It support · Production

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

The problem

AutoScout24's AI Platform engineers were spending up to 30% of their time on repetitive support tasks—answering questions, granting tool access, and locating documentation—while separate engineering teams were building AI agents in fragmented, non-standardized ways that prevented enterprise-wide scaling.

Workflow diagram · grounded in source
1
Developer posts Slack message
trigger
“User interaction via Slack: A developer posts a message in a support channel, for example, "@SupportBot, can I get a GitHub Copilot license?"”
2
Cryptographic signature verification
validation
“This function performs an essential security check, verifying the request's cryptographic signature to confirm it's authentically from Slack.”
3
SQS FIFO queue decoupling
integration
“The verified request is placed onto an Amazon SQS First-In, First-Out (FIFO) queue. This decouples the front-end from the agent, improving resilience. Using a FIFO queue with the message's thread timestamp as the MessageGroupId makes sur…”
4
Orchestrator routes to specialized agent
routing
“The Orchestrator Agent's logic, built with Strands Agents, analyzes the user's prompt and determines the correct specialized agent to invoke—either the Knowledge Base Agent for a question or the GitHub Agent for an action request.”
5
RAG knowledge retrieval
ai_action
“Knowledge retrieval: Answering "how-to" questions by searching internal documentation, a capability known as Retrieval Augmented Generation (RAG).”
6
GitHub action execution
ai_action
“Action execution: Performing tasks in other systems, such as assigning a GitHub Copilot license, which requires secure API integration, or "tool use."”
7
Response delivered to Slack
output
“the response back to Slack”
Reported outcome

The team deployed a production-ready Slack support bot that is actively reducing the manual support load on the AI Platform Engineering team, addressing the 30% of time previously spent on repetitive tasks, and produced a reusable Bot Factory blueprint that allows other teams across AutoScout24 to build new agents faster without reinventing infrastructure.

Reported metrics
Engineer time on repetitive support tasksup to 30%
Manual support loadactively reducing the manual support load
Innovation speed across teamssignificantly accelerates innovation
Reported stack
Amazon BedrockAmazon Bedrock AgentCoreAmazon API GatewayAWS X-RayAWS Secrets ManagerIAMSlackGitHub
Source
https://aws.amazon.com/blogs/machine-learning/how-autoscout24-built-a-bot-factory-to-standardize-ai-agent-development-with-amazon-bedrock?tag=soumet-20
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

The team deployed a production-ready Slack support bot that is actively reducing the manual support load on the AI Platform Engineering team, addressing the 30% of time previously spent on repetitive tasks, and produc…

What tools did this team use?

Amazon Bedrock, Amazon Bedrock AgentCore, Amazon API Gateway, AWS X-Ray, AWS Secrets Manager, IAM, Slack, GitHub.

What results were reported?

Engineer time on repetitive support tasks: up to 30%; Manual support load: actively reducing the manual support load; Innovation speed across teams: significantly accelerates innovation (source-reported, not independently verified).

How is this it support AI workflow structured?

Developer posts Slack message → Cryptographic signature verification → SQS FIFO queue decoupling → Orchestrator routes to specialized agent → RAG knowledge retrieval → GitHub action execution → Response delivered to Slack.

Related it support cases