How AutoScout24 built a Bot Factory to standardize AI agent development with Amazon Bedrock

This article describes how AutoScout24 built a standardized "Bot Factory" framework for developing AI agents using Amazon Bedrock, moving from fragmented experiments to enterprise-grade infrastructure.

• AutoScout24 engineers spent 30% of time on repetitive support tasks like access grants and documentation searches
• Bot Factory uses event-driven serverless architecture: Slack → API Gateway → SQS → Amazon Bedrock AgentCore
• Orchestrator agent delegates to specialized agents (Knowledge Base agent, GitHub agent) using Strands Agents SDK
• Session isolation via unique sessionId per Slack thread ensures conversational context and prevents data leakage
• Amazon Bedrock AgentCore handles session management, scaling, and security without custom infrastructure
• Knowledge Base agent performs RAG for documentation queries; GitHub agent executes actions like license assignments
• Deployed production support bot reducing manual support load significantly
• Reusable blueprint enables other teams to build agents focusing on business logic, not infrastructure

AutoScout24 successfully transformed AI experimentation into a scalable, standardized framework that reduces operational overhead and accelerates enterprise agent development.