AWS Partners Demonstrate Enterprise AI Agent Solutions with Amazon Bedrock AgentCore

By: Veena Vasudevan, Principal Partner Solutions Architect, Data & AI – AWS
by: Salman Taherian, Senior Partner Development Specialist, Data & AI – AWS

Following the general availability of Amazon Bedrock AgentCore on October 13, 2025, AWS Partners are seizing first-mover advantage in the enterprise AI agent market, rapidly turning proof-of-concept innovations into production solutions. AgentCore provides the foundation to build, deploy, and operate agents using any framework or model without infrastructure management by offering services for Runtime, Memory, Gateway, Identity, and Observability.

Amazon Bedrock AgentCore is driving unprecedented enterprise transformation. These select success stories showcase how early-adopter partners are delivering immense value for their customers today.

Caylent Optimizes CloudZero Advisor

CloudZero provides a FinOps solution that enables enterprises to understand and optimize cloud costs in real time by connecting engineering decisions directly to business outcomes. As more companies onboarded to their platform, CloudZero faced growing challenges scaling their AI-driven cloud cost optimization and infrastructure analysis capabilities. Operations were increasingly slowed by complexity and high developer cognitive load required to manage and tune custom agentic AI workflows processing petabytes of cost and usage data.

Caylent implemented CloudZero Advisor, an agentic platform built on AgentCore Runtime that orchestrates five specialized agents—Cost/Billing, Cloud Pricing, Benchmarking, CostFormation, and Knowledge Base—to deliver comprehensive cloud cost analysis through natural language interactions. AgentCore Memory manages context across cost analysis sessions, while AgentCore Gateway services enable seamless integration with cloud resources. By migrating from Amazon Bedrock Agents to AgentCore, CloudZero removed the complexity of maintaining custom memory systems and achieved 5x faster response times, delivering 2–4s Time to First Token (TTFT) per step versus 30+s previously.

“AgentCore enabled us to build CloudZero Advisor into a production-ready, agentic platform that delivers real speed and efficiency. The performance gains we’re seeing turn agentic AI from a proof of concept into a core part of CloudZero’s data platform, unlocking use cases that were previously too slow or costly and scaling seamlessly to meet 50x growth,” said Randall Hunt, CTO, Caylent.

The implementation reduced developer cognitive load by 75% through automated memory management and standardized CI/CD pipelines that simplified code maintenance. DeepEval validation and built-in failover mechanisms strengthened system resilience, enabling seamless multi-agent collaboration with secure multi-tenant isolation for complex cloud optimization workloads.

Cisco Brings Zero Trust Security to AI Agents for Enterprise Customers

Traditional identity and security systems weren’t designed for AI agents, creating governance challenges as enterprises adopt agentic AI. Organizations need visibility over agent-tool communications and data exfiltration prevention.

Cisco partners with AWS, integrating Cisco Duo and Cisco Secure Access with AgentCore to build comprehensive zero trust architectures for agentic AI. The solution combines Cisco Duo with AgentCore Identity to create an identity fabric with granular access policies and just in time access for human and non-human identities, including machine and agentic identities. Cisco Secure Access’s MCP semantic inspection proxy, integrated with AgentCore Gateway, provides a unique capability—the ability to intercept and inspect all MCP messages between AI agents and tools, detecting threats such as tool poisoning and prompt injections while preventing data exfiltration through DLP controls.

The integration extends zero trust principles to AI agents with the same rigorous policies as humans. AgentCore Gateway facilitates secure discovery and access to MCP servers and tools with zero trust policies and user consent. Cisco Duo integration provides strong authentication and adaptive access controls for both human users and AI agents. Organizations monitor MCP communications, view threats detected and blocked and enforce policies through Cisco Secure Access’s dashboard. AgentCore Observability tracks agent behavior and detects anomalies.

“Cisco is teaming up with AWS AgentCore and creating secure identities for AI agents“, said Pradeep Musugu, Sr. Director, IT Security, Cisco. “We need clear rules to establish and manage access from human and non-human identities, and constant inspection of the meaning/intent of each MCP message. The combined approach of extending Duo’s identity provider capabilities and Secure Access’ MCP message security into AWS AgentCore will allow us to build and use AI confidently, keeping both our AI agents, team members, and data secure as we grow with AI.“

Customers can leverage Cisco Duo IAM to integrate with AWS AgentCore for inbound and outbound authentication today. Cisco’s MCP semantic inspection proxy is in production with selected customers, with broader integration planned for early 2026. This enables organizations to build and scale AI agent deployments with strict security and compliance.

Genpact Enhances Financial Crime Detection at Scale for Apex Fintech Solutions

Apex Fintech Solutions needed to transform financial crime investigation workflows across their digital clearing and custody platform. Their existing process struggled with siloed reporting structures, making it difficult to analyze patterns across accounts, entities, and transactions—limiting their ability to detect and prevent financial crimes while maintaining regulatory compliance.

Genpact implemented their riskCanvas® Data Explorer solution with a sophisticated agent-to-agent communication system built on AgentCore Runtime and leveraging AgentCore Observability. The implementation features a Supervisor Agent pattern coordinating specialized agents for accounts, entities, items, and transactions analysis. This agent intelligently translates domain-specific questions into data query and analysis tasks routed to the appropriate specialist agent. AgentCore Memory maintains context across complex investigations, transforming previously disconnected reporting structures into an integrated intelligence platform that delivers insights through natural language queries.

“Moving from efficiency gains to enterprise value requires a holistic approach to agent orchestration and lifecycle management – and that’s exactly what we did for Apex leveraging Genpact’s Agentic Development Lifecycle with Amazon Bedrock AgentCore to deliver a riskCanvas® Data Explorer roll-out that helped them improve financial crime detection and compliance“, said Matt Arellano, AWS Practice Lead at Genpact.

The solution enables financial crime analysts to process daily trades across millions of brokerage accounts through natural language interactions. Following successful deployment, Apex plans to expand AgentCore across all financial crime detection workflows including AML monitoring, fraud detection, and compliance reporting.

PwC Australia Accelerates Digital Modernization for Major Australian Bank

PwC Australia is leading an Agentic transformation for a major Australian bank’s modernization, streamlining architecture functions to accelerate projects, improve budgets, and prevent rework.

By leveraging AgentCore Runtime and Gateway services, PwC is integrating strategic architecture AI agents within the bank’s existing business processes to automate complex architectural reviews and ensure alignment with the bank’s governance framework and pre-defined implementation patterns. The agents replace manual review processes with AI-driven pattern compliance checks and human-in-the-loop oversight for critical decision points.

“PwC is excited to be the transformation partner for pivotal Agentic transformation initiatives in the banking industry. These strategic initiatives will streamline technology development and modernisation programs – unlocking consistent improvements, returns on investment, along with business outcomes such as speed to market, competitive positioning, and the ability to launch innovative products at scale. PwC is leveraging AgentCore runtime and gateway services, to build an initial architecture agent, within a bank’s existing processes, to target an estimated productivity improvement of over 30%, streamlining the bank’s architecture functions, to accelerate projects, reduce budgets, and prevent the late emergence of critical issues,” said Noel Williams, Banking and Capital Markets Lead Partner, PwC Australia.

AgentCore’s serverless runtime enables PwC to deploy agents without managing infrastructure complexity, while Gateway services provide seamless integration with the bank’s enterprise systems.

Reply Transforms Totemia’s Vacation Discovery with Conversational AI

Totemia, France’s leading childhood vacation camp platform, faced a critical challenge. Parents were overwhelmed with over 60 generic search results, couldn’t handle multi-child searches, and saw no real-time availability, leading to customer abandonment and conversion losses.

Reply partnered with Totemia to develop an Offer Agent built on Strands SDK and hosted on AgentCore Runtime. The solution leverages AgentCore Gateway to invoke AWS Lambda functions generating dynamic SQL queries via Amazon Bedrock foundation models (Anthropic Claude) based on client criteria. These functions connect to Amazon RDS to fetch real-time offers and return the top 5 matches. AgentCore Memory enables multi-turn conversations, refining recommendations based on previous interactions. When no exact match exists, the agent proactively suggests adjusting criteria through iterative refinement. AgentCore Identity via Amazon Cognito provides authentication, while AgentCore Observability tracks performance and reliability.

“With AgentCore, we can easily develop and test our solutions using different Python frameworks, giving us flexibility and faster experimentation,” said Xihui Wang, Reply Data Scientist. “The built-in gateway enables us to securely integrate Lambda functions as tools while also making them reusable across multiple agents. In addition, AgentCore provides built-in memory management, observability, and identity services, which reduce development overhead and simplify scaling to production.“

AgentCore’s ready-to-use building blocks eliminated custom infrastructure development, enabling Reply to move from prototyping to production rapidly. The agent narrows results from 60 to 5-10 tailored recommendations, achieving 65% reduced search time, 40% increased bookings, 24/7 availability, 25% conversion growth, and supporting 30,000 users monthly—projecting 200-300% ROI within year one.

Conclusion

These early-adopter success stories demonstrate how AWS Partners are capitalizing on AgentCore to deliver transformative value across industries – from financial services to travel, security to banking. As enterprise demand for AI agents accelerates, AgentCore enables partners to rapidly scale production-ready solutions while maintaining enterprise-grade security and performance.

The opportunity is immediate. Visit Amazon Bedrock AgentCore to start building today. Partners can access exclusive resources and latest capabilities through AWS Partner Central’s Agentic AI Hub, and showcase their expertise through the AWS Generative AI Competency program.