Financial institutions advance mission-critical workloads and Agentic AI at re:Invent 2025

re:Invent 2025 addressed a wide range of financial services challenges, from mission-critical infrastructure to customer experience transformation. Agentic AI took center stage with over 30 dedicated sessions, reflecting the industry’s adoption of AI agents in production environments. From Matt Garman’s (CEO, AWS) keynote featuring Nasdaq, Visa, and National Australia Bank (NAB’s) work with Amazon Bedrock AgentCore, to BlackRock bringing Aladdin investment management technology platform to AWS, the event showcased how the industry’s key participants are deploying AI agents at scale.

A shift emerged in how financial services executives approach agentic AI. The conversation has evolved from whether to adopt agentic AI to how quickly they can deploy it to maintain competitive advantage. Meanwhile, developers have experienced their own transformation, moving from skepticism about AI-generated code quality to embracing agentic frameworks that build production-ready applications in days rather than months. Financial institutions leveraging agentic AI are already reducing operational costs, creating additional revenue streams, and improving customer experiences.

Scott Mullins, Managing Director of Financial Services at AWS, noted during his Innovation Talk that financial institutions are well-positioned for this shift to an agentic future: “One reason that we now stand on the threshold of a new frontier today where agents won’t only chat with us, but act on our behalf is that financial institutions have been preparing for this moment for years. They’ve moved their data to AWS and set up processes and guardrails that would enable them to securely and confidently take advantage of machine learning and generative AI.”

This momentum is reflected in industry forecasts: Gartner predicts that 40% of enterprise applications will be integrated with task-specific AI agents by the end of 2026, up from less than 5% today¹. Financial institutions that have already migrated their data and workloads to AWS are ready to build these agentic applications on secure, scalable infrastructure with the guardrails and governance frameworks required for regulated industries.

Deploying agentic AI at scale requires mission-critical infrastructure, enterprise-grade resiliency, and data foundations, and re:Invent 2025 demonstrated how financial institutions are building these foundations.

Three Key Themes from re:Invent 2025

Mission-critical applications on the cloud and resiliency at scale
Financial institutions accelerated their migration of mission-critical workloads to AWS in 2025, demonstrating trust in cloud infrastructure for their most demanding applications. Itaú Unibanco migrated its 50-year-old mainframe-based checking account authorization platform serving 70 million customers to AWS while maintaining 99.99% uptime and sub-100ms latency. Visa deployed its Tier 0 Visa Protect for Account-to-Account payments workload on AWS, providing real-time fraud scoring with sub-250ms latency and 99.99% availability. Beyond migration, institutions showcased advanced resiliency capabilities. Capital One described a multi-region resiliency program built around automated dependency discovery, centralized recovery tooling, and integration with AWS primitives such as ARC Region Switch. Nasdaq explained how it builds ultra-low-latency, mission-critical market systems on AWS using Outposts, multi-site deployments, and engineered failure handling, so that trading continues under severe failures or network isolation.

Data foundations for AI
Financial institutions recognize that a robust data infrastructure is a critical enabler for successful AI implementations. Fidelity Investments showcased its enterprise-scale text-to-SQL solution achieving 93-95% execution accuracy with 28-second response times, enabling non-technical users to query complex databases using natural language through AI-ready data structures, semantic layers, and dynamic workflows. London Stock Exchange Group (LSEG) demonstrated massive-scale AI capabilities, processing 274 billion daily market updates across 575+ exchanges. All this while migrating 75 petabytes of historical data to Amazon S3, achieving 5X cost reduction and improving anomaly detection from days to minutes using a serverless architecture with Amazon Bedrock, AWS Glue, and Amazon Athena.

Agentic AI in production

• Insurance: Prudential built a life insurance advisor assistant using a multi-agent architecture where an orchestration agent routes requests to specialized sub-agents for quotes, forms, products, illustrations, and book-of-business management. Separately, Allianz Technology SE implemented a multi-agent claims processing system that reduced processing time by 80% from approximately 100 days using a workflow managed by seven specialized agents. The multi-agent system handles everything from coverage verification to fraud detection to automated payouts, with audit trails for regulatory compliance.
• Trading and Capital Markets: Sessions explored how financial institutions are deploying trading assistants and high-frequency trading analytics powered by agentic AI for real-time news sentiment analysis and market impact assessment. Robinhood described its tuning and evaluation strategy for production AI agents including Customer Experience support, Cortex Digest, and natural-language trading tools, cutting latency and cost while improving quality.
• Banking: Danske Bank is applying AWS Transform and Kiro to analyze and re-imagine core mainframe workloads, with AI-powered code analysis, reverse-engineering of undocumented business rules, and decomposition of monoliths into domain-aligned services.
• Payments: Coinbase demonstrated agentic payments capabilities, while multiple sessions featured agentic commerce solutions integrating payment processing with conversational AI. Visa and AWS announced next-generation agentic commerce capabilities available in AWS Marketplace and Amazon Bedrock AgentCore to help deliver secure, reliable payment experiences.

To support the continued evolution of mission-critical workloads, data foundations, and agentic AI deployments, AWS introduced several service enhancements at re:Invent 2025. Below, we highlight a few that are most relevant to financial institutions.

Service Announcements for Financial Services

Mission-critical applications on the cloud and resiliency at scale

• AWS Interconnect – Multicloud simplifies multi-cloud connectivity between AWS and other major cloud service providers, starting with Google Cloud. This managed connection enables customers to establish private, resilient, high-speed network connections with dedicated bandwidth between their Amazon Virtual Private Cloud (VPCs) and other cloud environments. Use cases: Financial institutions can connect trading platforms across AWS and Google Cloud with low-latency connectivity to ensure regulatory compliance and minimize execution delays for secure trading. They can facilitate disaster recovery and business continuity by maintaining synchronized databases and applications across multiple cloud providers. They can also support hybrid cloud payment processing systems that need to interact with banking systems in Google Cloud while leveraging AWS’s microservices architecture, maintaining consistent performance and security standards.
• Amazon Route 53 Accelerated Recovery for Managing Public Domain Name Service (DNS) Records: This is a new DNS business continuity feature designed to provide a 60-minute recovery time objective (RTO) during service disruptions in the US East (N. Virginia) AWS Region, ensuring customers continue making DNS changes and provisioning infrastructure even during regional outages. Use cases: While AWS maintains availability across our global infrastructure, Financial Services companies gain the confidence that they will be able to make DNS changes during unexpected regional disruptions, allowing them to provision standby cloud resources or redirect traffic quickly when needed. Learn more from our launch blog.

Data foundations for AI

• Amazon Aurora PostgreSQL Dynamic Data Masking: Amazon Aurora PostgreSQL now supports dynamic data masking through the new pg_columnmask extension. This enables column-level protection that complements PostgreSQL’s native row-level security, allowing companies to control access to sensitive data through SQL-based masking policies. Use cases: Financial institutions use pg_columnmask to comply with General Data Protection Regulation (GDPR), Health Insurance Portability and Accountability Act (HIPAA), and Payment Card Industry Data Security Standard (PCI DSS) requirements by masking personally identifiable information (PII) at the database level for regulatory compliance and data privacy. Banks and insurance companies can protect sensitive customer data during support interactions by implementing column-level masking for customer service and support operations. They can also use production-like data in non-production environments without compliance risks by applying SQL-based masking policies for development and testing environments. Learn more from our launch blog.
• AWS Glue Catalog Federation for Remote Apache Iceberg Catalogs: Catalog federation in AWS Glue provides direct and secure access to Iceberg tables stored in Amazon S3 and cataloged in remote catalogs. It synchronizes metadata across AWS Glue Data Catalog and remote catalogs, supported by Amazon Redshift, Amazon EMR, Amazon Athena, AWS Glue, and third-party engines such as Apache Spark. Use cases: Financial institutions can connect AWS analytics directly to Databricks and Snowflake catalogs, enabling them to query enterprise data from multiple platforms through a single interface while maintaining automated metadata synchronization for unified data discovery and access. Learn more by visiting this webpage.
• Amazon S3 Tables Intelligent Tiering and Multi-Region Replication: Amazon S3 Tables replicate complete table structure, including all snapshots and metadata to reduce query latency and improve data accessibility for global analytics workloads. The Intelligent-Tiering storage class automatically tiers data to the most cost-effective access tier based on access patterns. Use cases: With converging feature parity between Amazon S3 and Amazon S3 Tables, financial institutions can quickly migrate to and adopt AWS-managed Iceberg for structured data storage in their data lakes. Learn more from our launch blog.

Agentic AI in production

• Amazon Nova Forge: Amazon Nova Forge enables organizations to build custom frontier models using checkpoints across pre-training, mid-training, and post-training phases. It allows blending proprietary data with Amazon Nova curated data while preserving general capabilities. Use cases: Financial services organizations can build specialized models trained on proprietary research, market analysis, and trading data to create domain-expert AI that understands financial instruments and market dynamics while maintaining reasoning capabilities. They can also develop custom regulatory compliance AI incorporating institution specific compliance procedures and legal frameworks to automate compliance reviews and risk assessments.
• Frontier Agents: Frontier agents represent a new class of AI agents that are autonomous, scalable, and work for hours or days without constant intervention. Kiro autonomous agent is the virtual developer; AWS Security Agent is the virtual security engineer; and AWS DevOps Agent is the virtual operations team member. Use cases: Financial institutions can leverage Kiro autonomous agent to identify security vulnerabilities and compliance violations, prioritize bugs by regulatory impact (GDPR, Health Insurance Portability and Accountability Act (HIPAA), and Sarbanes Oxley (SOX)), and create tested pull requests in isolated sandboxes for bug triaging. AWS Security Agent enables automated penetration testing by returning validated findings with remediation code, scaling to test multiple applications simultaneously. AWS DevOps Agent provides automated incident root cause analysis by responding instantly to incidents and correlating telemetry, code, and deployment data across observability tools like Amazon CloudWatch to identify root causes.
• Amazon Bedrock AgentCore Policy + Evaluations: Policy in Amazon Bedrock AgentCore integrates with AgentCore Gateway to intercept every tool call in real time, ensuring agents stay within defined boundaries. Teams can create policies using natural language that automatically convert to Cedar, the AWS open-source policy language. AgentCore Evaluations helps developers test and monitor agent performance based on real-world behavior. Use cases: Financial institutions can deploy AI agents with policy controls that enforce trading limits, regulatory boundaries, and approval workflows while maintaining audit trails to help with regulatory compliance. They can also create customer service agents with episodic memory that learn individual client preferences, financial goals, and risk tolerances to provide personalized wealth management guidance. Learn more from our launch blog.
• Amazon Connect AI Agents – New Features: Amazon Connect introduced multiple enhancements: Amazon Nova 2 Sonic for real-time conversational AI with best-in-class streaming speech understanding, new first-party agents pre-built for common contact center scenarios, third-party agent integration through Model Context Protocol (MCP), and a recommendations engine that applies AI to unified customer profiles for real-time, personalized suggestions. Use cases: Financial institutions can deliver personalized experiences and surface insights throughout the customer journey, from consideration to loyalty, enhancing customer satisfaction. AI-powered recommendations identify cross-sell and upsell opportunities across financial products, empowering human agents with insights during customer interactions. Organizations can also achieve operational efficiency through automated case summarization that eliminates manual note-taking, agentic manager assistants that provide natural language data exploration, and predictive analytics that enable real-time sentiment analysis, churn prediction, and lead qualification with reduction in review time. Learn more by visiting these webpages: Nova 2 Sonic, MCP Support, and Predictive Insights.

Conclusion

re:Invent 2025 showcased the breadth of innovation happening across financial services. The announcements spanning infrastructure resiliency, data analytics, generative AI, and customer experience reflect AWS’s commitment to addressing the unique challenges of regulated industries. Organizations should evaluate these capabilities, consider which announcements align with their strategic priorities and how they can leverage AWS’s proven track record in financial services to accelerate their modernization journey.

Learn More

• Watch AWS re:Invent 2025 session recordings on the AWS Events YouTube Channel.
• Review top announcements from re:Invent 2025.
• Visit aws.amazon.com/financial-services or Get in touch with an AWS Financial Services expert.
• Follow us on LinkedIn – https://www.linkedin.com/showcase/aws-for-financial-services/.

[1] Gartner, “Gartner Predicts 40% of Enterprise Apps Will Feature Task-Specific AI Agents by 2026, Up from Less Than 5% in 2025,” August 27, 2025. https://www.gartner.com/en/newsroom/press-releases/2025-08-26-gartner-predicts-40-percent-of-enterprise-apps-will-feature-task-specific-ai-agents-by-2026-up-from-less-than-5-percent-in-2025