Inspiration

An error alert tells you something broke, but figuring out why often means jumping between stack traces, logs, and requests. We wanted to shorten that investigation and help developers move from a crash to a fix they can actually verify.

What it does

Aftermath is an AI debugging assistant designed to connect Sentry errors, logs, and traces into a clear incident timeline. It helps developers investigate the likely cause, inspect the supporting evidence, and draft a code fix with a regression test.

The goal is simple: show what happened, explain why it might have happened, and give developers a concrete next step.

How we're building it

Our planned stack combines a React and TypeScript interface with a Python backend. Sentry provides error monitoring, Logs, and Tracing, while the OpenAI API helps interpret incident evidence and propose fixes.

The investigation workflow starts with an error, gathers related events, and presents a diagnosis linked to its sources. Suggested changes stay available for developer review.

Challenges we're tackling

The hardest part is distinguishing a plausible explanation from one supported by evidence. Logs can be incomplete, and events that happen close together aren't necessarily related.

We're focusing on connecting events through request and trace identifiers, making uncertainty visible, and designing regression tests that reproduce the actual failure.

What we're learning

Our central question is how to make AI-assisted debugging trustworthy. We're exploring what context a model needs, how to present evidence clearly, and how tests can help validate a proposed fix.

What's next

Our first milestone is a complete live demo: trigger a bug in a sample application, investigate it using Sentry Logs and Tracing, propose a fix, and verify the behavior with a regression test.