Miro BugManager uses Amazon Bedrock and Claude Sonnet 4 for automated bug routing

AI Impact Summary

Miro's BugManager is an LLM-powered bug triage system that leverages Amazon Bedrock and Retrieval Augmented Generation to auto-route bug reports to the correct teams. It first parses media (screenshots/videos) with Amazon Nova Pro, enriches the report with context from knowledge bases (GitHub issues, Jira tickets, Confluence docs, READMEs) via Bedrock Knowledge Bases, and uses Anthropic Claude Sonnet 4 to produce an optimized routing prompt. The solution runs as a Python microservice in an Amazon EKS cluster and exposes a Slack workflow for engineers, with optional root-cause analysis drawn from the codebase. This architecture delivers zero-training adaptability in a dynamic org structure and achieves 6x fewer team reassignments and 5x shorter time-to-resolution, reducing SLA risk and boosting developer productivity.