Building Self-Healing Agents: Automating the Feedback Loop

Software teams frequently encounter bugs that require manual log analysis and reproduction before a fix is pushed. This session details a system where Droid triages its own bug reports, queries production observability data, reproduces failures in a sandboxed terminal, and ships fixes as pull requests. This talk walks through the tools that make this possible for engineers who want a practical playbook for making systems self-diagnosing and self-healing. 𝐊𝐞𝐲 𝐓𝐚𝐤𝐞𝐚𝐰𝐚𝐲𝐬 • How to enable agents to query production logs and traces • Reproducing failures automatically in sandboxed terminals • A roadmap for making AI systems self-diagnosing and self-healing Subscribe to our calendar: https://luma.com/AgenticAIObservability