveraging stable test sets, Git hash tracking, and automated metric pipelines to support proactive interventions to ensure that performance deviations are caught early and addressed before they impact users or compliance standards.

This layered approach ensures that DB Lumina is not only accurate and efficient but also aligned with Deutsche Bank’s internal standards, achieving a balanced and rigorous evaluation strategy that supports both innovation and accountability.

aside_block: <ListValue: [StructValue([('title', '2025 ROI of AI in Financial Services'), ('body', <wagtail.rich_text.RichText object at 0x7f8a7bb34340>), ('btn_text', 'Read now'), ('href', 'https://cloud.google.com/resources/content/roi-of-ai-financial-services'), ('image', None)])]>

Bringing new benefits to the business

We developed an initial pilot for DB Lumina with Google Cloud Consulting, creating a simple prototype early in the use case development that used only embeddings without prompts. Though it was later surpassed by later versions, this pilot informed the subsequent development of DB Lumina’s RAG architecture.

The project transitioned then through our development and application testing environments to our production deployment, eventually going live in September 2024. Currently, DB Lumina is already in the hands of around 5,000 users across Deutsche Bank Research, specifically in divisions like Investment Bank Origination & Advisory and Fixed Income & Currencies. We plan to roll it out to more than 10,000 users across corporate banking and other functions by the end of the year.

DBLumina is expected to deliver significant business benefits for Deutsche Bank:

Time savings: Analysts reported significant time savings, saving 30 to 45 minutes on preparing earnings note templates and up to two hours when writing research reports and roadshow updates.
Increased analysis depth: One analyst increased the analysis in an earnings report by 50%, adding additions sections by region and activity, as well as a summary section for forecast changes. This was achieved through summarization of earnings releases and investor transcripts and subsequent analysis through conversational prompts.
New analysis opportunities: DB Lumina has created new opportunities for teams to analyze new topics. For example, the U.S. and European Economics teams use DB Lumina to score central bank communications to assess hawkishness and dovishness over time. Another analyst was able to analyze and compare budget speeches from eight different ministries, tallying up keywords related to capacity constraints and growth orientation to identify shifts in priorities.
Increased accuracy: Analysts have also started using DB Lumina as part of their editing process. One supervisory analyst noted that since the rollout, there has been a noted improvement in the editorial and grammatical accuracy across analyst notes, especially from non-native English speakers.

Building the future of gen AI and RAG in finance

We've seen the power of RAG transform how financial institutions interact with their data. DB Lumina has proved the value of combining retrieval, gen AI, and conversational AI, but this is just the start of our journey. We believe the future lies in embracing and refining the “agentic” capabilities that are inherent in our architecture. We envision building and orchestrating a system where various components act as agents — all working together to provide intelligent and informed responses to complex financial inquiries.

To support our vision moving forward, we plan to deepen agent specialization within our RAG framework, building agents designed to handle specific types of queries or tasks across compliance, investment strategies, and risk assessment. We also want to incorporate the ReAct (Reasoning and Acting) paradigm into our agents’ decision-making process to enable them to not only retrieve information but also actively reason, plan actions, and refine their searches to provide more accurate and nuanced answers.

In addition, we'll be actively exploring and implementing more of the tools and services available within Vertex AI to further enhance our AI capabilities. This includes exploring other models for specific tasks or to achieve different performance characteristics, optimizing our vector search infrastructure, and utilizing AI pipelines for greater efficiency and scalability across our RAG system.The ultimate goal is to empower DB Lumina to handle increasingly complex and multi-faceted queries through improved context understanding, ensuring it can accurately interpret context like previous interactions and underlying financial concepts. This includes moving beyond simple question answers to providing analysis and recommendations based on retrieved information. To enhance DB Lumina’s ability to provide real-time information and address queries requiring up-to-date external data, we are planning to integrate a feature for grounding responses with internet-based information.

By focusing on these areas, we aim to transform DB Lumina from a helpful information retriever into a powerful AI agent capable of tackling even the most challenging financial inquiries. This will unlock new opportunities for improved customer service, enhanced decision-making, and greater operational efficiency for financial institutions. The future of RAG and gen AI in finance is bright, and we're excited to be at the forefront of this transformative technology.

How Mr. Cooper assembled a "team" of AI agents to handle complex mortgage questions

Thu, 18 Sep 2025 16:00:00 +0000

In today’s world where instant responses and seamless experiences are the norm, industries like mortgage servicing face tough challenges. When navigating a maze of regulations, piles of financial documents, and the high stakes of homeownership, consumers quickly find that even simple questions can turn into complicated issues. And the same can be true for the customer reps trying to help them navigate all that complexity.

Like many enterprises, Mr. Cooper is exploring how agentic AI and advanced AI agents can help both our customers and employees meet their needs with confidence. In our work to develop just such an agent with Google Cloud, one of our curious discoveries has been that like a good team, the best AI agents may just be made up of groups of agents with distinct skillsets and abilities, and we come to the best results when they’re working in concert.

At Mr. Cooper, our mission is to “Keep the dream of homeownership alive.” We're here to simplify the journey, provide clarity, and ensure our customers feel confident every step of the way. That confidence is key when they’re making one of the most consequential purchases, and decisions, of their lives.

With those dual goals of simplicity and certainty in mind, we partnered with Google Cloud Consulting to develop an agentic AI framework designed to complement and support our team. We call it the Coaching Intelligent Education & Resource Agent, or CIERA. We asked ourselves how to implement a chatbot that could effectively collaborate with our human agents to streamline both sides of the customer service experience.

And just as we prioritize hiring great groups of customer reps and mortgage agents, we’ve discovered how important it is to put together the right group of agents to effectively meet the needs of all our users. CIERA is designed to do exactly that, handling routine and time-consuming tasks to enhance efficiency, while empowering our people to focus on delivering what they do best — empathy, judgment and meaningful human connection.

CIERA represents an exciting step forward in blending human expertise with AI capabilities, creating a collaborative approach that elevates both the customer experience and our team's impact. And just as important as this work is for Mr. Cooper, CIERA also demonstrates how our multi-agent approach can serve as a model for companies across industries. Read on to learn how we did it, and how you can, too.

The challenge: Beyond the reach of traditional automation

Mortgage servicing is uniquely complex, where a customer might have a single question that requires an agent to cross-reference multiple documents.

This presents several challenges for traditional automation:

Siloed Knowledge: Scattered information makes it hard to see the full picture, but AI surfaces key data, helping agents make faster, smarter decisions for customers.

Lack of Understanding: Traditional systems rely on rigid keywords and decision trees, often missing the true intent behind customer inquiries. Our AI framework uncovers context and intent, equipping agents with the insights they need to respond with empathy and accuracy.

Inflexible Processes: When conversations take unexpected turns, legacy automation often fails, creating dead ends for customers and the team. AI provides real-time adaptive guidance, helping agents navigate these twists seamlessly.

To truly elevate the customer experience, we needed a solution capable of reasoning, orchestrating, and understanding context — one that enhances and amplifies our capabilities to deliver exceptional service.

The vision: Introducing CIERA, a collaborative AI agent workforce

Our vision was to create an agentic framework that supports our call center agents by leveraging Google Cloud’s Vertex AI platform. CIERA’s AI agents handle repetitive and complex tasks, allowing our team to focus on what technology can’t.

Guided by the principle that AI enhances human performance, these digital collaborators are designed to deliver accurate, comprehensive, and human-centered solutions.

Building the agent workforce: Our architectural blueprint
Our modular architecture assigns distinct roles to each AI agent, creating a scalable, efficient. and manageable system that seamlessly collaborates with people to make work smoother and more rewarding.

Meet the key players of our digital team and the solutions they deliver for team members and customers:

Sage, the Head Agent: Sage monitors how all other AI agents perform. By learning from patterns across workflows, Sage helps ensure that each AI agent works in harmony with human teams. Key abilities include intelligent agent monitoring, recognizing useful trends and fine-tuning orchestration.

Ava, the Orchestrator: Ava serves as the team’s coordinator, managing complex customer inquiries by breaking them into manageable tasks and assigning them to the appropriate AI assistants. While Ava doesn’t interact directly with customers, it ensures processes run smoothly, empowering human agents to remain central to delivering solutions.
Lex, the Task Specialist: Lex specializes in complex tasks, helping human agents during customer calls by quickly offering insights to questions around loan applications or escrow analyses. Working behind the scenes, Lex provides insights that allow people to focus on connecting with customers and making informed decisions.
Sky, the Data Specialist: Sky helps human teams navigate internal knowledge bases and FAQs. For questions about policies, procedures, or definitions, Sky provides accurate and timely information, freeing people to spend more time on meaningful interactions, rather than searching for data.
Remy, the Memory Agent: Remy assists by remembering past actions and outcomes, which helps personalize workflows and inform future decisions. Remy’s memory supports ongoing learning and training, making it easier for human agents to access shared knowledge and continuously improve their skills.
Iris, the Evaluation Agent: By evaluating confidence scores, detecting hallucinations, and grounding responses with Model Armor, Iris ensures consistency and authenticity, helping human agents provide trustworthy customer support.

A sample analysis performed by CIERA.

How it works in practice: A real-life scenario

Imagine a customer initiates a call asking, "I received a notice my escrow payment is increasing. Can you explain why and tell me what my new total payment will be?"

Instead of relying solely on automated responses, CIERA ensures every step is grounded in close partnership between AI agents and human team members:

Orchestration: Ava receives the query, understands the two distinct parts (the "why" and the "what"), and creates a plan. Ava consults with a human agent, confirms the correct context and then delegates tasks to the Lex agents.
Parallel Processing: With human oversight, Ava assigns the "why" task to Lex, pointing it to the customer's most recent escrow analysis document. Simultaneously, it tasks another Lex agent to calculate the new total payment based on data from our systems.
Synthesis: The Lex agent reads the document and reports back to the human agent: "The increase is due to a $200 annual rise in property taxes." The other agent confirms the new total payment. The human similarly reviews the payment calculation before moving ahead.
Resolution: Ava gathers all AI-generated insights, but the human agent validates and personalizes the final response as needed to ensure clarity, empathy, and accuracy before delivering it to the customer.

This human-in-the-loop approach ensures complex, multifaceted questions are resolved with both the efficiency of advanced AI and understanding nuances with the trust that only people can provide. The partnership guarantees every answer is not just quick, but also trustworthy and tailored to the customer’s needs.

Ensuring quality and trust: The "agentic pulse" and human oversight

In a regulated industry like ours, trust and accuracy are non-negotiable. Deploying advanced AI requires an equally advanced framework for evaluation and governance. To achieve this, we developed two key concepts:

The "Agentic Pulse" Dashboard: Our central command center for monitoring the health and performance of our agent workforce. Powered by model-based evaluation services within the Vertex AI platform , it goes beyond simple metrics. We track:
- Faithfulness: Is the agent's answer grounded in the source documents
- Relevance: Does the answer directly address the customer's question
- Safety: Does the agent avoid generating harmful or inappropriate content
- Business Metrics: How do we correlate these quality scores with classic KPIs like average handle time (AHT) and customer satisfaction (CSAT)
The "Sandbox" for HITL: Our "Sandbox" environment provides space for our business and technical teams to safely review, test and refine agent processes. Additionally, if the "Agentic Pulse" flags an interaction for review, a human expert can analyze the agent's reasoning and provide feedback, ensuring a continuous cycle of improvement and learning.

This robust governance framework gives us the confidence to deploy these powerful tools responsibly.

Example of a theoretical loan analysis assisted by CIERA.

Projected impact: From complex processes to clear wins

While CIERA is on its journey towards full production, our projections based on extensive testing and modelling point to historic and transformative gains across the board:

For our customers: We project a reduction in wait times and a higher rate of first-contact resolution, so customers can get answers quicker and with the benefits of round-the-clock support for many complex scenarios.

For our human agents: By automating tedious research, CIERA will free up our human agents to focus on sensitive and complex customer relationships that require a human touch and create better tools and resources for more engaging work.

For our business: We anticipate a major reduction in average handling times for a large segment of inquiries and faster, more accurate resolutions that are a direct driver of customer happiness and loyalty.

Beyond mortgages: A blueprint for any complex industry

The architectural patterns developed with CIERA are not limited to mortgage servicing. This agentic approach — of using an orchestrator to manage a team of specialized AI agents—is a powerful blueprint that can be applied to any industry, including healthcare, logistics and manufacturing, by grappling with information and task complexity.

A typical workflow with CIERA.

The future is agentic and collaborative

Our journey with CIERA is just beginning, but it has already solidified our belief that the future of customer service is agentic driven. By combining Mr. Cooper's deep industry expertise with Google Cloud's world-class AI infrastructure, we are not just building bots, we are cultivating a digital workforce.

This collaboration is about more than just lowering costs or improving efficiency — it's about building trust, delivering clarity, and creating a customer experience truly worthy of the dream of homeownership.

^{The team would like to thank Googlers Sumit Agrawal and Crispin Velez and the GSD AI Incubation team for their support and technical leadership on agents and agent frameworks as well as their deep expertise in ADK, MCP, and large language model evaluations.}

Setting new expectations: Benchmarking high-performance trading with C3 machines

Wed, 17 Sep 2025 10:00:00 +0000

Trading in capital markets demands peak compute performance, with every microsecond impacting critical decisions and market outcomes. At Google Cloud, we’re committed to providing global markets with the cutting-edge infrastructure they need to create and participate in digital exchange ecosystems. Our industry investments enable a purpose-built, cloud-native market infrastructure solution leveraging a global network that was built for security and scale, data, and AI capabilities. At the same time, we’re building industry-specific innovations that offer performant, scalable, and resilient environments for exchanges and trading participants, ultimately transforming how they access markets, utilize data, and manage risk.

Following our investments in optimized infrastructure for digital exchanges, we’ve continued to push the boundaries of what's possible in cloud-based trading.

The general availability of our C3 and C4 machine series, powered by the latest Intel Xeon Scalable processors and our custom Titanium offload network investments, represents a significant leap forward for latency-sensitive trading applications. In addition, Citadel Securities recently joined us as part of our Cloud WAN layer 2 solution that enables point-to-point connections over Google’s proprietary global network and complements our existing Network Connectivity Center layer 3 solution. Together, these offerings help trading participants achieve low-latency compute and connectivity for their globally distributed trading infrastructure.

Building on these foundational investments, we’re excited to announce new benchmarks specifically tailored for trading participants who require minimal latency and jitter with maximal throughput to handle the increasing market velocity of their trading infrastructures. In collaboration with 28Stone, a consultancy with expertise in capital markets technology and electronic trading solutions, we’ve tested and validated the performance of our C3 machine types to meet the needs of trading participants.

28Stone’s published report highlights that Google Cloud can achieve a round-trip trading decision in less than 50 microseconds at P99 across a range of compute profiles.

“As a 24x7 digital exchange focused on delivering institutional exchange level consistency in the cloud, we are excited to see that Bullish’s participants can immediately leverage what 28Stone and Google Cloud have publicly demonstrated: the technical capabilities to rapidly spot opportunity and engage with confidence.” - Alan Fraser, VP of Platform & Operations, Bullish

aside_block: <ListValue: [StructValue([('title', '2025 ROI of AI in Financial Services'), ('body', <wagtail.rich_text.RichText object at 0x7f8a7bba9bb0>), ('btn_text', 'Read now'), ('href', 'https://cloud.google.com/resources/content/roi-of-ai-financial-services'), ('image', None)])]>

The role of latency, jitter, and throughput

In today's hyper-competitive electronic trading landscape, the performance of underlying technology infrastructure is not just a contributing factor to success — it's fundamental. Three key performance metrics stand out for their impact on trading outcomes: latency, jitter, and throughput.

Latency: the race

In the context of trading, latency refers to the time it takes to receive, understand, and decide an action from a single datum. For trading systems, this means the time it takes for market data to reach the trading algorithm, for that algorithm to make a decision, and finally send a response back to the exchange. In a world of high-frequency trading (HFT) and algorithmic execution, single-microsecond delays can mean the difference between a profitable trade and a missed opportunity, a less favorable execution price, or getting filled on a resting order.

28Stone demonstrated latency for participants to make a straightforward trade decision — from receiving the ticks to processing a trade decision — of between 1.5µs and 3.5µs (P50 and P99, respectively) for normal replay speed of CME Group Equity market pcap files using open source Data Plane Development Kit (DPDK) network acceleration. Additionally, they demonstrated similar latency profiles with data rates increasing by up to 100x.

Jitter: the quest for consistency

Jitter refers to the variation in latency over time. While low latency is critical, high jitter — meaning unpredictable and inconsistent delays — can negatively impact trading performance. If the time it takes for an order to reach the exchange varies significantly, it becomes incredibly difficult to predict execution outcomes, manage risk, or implement trading strategies that rely on time certainty. If you were to use UberEats to order lunch but were told it would be delivered between 12 and 5pm, you would be unlikely to choose that option.

28Stone demonstrated that participants can expect their experience in Google Cloud to be uniform, regardless of volatile market conditions. Google Cloud infrastructure delivers low jitter, allowing trading algorithms to operate with a higher degree of certainty, leading to more stable and reliable market mechanics – as well as profitability from good trading signals.

As illustrated below for various C3 instance sizes using C++ with or without DPDK, the percentiles demonstrate strong consistent message performance. The increase in latency between each percentile demonstrates network and compute consistency measured in single and low tens of microseconds. The 28Stone report contains complete histograms for various configurations, allowing customers to see how to balance their specific latency and jitter requirements against the configurations’ cost profiles.

Throughput: handling the pressure of information

Throughput measures the amount of data that can be processed or transmitted within a given window of time (typically one second). In trading, throughput is a system's capacity to handle large volumes of market data updates, process numerous events simultaneously, and aggregate trade events efficiently — especially during periods of high market volatility or peak trading hours. Insufficient throughput can lead to data queues, order rejections, increased risk exposure, and an inability to keep pace with market activity.

The C3 machine series leverages Google Cloud's high-bandwidth networking capabilities, which include up to 200 Gbps per VM Tier 1 networking and services such as Cloud WAN. These machines are designed to provide the high throughput traders need to ingest vast streams of market data and execute on a large number of orders per second. The result is a trading system that performs optimally under strenuous load.

As highlighted, 28Stone tested increasingly higher replay speeds, resulting in higher bit rates that various trading systems may see in a variety of markets.

In summary, minimizing latency, reducing jitter, and maximizing throughput aren’t abstract technical pursuits but about consistency and certainty, akin to your lunch order arriving near your lunch break and not in the evening. Modern capital markets trading participants and digital exchanges demand these capabilities to enable market quality, fairness, and operational resilience.

Embracing new market dynamics with cloud

Financial markets are in a perpetual state of flux, characterized by evolving regulations, a proliferation of new asset classes (including those that trade 24/7 like FX and digital assets), and sudden, sharp spikes in trading volumes. In this dynamic environment, the ability to scale infrastructure rapidly, operate with flexibility, and manage costs effectively is not just an advantage — it's a prerequisite for survival and growth.

With Google Cloud, exchanges and participants enjoy several benefits:

Elastic scalability: Automatically scale resources up or down based on real-time demand. This means that during unexpected volume spikes — driven by market news, geopolitical events, or algorithmic trading activity — trading infrastructure can dynamically access additional compute power to maintain optimal performance. When volumes normalize, resources can be scaled back down, so that firms only pay for what they use.
Flexibility for continuous trading: Cloud infrastructure provides the resilience and availability that continuous 24/7 trading requires, as demonstrated daily by critical workloads for global 24/7 industry platforms like air travel, retail banking, media, and retail. Google Cloud's global network and multiple regions help ensure high uptime and fault tolerance, both critical for markets that never sleep. As seen with Deutsche Börse's development of a digital asset trading platform on Google Cloud, the architecture is designed for 24/7 availability and can be rolled out quickly to new markets. Google Cloud’s continuous operations allow firms to capitalize on opportunities around the clock without the massive investment and operational overhead of maintaining private data centers with equivalent N+1 redundancy globally.
Rapid engagement in new markets: Historically, expanding into new geographical markets or launching new asset classes involved lengthy and costly infrastructure build-outs. With Google Cloud's global regions, firms can deploy trading infrastructure in new regions in a fraction of the time. This agility allows for rapid market entry, thereby enabling businesses to seize new revenue streams and diversify their business with lower upfront investment costs and risk. Quickly provisioning and de-provisioning resources also means firms can experiment with new market strategies more freely, knowing they are not locked into long-term hardware commitments.

Want to see it for yourself?

By harnessing the scale, flexibility, and compelling cost-to-performance ratio of Google Cloud, including the powerful C3 and C4 family instances, trading participants can transform market volatility from a threat into an opportunity. They can confidently handle volume surges, support round-the-clock trading, and swiftly enter new markets, all while maintaining tight control over their operational expenditure and maximizing their competitive edge.

Want to apply these capabilities to the announced CME Group market migration to Google Cloud? Register with Google Cloud for notifications and engagements.

How Keeta processes 11 million financial transactions per second with Spanner

Thu, 14 Aug 2025 16:00:00 +0000

Keeta Network is a layer‑1 blockchain that unifies transactions across different blockchains and payment systems, eliminating the need for costly intermediaries, reducing fees, and enabling near‑instant settlements. By facilitating cross‑chain transactions and interoperability with existing payment systems, Keeta bridges the gap between cryptocurrencies and fiat, enabling a secure, efficient, and compliant global financial ecosystem.

Founded in 2022 and backed by Eric Schmidt, the former CEO of Google, Keeta has engineered its network to meet the stringent regulatory and operational requirements of financial institutions. Its on‑chain compliance protocols, including Know Your Customer (KYC) and Anti-Money Laundering (AML), ensure security and regulatory adherence. Keeta’s architecture also natively supports asset tokenization and digital identity, making it an ideal platform for stablecoins and real‑world asset transfers.

Recently, the company conducted a public, verified stress test of its network, which runs on Spanner, Google Cloud’s horizontally scalable, highly available operational database. The test demonstrated Keeta Network is capable of over 11 million transactions per second (TPS), significantly outperforming traditional layer-1 blockchains and opening new opportunities for what is possible with blockchain technology.

Keeta chose Spanner to power its distributed ledger due to its availability and elastic scalability, allowing the team to scale up or down as needed without downtime, costly over-provisioning, or risky manual administration. Google Cloud was also instrumental in helping to prepare and execute Keeta’s stress test, providing world-class infrastructure and technical guidance that helped validate the network’s real-world performance.

With Spanner’s fully managed operations and familiar relational developer experience, Keeta was able to focus on its network — not database infrastructure or distributed systems. At peak, Spanner handled 300,000 queries per second, reading and writing durable state to read balances, check permissions, resolve conflicts, and publish votes.

aside_block: <ListValue: [StructValue([('title', 'Try Google Cloud for free'), ('body', <wagtail.rich_text.RichText object at 0x7f8a7ba92520>), ('btn_text', 'Get started for free'), ('href', 'https://console.cloud.google.com/freetrial?redirectPath=/welcome'), ('image', None)])]>

Building one network, ready for anything

On its mission to be the blockchain that connects all networks, Keeta created a unified platform that serves as a common ground for all payment networks and assets. Keeta Network’s underlying architecture is built on a directed acyclic graph (DAG) structure. Unlike traditional blockchain architecture, a DAG can process transactions in parallel across many individual accounts, reducing latency and avoiding common bottlenecks that plague other existing solutions.

The network utilizes a two-step voting process to approve or deny operations. Each transaction must be verified by a set of voting representatives, which occurs prior to updating any ledger. Individual steps rely on Spanner’s ACID (atomicity, consistency, isolation, durability) transactions and strict external consistency to ensure correctness as well as durability in the event of an outage or a network partition.

Keeta Network is unbounded by design, enabling it to scale horizontally to handle the increasing demand of participants. Similarly, Spanner’s scale-out architecture allows for linear read and write scaling in dozens of regions globally, all while maintaining consistency and latency.

Furthermore, representatives can be configured to scale down as throughput requirements decrease. Spanner ensures that scaling up or scale down are always online operations, even under the heaviest load. By dynamically adjusting the size of its Spanner instances based on actual demand, Keeta saves money.

Live test results showing more than 10M transactions per second at peak

Testing Keeta’s performance in the real world

Keeta’s Test Network consisted of four representative nodes, each issuing votes on the network. To process the targeted number of transactions, more than 30 million synthetic accounts generated over 25 billion transactions, reading and writing data to Spanner instances in four separate regions. It’s important to note that adding additional representative nodes did not materially change the complexity of the confirmation process.

In order to put the network through its paces, the stress test utilized a “fan out” approach to demonstrate its parallel throughput and immense scale. One account was used to begin the process of distributing funds to every account. This initial source account created numerous blocks, each containing 20 transactions each, which were then used to fund an additional 60,000 to 120,000 accounts. Each of these accounts, in turn, sent additional transactions. This process was repeated many times to reach the 30 million total accounts used during the test.

Verified independently by Chainspect, the network successfully reached 11,122,116 transactions per second (TPS), far exceeding Keeta’s targeted goal of 10 million TPS.

Taking blockchain technology to new heights

By showcasing this scalability, Keeta is one step closer to its vision – connecting the fragmented global economy. Existing solutions lack the scalability required for traditional financial traffic, making it impossible to connect global finance. Spanner and Google Cloud provide Keeta peace of mind and a significant technical leg up, delivering an infrastructure that can grow at the same pace as its network without significant rebuilding or unpredictable costs.

Keeta shows that blockchain technology is now capable of improving critical operations like cross-border payments, point-of-sale transactions, and asset transfers of all types. To put the addressable market into perspective, consider this: Tens of trillions of dollars worth of value are transferred across outdated financial systems daily — and Keeta Network has proven it has the speed, scale, and security to be the foundation for a new, interconnected ecosystem.

Leaders in industries like financial services, retail, and entertainment and media already rely on Spanner to power their most critical operational workloads. Learn more about how Spanner can help take the stress out of your organization’s next growth milestone and set your development teams up for success.

How Wells Fargo is using Google Cloud AI to empower its workforce with agentic tools

Tue, 05 Aug 2025 12:30:00 +0000

Editor's note: Agentspace is now part of Gemini Enterprise. The agent creation and orchestration technology behind Agentspace is now powering the core functionalities of the Gemini Enterprise platform. Learn more.

The financial services industry has hit a technology tipping point. AI is fundamentally reshaping how people interact with their financial institutions, and this has forced banks to deliver unprecedented agility, efficiency, and personalization.

Today, Wells Fargo and Google Cloud are excited to announce an expansion of their strategic relationship, which will transform how Wells Fargo uses and deploys agentic AI at scale.

This expanded collaboration will equip Wells Fargo employees — including branch bankers, investment bankers, marketers, and customer relations and corporate teams — with AI agents and tools from Google Cloud.

Wells Fargo is an early adopter of Google Agentspace, Google Cloud's unified, secure platform to build, manage, and adopt AI agents at scale. Wells Fargo employees and teams will now be able to reach meaningful insights faster and unlock new levels of efficiency and innovation using agents. These agents will help them find and synthesize information faster, automate tasks and workflows, and increase organizational agility.

This collaboration marks a defining moment for agentic deployment in financial services, underscored by Wells Fargo's commitment to leveraging AI to enhance customer experiences, streamline operations, and cultivate a culture of innovation powered by its people.

aside_block: <ListValue: [StructValue([('title', 'Try Google Cloud for free'), ('body', <wagtail.rich_text.RichText object at 0x7f8a7a756280>), ('btn_text', 'Get started for free'), ('href', 'https://console.cloud.google.com/freetrial?redirectPath=/welcome'), ('image', None)])]>

Real-World Agentic Applications Across Wells Fargo

By providing easier access to information with agentic search and the ability to customize AI agents, Wells Fargo is helping its employees unlock significant new levels of efficiency and innovation across the entire bank.

For its corporate and investment bank, Wells Fargo is building agents to help employees answer, triage, and summarize complex foreign exchange post-trade inquiries, as well as navigate policies and procedures across internal data sources and systems. Once deployed, these tools can free up time for bankers and traders, allowing them to focus more intently on client relationships. Additionally, these agents can provide bankers with real-time market insights, ultimately enhancing the overall client experience.

Consider contract management, where Wells Fargo handles roughly a quarter of a million documents related to vendor agreements. A custom agent can rapidly query these extensive documents — identifying contracts with specific clauses, payment terms, contract types, and other important contract information, driving consistency and efficiency.

AI agents are set to revolutionize customer service for Wells Fargo, across all channels – including digital, branch, and call centers. Agents can automate routine tasks like balance inquiries and debit card replacements, significantly reducing wait times and freeing bankers to focus on complex tasks and deepening customer relationships. These agents also offer 24/7 hyper-personalized experiences at scale, analyzing vast datasets to provide tailored advice and product recommendations.

Additional capabilities that Wells Fargo will leverage using Agentspace include:

Deeper insights and efficiency: Google’s NotebookLM, offered in Agentspace, will be a cornerstone for various functions within Wells Fargo. This powerful tool lets employees upload websites, documents, presentations, spreadsheets, and other materials, which they can then easily query, interact with, and analyze. This can help with a range of tasks including sophisticated research; content creation; and generating summaries and actionable insights. Employees across the enterprise will leverage NotebookLM for quick, accurate answers, allowing them to work faster and smarter, ultimately freeing up valuable time to focus on more strategic work.

Intelligent information discovery: Moving beyond traditional keyword searches, Wells Fargo will deploy Google-quality, multimodal search. Employees will be able to conversationally interact with enterprise data, asking complex questions and receiving relevant, synthesized information from a vast array of internal sources; these could include employee handbooks, corporate policies, and operational tools. This allows for quick, accurate responses, whether for a nuanced policy query or details on current service offerings.

Responsible AI: a core tenet

Both Wells Fargo and Google Cloud are deeply committed to the responsible development and deployment of AI. This strategic collaboration is underpinned by rigorous ethical and regulatory frameworks so that these powerful tools are used in a way that promotes accuracy, fairness, transparency, accountability, and security.

Wells Fargo has established strong AI governance to align implementation with the company’s values and regulatory priorities, building trust with both employees and customers.

Transforming operations, accelerating innovation

Ultimately, the expanded collaboration is about Wells Fargo remaining at the forefront of the financial services industry through its embrace of new technology to better serve its customers. By enabling these advanced AI products internally first, Wells Fargo is building a strong foundation to accelerate the delivery of new capabilities and enhanced experiences to its customers.

The journey is just beginning, but the vision is clear: a future where generative AI empowers every employee at Wells Fargo, transforming how they work, collaborate, and serve customers. Together, Wells Fargo and Google Cloud are building that future.

Learn more about how Google Agentspace can help transform your operations, offerings, customer experience, and more.

Build a multi-agent KYC workflow in three steps using Google’s Agent Development Kit and Gemini

Mon, 16 Jun 2025 16:00:00 +0000

Know Your Customer (KYC) processes are foundational to any Financial Services Institution's (FSI) regulatory compliance practices and risk mitigation strategies. KYC is how financial institutions verify the identity of their customers and assess associated risks. But as customers expect instant approvals, FSIs face pressure to streamline their manual, time-consuming and error-prone KYC processes.

The good news: As LLMs get more capable and gain access to more tools to perform useful actions, employing a robust 'agentic' architecture to bolster the KYC process is just what FSIs need. The challenge? Building robust AI agents is complex. Google's Agent Development Kit (ADK) gives you essential tooling to build multi-agent workflows. Plus, combining ADK with Search Grounding via Gemini can help give you higher fidelity and trustworthiness for tasks requiring external knowledge. Together, this can give FSIs:

Improved efficiency: Automate large portions of the KYC workflow, reducing manual effort and turnaround times.
Enhanced accuracy: Leverage AI for consistent document analysis and comprehensive external checks.
Strengthened compliance: Improve auditability through clear reporting and source attribution (via grounding).

To that end, this post illustrates how Google Cloud's cutting-edge AI technologies - the Agent Development Kit (ADK), Vertex AI Gemini models, Search Grounding, and BigQuery - can be leveraged to build such a multi-agent KYC solution.

aside_block: <ListValue: [StructValue([('title', '$300 in free credit to try Google Cloud AI and ML'), ('body', <wagtail.rich_text.RichText object at 0x7f8a7a756760>), ('btn_text', 'Start building for free'), ('href', 'http://console.cloud.google.com/freetrial?redirectPath=/vertex-ai/'), ('image', None)])]>

Tech stack from Google Cloud

This multi-agent architecture we’ll show you today effectively utilizes several key Google Cloud services:

Agent Development Kit (ADK): Simplifies the creation and orchestration of agents. ADK handles agent definition, tool integration, state management, and inter-agent communication. It’s a platform and model-agnostic agentic framework which provides the scaffolding upon which complex agentic workflows can be built.
Vertex AI & Gemini models: The agents are powered by Gemini models (like gemini-2.0-flash) hosted on Vertex AI. These models provide the core reasoning, instruction-following, and language understanding capabilities. Gemini's potential for multimodal analysis (processing images in IDs or documents) and multilingual support further enhances the KYC process for diverse customer bases.
Search Grounding: The google_search tool, used by the Resume_Crosschecker and External_Search agents, leverages Gemini’s Google Search grounding capabilities. This connects the Gemini model's responses to real-time information from Google Search, significantly reducing hallucinations and ensuring that external checks are based on up-to-date, verifiable public data. The agents are instructed to cite sources (URIs) provided by the grounding mechanism, enhancing transparency and auditability.
BigQuery: The search_internal_database custom tool demonstrates direct integration with BigQuery. The KYC_Agent uses this tool early in the workflow to check if a customer profile already exists within the institution's internal data warehouse, preventing duplicate entries and leveraging existing information. This showcases how agents can securely interact with internal, structured datasets.

Deep dive: How to build a KYC agent in three steps

Our example KYC solution utilizes a root agent (KYC Agent) that orchestrates several specialized sub-agents:

Document Checker: Analyzes uploaded documents (ID, proof of address, bank statements, etc.) for consistency, validity, and potential discrepancies across documents.
Resume Crosschecker: Verifies information on a customer's resume against public sources like LinkedIn and company websites using grounded web searches.
External Search: Conducts external due diligence, searching for adverse media, Politically Exposed Person (PEP) status, and sanctions list appearances using grounded web searches.
Wealth Calculator: Assesses the client's financial position by analyzing financial documents, calculating net worth, and verifying the source of wealth legitimacy.

The root KYC_Agent manages the overall workflow, calling these child agents sequentially and handling tasks like checking if the customer is already present in internal databases and generating unique case IDs to track KYC requests.

Diagram showing the KYC Agent’s structure with sub-agents and tools

Step 1: Define your root agent (which receives the initial request from the user) and the child agents which handle the specialised tasks involved in the KYC process.

code_block: <ListValue: [StructValue([('code', '# kyc_agent/agent.py (Illustrative Snippet)\r\n\r\n# Child Agents Definitions (Simplified)\r\ndocument_checker_agent = Agent(\r\n model=MODEL, # e.g. gemini-2.0-flash-001\r\n name="Document_Checker",\r\n description=\'Analyses documents and finds discrepancies...\',\r\n instruction=instructions_dict[\'Document_Checker\'],\r\ngenerate_content_config=GenerateContentConfig(temperature=0.27),\r\n)\r\n\r\nresume_crosschecker = Agent(\r\n model=MODEL,\r\n name=\'Resume_Crosschecker\',\r\n description=\'Uses `google_search` tool for verifying resume...\',\r\n instruction=instructions_dict[\'Resume_Crosschecker\'],\r\n tools=[google_search], # Leverages Search Grounding\r\n generate_content_config=GenerateContentConfig(temperature=0.27),\r\n)\r\n\r\nexternal_search_agent = Agent(\r\n model=MODEL,\r\n name="External_Search",\r\n description=\'Uses `google_search` tool to find negative news...\',\r\n instruction=instructions_dict[\'External_Search\'],\r\n tools=[google_search], # Leverages Search Grounding\r\n generate_content_config=GenerateContentConfig(temperature=0.27),\r\n)\r\n\r\nwealth_calculator_agent = Agent(\r\n model=MODEL,\r\n name="Wealth_Calculator",\r\n description="Assesses the client\'s financial position...",\r\n instruction=instructions_dict[\'Wealth_Calculator\'],\r\n generate_content_config=GenerateContentConfig(temperature=0.27),\r\n)\r\n\r\n# Wrap Resume_Crosschecker Agent\r\nresume_crosschecker_tool = AgentTool(agent=resume_crosschecker_agent)\r\n\r\n# Wrap External_Search Agent\r\nexternal_search_tool = AgentTool(agent=external_search_agent)\r\n\r\n# Root KYC Agent orchestrating the workflow\r\nroot_agent = Agent(\r\n model=MODEL,\r\n name="KYC_Agent",\r\n description="KYC Onboarding Assistant",\r\n # Add the AgentTool wrappers to the tools list, alongside the original tools\r\n tools=[\r\n generate_case_id,\r\n search_internal_database,\r\n resume_crosschecker_tool, # AgentTool\r\n external_search_tool # AgentTool\r\n ],\r\n sub_agents=[\r\n document_checker_agent,\r\n wealth_calculator_agent\r\n ],\r\n generate_content_config=GenerateContentConfig(temperature=0.27),\r\n instruction=instructions_dict[\'KYC_Agent\'], # Instructions should still guide the LLM to call the tools by name\r\n global_instruction=\'You will always give detailed responses and follow instructions\'\r\n)'), ('language', 'lang-py'), ('caption', <wagtail.rich_text.RichText object at 0x7f8a7a781c10>)])]>

Step 2: Define the tools needed by your agents in order to perform their respective tasks

code_block: <ListValue: [StructValue([('code', '# kyc_agent/custom_tools.py (Illustrative Snippet)\r\n\r\ndef search_internal_database(input_name: str) -> Dict[str, Any]:\r\n """\r\n Finds names in an internal BigQuery table...\r\n """\r\n try:\r\n client = bigquery.Client(project=PROJECT_ID)\r\n query = f"""\r\n SELECT `Full Name`, `UID`, `Risk Level`, `Citizenship`, `Networth`\r\n FROM `{TABLE_NAME}` # Defined in constants.py\r\n WHERE LOWER(`Full Name`) LIKE LOWER(\'%{input_name}%\')\r\n """\r\n query_job = client.query(query)\r\n results = query_job.result()\r\n df = results.to_dataframe()\r\n return df.to_dict(\'records\')\r\n except Exception as e:\r\n error_message = f"An error occurred with BigQuery: {e}"\r\n # Handle errors, potentially fallback to alternate data source\r\n # Fallback logic would go here if needed\r\n return {"error": error_message}'), ('language', 'lang-py'), ('caption', <wagtail.rich_text.RichText object at 0x7f8a7a781a90>)])]>

Step 3: Run your agent locally using the command “adk web”. ADK provides a built-in UI for developers to visualise and debug the agent during the development process:

Screenshot of the ADK Dev UI used for developing agents

Start building now

This multi-agent KYC architecture demonstrates the power of combining ADK, Gemini, Search Grounding, and BigQuery. It provides a blueprint for building intelligent, automated solutions for complex business processes.

Learn more: Dive deeper into the technologies used:

Build your own: Adapt this pattern to your specific KYC requirements and integrate it with your existing systems on Google Cloud using services like Cloud Run for deployment.
Contact us: Reach out to Google Cloud Sales for a deeper discussion on implementing AI-driven KYC solutions tailored to your organization.

By embracing a multi-agent approach powered by Google Cloud's AI stack, FSIs can transform their KYC processes, achieving greater efficiency, accuracy, and compliance in an increasingly digital world.

How Alpian is redefining private banking for the digital age with gen AI

Tue, 03 Jun 2025 16:00:00 +0000

As the first fully cloud-native private bank in Switzerland, Alpian stands at the forefront of digital innovation in the financial services sector. With its unique model blending personal wealth management and digital convenience, Alpian offers clients a seamless, high-value banking experience.

Through its digital-first approach built on the cloud, Alpian has achieved unprecedented agility, scalability, and compliance capabilities, setting a new standard for private banking in the 21st century. In particular, its use of generative AI gives us a glimpse of the future of banking.

The Challenge: Innovating in a Tightly Regulated Environment

The financial industry is one of the most regulated sectors in the world, and Switzerland’s banking system is no exception. Alpian faced a dual challenge: balancing the need for innovation to provide cutting-edge services while adhering to stringent compliance standards set by the Swiss Financial Market Supervisory Authority (FINMA).

Especially when it came to deploying a new technology like generative AI, the teams at Alpian and Google Cloud knew there was virtually no room for error.

Tools like Gemini have streamlined traditionally complex processes, allowing developers to interact with infrastructure through simple conversational commands. For instance, instead of navigating through multiple repositories and manual configurations, developers can now deploy a new service by simply typing their request into a chat interface.

This approach not only accelerates deployment times — reducing them from days to mere hours — it's also empowered teams to focus on innovative rather than repetitive tasks.

There are limits, to be sure, both to ensure security and compliance, as well as focus on the part of teams.

Thanks to this platform with generative AI, we haven’t opened the full stack to our engineers, but we have created a defined scope where they can interact with different elements of our IT using a simplified conversational interface. It’s within these boundaries that they have the ability to be autonomous and put AI to work.

Faster deployment times translate directly into better client experiences, offering quicker access to new features like tailored wealth management tools and enhanced security. This integration of generative AI has not only optimized internal workflows but also set a new benchmark for operational excellence in the banking sector.

aside_block: <ListValue: [StructValue([('title', 'Try Google Cloud for free'), ('body', <wagtail.rich_text.RichText object at 0x7f8a7a6b6820>), ('btn_text', 'Get started for free'), ('href', 'https://console.cloud.google.com/freetrial?redirectPath=/welcome'), ('image', None)])]>

A Collaborative Journey to Success

Alpian worked closely with its team at Google Cloud to find just the right solutions to meet it's evolving needs. Through strong trust, dedicated support and expertise, they were able to optimize infrastructure, implement scalable solutions, and leverage AI-powered tools like Vertex AI and BigQuery.

"Google Cloud’s commitment to security, compliance, and innovation gave us the confidence to break new ground in private banking,” Damien Chambon, head of cloud at Alpian, said.

Key Results

Alpian’s cloud and AI work has already had a meaningful impact on the business:

25% faster feature deployment, ensuring quicker time-to-market for innovative banking products.
Enhanced developer productivity with platform engineering, enabling more independence and creativity within teams.
Automated compliance workflows, aligning seamlessly with FINMA’s rigorous standards.
Simplified deployment processes, reducing infrastructure complexity with tools like Gemini

These achievements have enabled Alpian to break down traditional operational silos, empowering cross-functional teams to work in harmony while delivering customer-focused solutions.

Shaping the Future of Private Banking

Alpian’s journey is just beginning. With plans to expand its AI capabilities further, the bank is exploring how tools like machine learning and data analytics can enhance client personalization and operational efficiency. By leveraging insights from customer interactions and integrating them with AI-driven workflows, Alpian aims to refine its offerings continually and remain a leader in the competitive digital banking space.

By aligning technological advancements with regulatory requirements, Alpian is creating a model for the future of banking — one where agility, security, and customer-centricity can come together seamlessly and confidently.

10 months to innovation: Definity's leap to data agility with BigQuery and Vertex AI

Thu, 13 Mar 2025 16:00:00 +0000

At Definity, a leading Canadian P&C insurer with a history spanning over 150 years, we have a long tradition of innovating to help our customers and communities adapt and thrive. To stay ahead in our rapidly evolving industry, we knew a unified data foundation was key to realizing the business and customer experience opportunities offered by modern analytics and AI.

While our legacy on-premises Cloudera platform had served us well, it could no longer support our growing needs for scale, innovation, and harnessing the power of data and AI. So, we embarked on a critical mission: modernizing our data infrastructure.

Legacy limitations stifling innovation

We faced a combination of interconnected challenges, which impact many organizations today:

Limited scalability and AI/ML workload support: Our existing infrastructure, constantly running at 80% utilization, was stretched thin. Processing billions of daily events for real-time analytics and scaling AI and ML workflows was a constant battle, limiting our ability to gain timely insights and develop innovative, data-driven products and experiences.
Data silos, fragmented insights: Our data resided in various systems, creating a fragmented view of our business. This made it difficult to get a holistic understanding of our customers and hindered initiatives like building a comprehensive customer 360º view and delivering personalized recommendations at a moment of relevance.
Escalating costs: Maintaining and scaling our Cloudera platform, which hosted massive data volumes (200TB compressed, 1PB uncompressed), was increasingly expensive and diverting valuable fiscal and people resources away from strategic priorities.

Faced with these pressing issues, the timing of our next renewal presented a strategic window of opportunity. We had a critical decision to make — migrate both technology and business platforms within 10 months or invest in upgrading our legacy Cloudera environment.

aside_block: <ListValue: [StructValue([('title', '$300 in free credit to try Google Cloud databases'), ('body', <wagtail.rich_text.RichText object at 0x7f8a7a65c670>), ('btn_text', 'Start building for free'), ('href', 'http://console.cloud.google.com/freetrial?redirectPath=/products?#databases'), ('image', None)])]>

Building a unified data and AI platform with Google Cloud

We chose Google Cloud and its powerful duo, BigQuery and Vertex AI, to build the Strategic Data Platform (SDP) — our new modern data analytics platform. BigQuery's serverless architecture, unmatched scalability, built-in ML capabilities, and seamless integration with Vertex AI made it the ideal solution to power our data-driven transformation.Our migration was a remarkably fast-paced effort, carried out in close collaboration with with Google Cloud and Quantiphi, a Google Cloud partner.

Like many enterprises, we adopted a hybrid approach. We retained Databricks on Google Cloud for specific ETL workloads, utilizing Quantiphi's expertise in converting legacy systems. At the same time, we migrated the bulk of our data processing to BigQuery for optimal cost-efficiency and performance. We also used Cloud Composer to orchestrate our complex data pipelines and ensure secure, private connectivity within our Google Cloud environment, a crucial requirement for handling sensitive customer data.

As a result, our dedicated team of over 100 Definity employees completed the migration in just ten months — 50% faster than the industry average. This rapid transition was aided by innovative tools, such as the "nifi-migration" solution built by Google Cloud Consulting. This open-source tool provided a visual and highly configurable way to automate real-time data flow between different systems, minimizing disruption and helping us surpass our initial migration timeline expectations.

Our CTO, Tatjana Lalkovic, who championed this effort to consolidate our structured and multimodal data to accelerate our AI/ML use cases, shared her perspective on the impact of our decision, saying:

“As we reimagined where data and AI could take our business, industry, and customer experience, Google Cloud BigQuery and Vertex AI stood out as modern, enterprise-ready serverless solutions prepared to meet the AI moment — not just today but for the foreseen future. The speed and success of this migration has created a lot of trust in our partnership and has been a significant boost to our digital transformation to streamline operations, improve products, and better serve customers.”

Strategic Data Platform - High level design on Google Cloud

Transforming insurance with data and AI

The results of our migration to BigQuery and Vertex AI has been transformative for Definity. We’ve seen exceptional user satisfaction, with the SDP achieving a remarkable Net Promoter Score (NPS) of 9.9 out of 10. The move has also saved us millions on our annual spend on non-strategic technologies and delivered a roughly 75% reduction in planned downtime for our digital platforms.

Performance has also dramatically improved, with processes to gain insights that once took days now completing in an average of 4.5 hours. Moreover, migrating to Google Cloud has helped us increase agility and innovation. We’re now able to double our business releases per year — achieving a 30% increase in testing automation, a 63% improvement in deployment time, and a 10x faster infrastructure setup.

By combining structured and unstructured data in BigQuery, we’ve unlocked new analytical possibilities and improved price-performance. This unified data foundation has empowered our business intelligence tools with richer, more comprehensive data, leading to more informed business decisions. The seamless integration with Vertex AI has enabled us to develop, deploy, and scale AI models, driving innovation in areas like fraud detection, automated intake, and personalized call center experiences. At the same time, we benefit from Google Cloud’s strong commitment to data security and privacy, helping us to strengthen our security posture and keep our customers safe.

As our VP of Data, Ron Mills, said:

"BigQuery's serverless architecture has been a game-changer. The 'nothing to manage' approach is a huge differentiator. For enterprises like us that are migrating from on-prem clusters constantly running at 80% capacity, it's like night and day."

Lessons learned from our migration journey

Migrating a core data platform is a significant undertaking, and we’ve learned a lot along the way. For other organizations considering the same journey, here are some key takeaways from our experience:

One team, one goal: Foster a collaborative environment where technology and business teams, vendors, contractors, and consultants work together seamlessly towards a shared objective.
Leadership trust and commitment: Executive leadership trust in the delivery team's decision-making is crucial for maintaining momentum and navigating challenges.
Be bold: Don't be afraid to think outside the box, make timely decisions, and be prepared to adapt quickly to unforeseen setbacks.
Plan for the unknown: Anticipate potential roadblocks and have a core team dedicated to developing alternative solutions and addressing unforeseen issues.
Strong business partnership: A trusted relationship with business teams is essential for smooth user acceptance testing, change management, and avoiding unnecessary disruptions during the migration.
Balanced governance: Independent governance should provide guidance and support calculated risk-taking, acting as a partner in problem-solving rather than a blocker.
Motivated team: Cultivate a team-oriented environment where ownership of the project extends beyond leadership to every team member.
Transparent communication: Maintain open and consistent communication among all stakeholders (in our case, over 250 people) to ensure everyone is aligned and informed.
Fast fail and incremental delivery: Avoid a "big bang" approach. Embrace incremental releases (we aimed for 2-5 daily releases) to learn quickly, adapt, and iterate.
Parallel run: Plan for a parallel run of your systems on both the legacy and target cloud platforms to ensure a smooth transition and validate the new environment.

A data-driven future with limitless potential

Our migration to BigQuery and Vertex AI is just the first step in Definity’s data transformation journey. With a modern, scalable, and AI-ready data foundation now in place, we are empowered to unlock even greater value from our data and continue to lead innovation in the insurance industry. We are excited about the possibilities that lie ahead and are already actively developing our next AI use cases, including several focused on legal summarization and IT functions. We are confident that our partnership with Google Cloud will be instrumental in helping us achieve our goals.

Learn how to start your own migration journey to BigQuery today.

How SIGNAL IDUNA supercharges customer service with gen AI

Thu, 13 Mar 2025 15:59:00 +0000

Today’s insurance customers expect more: simple digital services, instant access to service representatives when they want to discuss personal matters, and quick feedback on submitted invoices. Meeting these demands has become increasingly difficult for insurers due to rising inquiry volumes, a shortage of skilled workers, and the loss of expertise as employees retire.

Recognizing the growing need for immediate and accurate responses, SIGNAL IDUNA, a leading German full-service insurer, particularly prominent in health insurance, introduced a cutting-edge AI knowledge assistant, powered by Google Cloud generative AI.

“We’ve pioneered to unlock the power of human-AI collaboration: To redefine process efficiency by bringing together technology and subject matter experts to deliver exceptional customer experiences,” said Johannes Rath, board member for Customer, Service, and Transformation at SIGNAL IDUNA.

SIGNAL IDUNA, in collaboration with Google Cloud, BCG and Deloitte, has developed an AI knowledge assistant that empowers service agents to quickly and accurately resolve complex customer inquiries. This innovative solution uses Google Cloud AI, including Google’s multimodal Gemini models, to help agents find relevant documents and provide comprehensive answers 30% faster — ultimately, enhancing customer satisfaction.

aside_block: <ListValue: [StructValue([('title', '$300 in free credit to try Google Cloud AI and ML'), ('body', <wagtail.rich_text.RichText object at 0x7f8a79f3e070>), ('btn_text', 'Start building for free'), ('href', 'http://console.cloud.google.com/freetrial?redirectPath=/vertex-ai/'), ('image', None)])]>

The Challenge: Meeting modern expectations

Like many organizations in the insurance sector, SIGNAL IDUNA faced significant operational burdens. The complexity of insurance products, along with the growing demand for immediate and accurate responses, often leads to bottlenecks that can impact service experiences.

For example, prior to introducing its AI knowledge assistant, service agents had to manually search thousands of internal documents for hundreds of different tariffs to find the information needed to answer questions or resolve customer issues — including, insurance conditions, tariff information, guidelines, and standard operating procedures. As a result, 27% of inquiries required further escalation to other departments or specialists, resulting in delayed resolutions, increased costs, and potential damage to reputation.

Though complex, SIGNAL IDUNA prioritized this process as one of its top gen AI use cases, developing an AI assistant to help agents provide quick and accurate answers to customer inquiries, particularly those about health insurance. The AI knowledge assistant is grounded in more than 2,000 internal documents for more than 600 different tariffs, allowing agents to ask questions in natural language and receive accurate answers, significantly reducing the time spent searching for relevant information.

A deep dive into SIGNAL IDUNA's gen AI system

Working with Google Cloud, BCG, and Deloitte, SIGNAL IDUNA built a sophisticated generative AI architecture using Google Cloud’s AI platform, Vertex AI, and utilized Gemini 1.5 Pro’s long-context capabilities to develop an AI knowledge assistant that can provide quick and accurate access to the right information within a vast collection of documents. The system employs multiple steps to aggregate and process extensive information from diverse sources, ensuring agents can access the complete context necessary to effectively address customer inquiries.

Here’s a breakdown of the key steps:

An end-to-end architecture diagram

1. Data pre-processing and extraction
The knowledge base is built from various document types, which are typically in PDF format, including policy documents, operating procedures, and general terms and conditions.

SIGNAL IDUNA utilizes a hybrid approach that combines Layout Parser in Google Cloud Document AI and PDFPlumber to parse these PDFs and extract the text content. While the Layout Parser is responsible for extracting the text segments, SIGNAL IDUNA enhances the extraction of tables with PDFPlumber if the quality of the PDFs allows. The extracted texts are then cleaned, chunked by Google's Gecko multilingual embedding model, and enhanced with additional metadata, enabling the ability to process and analyze the information later effectively.

For storing the vectorized texts, Google Cloud SQL for PostgreSQL is used with the pgvector PostgreSQL extension, which provides a highly effective vector database solution for our needs. By storing vectorized text chunks in Cloud SQL, SIGNAL IDUNA benefits from its scalability, reliability, and seamless integration with other Google Cloud services, while pgvector empowers efficient similarity search functionality.

2. Query augmentation
Query augmentation generates multiple queries to improve the formulation of user questions for both document retrieval from the vector store and answer generation. The original question is reformulated into several variants, creating three versions in total: the original query, a rewritten query, and an imitation query. These are used then to retrieve relevant documents and generate the final answer.

For the rewritten query, the system uses Gemini Pro 1.5 to correct spelling errors in the original question. Additionally, the query is expanded by adding synonyms for predefined terms and tagging specific terms (e.g., "remedies," "assistive devices," "wahlleistung/selective benefits") with categories. The system also uses information about selected tariffs to enrich the query. For example, tariff attributes, such as brand or contract type, are extracted from a database and appended to the query in a structured format. These specific adjustments make it possible to handle special tariff codes and add further context based on tariff prefixes.

The imitation query uses Gemini Pro 1.5 to rephrase the question to mimic the language of technical insurance documents, improving the semantic similarity with the source material. It considers conversation history and handles age formatting.

3. Retrieval
First, the system checks the query cache, which stores previously answered questions and their corresponding correct answers. If the question, or one very similar to it, has already been successfully resolved, the cached answer is retrieved, helping to provide a rapid answer. This efficient approach ensures quick access to information and avoids redundant processing.

The accuracy of the cache is maintained through a user feedback loop, which identifies correctly answered questions to be stored in the cache through upvotes. A downvote on a cached answer triggers an immediate cache invalidation, ensuring only relevant and helpful responses are served. This dynamic approach improves the efficiency and accuracy of the system over time. If no matching questions are found in the query cache, the retrieval process falls back on the vector store, ensuring that the system can answer novel questions.

After retrieving any relevant information chunks from the query cache or vector store, the system uses the Vertex AI ranking API to rerank them. This crucial process analyzes various signals to refine the results, prioritizing relevance and ensuring the most accurate and helpful information is presented.

Ensuring complete and accurate answers is paramount during retrieval, and SIGNAL IDUNA found that some queries required information beyond what was available in the source documents. To address this issue, the system uses keyword-based augmentations to supplement the final prompt, providing a more comprehensive context for generating responses.

4. Generation
The answer generation process involves three key components: the user's question with multiple queries, retrieved chunks of relevant information, and augmentations that add further context. These elements are combined to create the final response using a complex prompt template.

Delivering a near real-time experience is crucial for service agents, so SIGNAL IDUNA also streams the generated response. During development, minimizing latency based on the input posed a significant technical hurdle. To address this issue, SIGNAL IDUNA reduced processing times using asynchronous APIs to help stream data and handle multiple requests. Currently, the system has achieved an average response time of approximately 6 seconds, and SIGNAL IDUNA is experimenting with newer faster models to reduce this time even further.

5. Evaluation
Rigorous evaluation is essential for optimizing Retrieval Augmented Generation (RAG) systems. SIGNAL IDUNA uses the Gen AI evaluation service in Vertex AI to automate the assessment of both response quality and the performance of all process components, such as retrieval. A comprehensive question set, created with input from SIGNAL IDUNA’s service agents, forms the basis of these automated tests.

The evaluation results flow seamlessly into Vertex AI Experiments and Google Cloud BigQuery. This enables SIGNAL IDUNA to visualize performance trends and gain actionable insights using dashboards with Looker on Google Cloud.

Here’s a closer look at how Looker helps evaluate the AI knowledge assistant:

Chunk retrieval: First, SIGNAL IDUNA evaluates retrieval of relevant information chunks. Metrics at this stage help assess how effectively the model identifies and gathers the necessary information from the source data. This includes tracking gen AI metrics, such as recall, precision, and F1-score, to pinpoint areas for improvement in the retrieval process. This is crucial as retrieving the correct information is the foundation of a good generated response.
Document reranking: Once the relevant chunks are retrieved, they’re reranked to prioritize the most pertinent information. The Looker dashboard allows monitoring the effectiveness of this reranking process.
Generated vs. expected response comparison: The final stage involves comparing the generated response with the expected response. SIGNAL IDUNA evaluates the quality, accuracy, and completeness of the generated output, utilizing large language models (LLMs) to score the similarity between the generated response and the expected response.
Explanation generation: To understand the reasoning behind an LLM’s evaluation, SIGNAL IDUNA generates explanations for its judgments. This provides valuable insights into the strengths and weaknesses of the generated responses, helping the developers identify specific areas for improvement.

This multi-stage evaluation approach provides SIGNAL IDUNA a holistic view of the model’s performance, enabling data-driven optimization at each stage. The Looker dashboard plays a vital role in visualizing these metrics, making it easier for the developers to identify areas where the model excels and where it needs improvement.

Real-world impact: AI-powered efficiency and productivity

To determine whether the AI assistant provided measurable added value for its workforce, SIGNAL IDUNA conducted an experiment with a total of 20 employees (internal and with external providers). During the experiment, customer requests were processed with and without the AI knowledge assistant to assess its impact.

One of the key benefits observed was a reduction in processing time. Searching across numerous data sources used to be a time-consuming process. The experiment showed that using the AI knowledge assistant reduced the core processing time (information search and response formulation) by approximately 30% and increased the quality of the response based on expert evaluations. The time saved was particularly notable for employees with less than two years of experience in health insurance.

In addition, the AI knowledge assistant significantly increased the case closure rate. Health insurance is a very complex field, and the use of external service providers means that not every employee can always answer every customer question. With support from the AI knowledge assistant, SIGNAL IDUNA’s case closure rate increased by approximately 24 percentage points, rising from 73% to almost 98%.

Scaling for the Future

“Together with Google, we at SIGNAL IDUNA have successfully applied gen AI to one of our core business processes” Stefan Lemke, CIO at SIGNAL IDUNA, said. “Now, it's time to scale this powerful technology across our entire organization. We're not just scaling a tool, we're scaling innovation, learning, and the possibilities of what we can achieve.”

Gen AI offers enormous potential for optimizing processes and developing innovative solutions. With its innovative approach — business teams experimenting with the technology in a decentralized manner and developing customized applications — SIGNAL IDUNA is primed to pioneer the next generation of insurance solutions and services.

At the same time, SIGNAL IDUNA is establishing central standards to scale insights gained across the company and tap into the combined power of its teams, resources, and lines of business. This strategic decision has helped create valuable resources like code libraries, infrastructure blueprints, and centrally offered services.

By combining agility with established standards and best practices, SIGNAL IDUNA can now react quickly to new requirements, setting a new standard for efficiency and customer satisfaction.

^{This project was delivered by the following core team members, Max Tschochohei, Anant Nawalgaria, and Corinna Ludwig by Google, and Christopher Masch, Michelle Mäding from SIGNAL IDUNA}

Financial Services

How SAP Concur automates expense reporting with agentic AI

Speed, scale, and ingenuity

Key design patterns for successful AI agents

Architecting for ambiguity: Google Cloud’s ecosystem advantage

Ready to build your next-generation agent?

PayPal's historically large data migration is the foundation for its gen AI innovation

Untapped potential of data

Legacy systems, modern ambitions

The journey to unified data

Benefits from BigQuery and beyond

AI-powered innovation unleashed

Lessons for the AI era

Build financial resilience with AI-powered tabletop exercises on Google Cloud

Our solution? Context-aware scenario modeling powered by Google AI

High-level sample scenario timeline

The execution: A cross-functional emergency drill

Impact & key results

Get started

A new era: Supporting customers as a critical ICT third-party provider under EU DORA

A smart investment: FINRA builds a culture of improvement with DORA

Using MCP with Web3: How to secure agents making blockchain transactions

The agent-controlled model

How it works

The inherent risks

The self-hosted variant

The transaction-crafter model

How it works

How to build the agent with Google Cloud tools

How the agent confirms intent

MCP servers should serve both realities

How CME Group builds a faster, smarter exchange on Cloud SQL

From butter and eggs to bandwidth

When Cloud SQL meets the trading floor

Cloud SQL is our new favorite teammate

Performance is our competitive advantage

Learn more:

How Global Payments built a resilient architecture for scale with Cloud SQL

Introducing an agentic commerce solution for merchants from PayPal and Google Cloud

How it works

Smooth, simple shopping journeys: The power of agent collaboration

Consumer trust at the core

Google Cloud and AMD at STAC Summit NYC: H4D VMs for Finance

H4D VMs for financial services

H4D: Superior performance for financial workloads

A full spectrum of financial services solutions

Come talk to experts!

How AI can scale customer experience — online and IRL

Why this architecture matters: Proven ROI

Platform capabilities designed for scale

The technical blueprint: Building intelligent customer experience systems on Google Cloud

The front-end layer

The orchestration and AI engine

The knowledge and integration layer

Putting it all together

Looking ahead: Empowering customer experience teams at scale

The oracles of DeFi: How to build trustworthy data feeds for decentralized applications

2025 is the year of strategic engagement

The need for trustworthy data

Building trustworthy oracle architecture

The Smart Derivative Contract: a DLT-based financial instrument relying on trustworthy data

Technical foundation: security layers

Applications beyond derivatives

Building production-ready blockchain infrastructure

Deutsche Bank delivers AI-powered financial research with DB Lumina

Creating AI-powered research experiences

DB Lumina architecture

RAG and document ingestion

User interface

An evaluation framework for gen AI

Bringing new benefits to the business

Building the future of gen AI and RAG in finance

How Mr. Cooper assembled a "team" of AI agents to handle complex mortgage questions

The challenge: Beyond the reach of traditional automation

The vision: Introducing CIERA, a collaborative AI agent workforce

How it works in practice: A real-life scenario

Ensuring quality and trust: The "agentic pulse" and human oversight

Projected impact: From complex processes to clear wins

The future is agentic and collaborative

Setting new expectations: Benchmarking high-performance trading with C3 machines