AutoScout24 details a Bedrock Bot Factory blueprint for standardized AI agents
Today, AutoScout24 outlined how it standardized internal AI agent development using Amazon Bedrock. The design targets repetitive developer-support work that the company says consumed up to 30% of AI platform engineers’ time.
Today, AutoScout24, a Europe-based automotive marketplace platform, published an architectural write-up on its Bot Factory approach for building and operating AI agents using Amazon Bedrock.
AutoScout24 said the first target was internal developer support, where AI platform engineers were spending up to 30% of their time on repetitive tasks, and it built the blueprint in a three-week bootcamp with AWS Prototype and Cloud Engineering (PACE).
The reference flow uses Slack as the entry point, validates events via Amazon API Gateway and AWS Lambda signature checks, decouples processing with an Amazon SQS FIFO queue, and runs the agent in Amazon Bedrock AgentCore with an orchestrator built using Strands Agents.
For retrieval and actions, the design uses Amazon Bedrock Knowledge Bases for RAG and tool calls for tasks like assigning a GitHub Copilot license, while thread context is keyed by a deterministic sessionId (Slack channel ID + thread timestamp) that can be traced end-to-end in AWS X-Ray.