Building AI agents for business support using Amazon Bedrock AgentCore

2026-05-27 · Source: Artificial Intelligence · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cloud Computing & IT Infrastructure, Robotics & Autonomous Systems · Depth: Intermediate, medium

Summary

AWS Generative AI Innovation Center (GenAIIC) collaborated with Works Human Intelligence (WHI) to develop two AI agents for business support using Amazon Bedrock AgentCore, achieving up to a 97% cost reduction and improved operational efficiency. The Commuting Allowance Agent, which automates application approvals, migrated from a monolithic LangGraph/Amazon ECS setup to AgentCore Runtime, enabling individual sub-agent launches and multi-tenancy with Amazon DynamoDB and Amazon Cognito. The Browser Operation Agent, designed to interact with WHI's "COMPANY" HR system, was built using Strands Agents. This agent initially achieved an 88% token reduction through prompt caching and optimized conversation history. Further cost optimization for the Browser Operation Agent involved using Amazon Bedrock's prompt caching, refining sub-agent prompts, and switching from Claude Sonnet 4.5 to Haiku 4.5, reducing cost per process from \$14.5 to \$0.4. This collaboration streamlined development and enabled WHI to focus on business logic.

Key takeaway

For AI Engineers or MLOps Engineers building or optimizing AI agents for business process automation, you should consider Amazon Bedrock AgentCore and Strands Agents. Migrating to this managed service can significantly reduce operational costs by up to 97% and streamline development efforts. Leverage features like prompt caching and evaluate cost-effective models such as Haiku to optimize your agent's behavior and resource consumption. This approach allows you to focus on core business logic rather than infrastructure management.

Key insights

Amazon Bedrock AgentCore and Strands Agents enable significant cost reduction and operational efficiency for AI agent development in business support.

Principles

• Modularize agents for scalability.
• Optimize prompts and models for cost.
• Use managed services for observability.

Method

The Browser Operation Agent workflow involves searching for optimal operation templates, creating a manual, operating the browser to check information, proposing changes, and executing them upon user approval.

In practice

• Migrate monolithic agents to AgentCore Runtime.
• Implement prompt caching for LLM calls.
• Evaluate models like Haiku for cost savings.

Topics

• Amazon Bedrock AgentCore
• AI Agents
• Strands Agents
• Cost Optimization
• Prompt Caching
• HR Automation
• Large Language Models

Code references

• aws-samples/sample-getting-started-with-amazon-agentcore

Best for: AI Engineer, MLOps Engineer, AI Architect

Related on AIssential

• Build AI agents for business intelligence with Amazon Bedrock AgentCore — AI agents on Amazon Bedrock AgentCore can autonomously process fragmented business data for real-time BI.
• Powering agentic AI sales strategy with Amazon Bedrock AgentCore — Orchestrating specialized AI agents with a unified interface significantly reduces cognitive load and improves operational efficiency.
• AWS Bedrock Agents: The Beginner’s Guide — AWS Bedrock Agents provide a managed platform for building and deploying AI agents with integrated orchestration and security.
• Semantic search across MCP tools with Amazon Bedrock AgentCore Gateway — Amazon Bedrock AgentCore Gateway enables scalable LLM agents through built-in semantic search for tool selection.
• Sage and AWS: Helping SMBs Embed Agentic AI — Sage and AWS partner to simplify agentic AI adoption for SMBs in financial workflows via cloud migration and a developer marketplace.
• AWS, Coinbase & Stripe Launch Autonomous Paying AI Agents — AWS Bedrock AgentCore Payments enables AI agents to autonomously transact for digital resources using Coinbase and Stripe.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Artificial Intelligence.