I am exploring how AI can help engineers investigate application failures. Sentinel is the first project in this area, with payment and external API failures as its initial focus.
An AI assistant for incident investigation.
Concept · early developmentSentinel is planned to help engineering teams follow evidence across logs, failed requests, and provider responses. Its goal is to identify a probable cause and produce a report that people can review.
Explore projectAI should help with a specific task and give people enough evidence to check its conclusions.
Begin with the failure. Identify what happened and what the engineer needs to know.
Follow the evidence. Use logs, API responses, and context to guide each next step.
Show the reasoning. Present a likely cause with supporting evidence and suggested next actions.
Tell me what makes it hard to trace. I am interested in real incident workflows as this project develops.