AI investigation
From evidence to a better hypothesis.
Investigate errors using your own model key, grounded in available telemetry and bounded repository context, with human control over the next step.
Start with the evidence
AI analysis can use the exception, stack trace, breadcrumbs and relevant repository context. Findings distinguish the symptom from the likely underlying cause.
Review the proposal
A proposed patch is reviewed against the code context and validation checks. Analysis is assistance, not proof; your team still reviews the change and runs the appropriate tests.
Control stays with your team
Bring your own supported model key and budget. Repository writing is not enabled in the current beta setup. When enabled, writing is separately gated and requires explicit human approval. Automatic merging is not part of the workflow.
Related
- Find the failure. Keep the context. Group exceptions into issues, inspect stack traces and breadcrumbs, and understand which users and releases are affected.
- Fix what matters to your users. Bring errors, performance and feature usage together to help prioritise engineering work by its effect on the product.
Bring your application into focus.
Tell us about your stack. We will help define the right beta setup for your first project.