Autonomous software remediation with capabilities, gates and proof.

What is autonomous remediation?

Autonomous remediation is the controlled use of software and AI to investigate an incident, take an approved recovery or repair action, test the result and verify service health. It may involve an operational action, a source-code change or both.

Operational recovery and code repair are different

Restarting an approved service may restore availability. It does not prove that the underlying defect is fixed. Remedy distinguishes temporary recovery from permanent repair and can escalate repeated or unresolved incidents into the repository workflow.

Bounded autonomy

The Remedy Agent is designed around explicit capabilities—not unrestricted shell access. Repository scope, operational actions, checks, reviewers and deployment permissions are set by customer policy. Autonomy exists inside those boundaries.

Choose the response before generating code

A failing service does not always require a source change. The immediate response may be a permitted recovery step, while an unresolved or recurring problem needs engineering investigation. Each action needs an authorised scope and an observable success condition.

When a diagnosis requires a code change, the workflow moves into isolated AI code repair. That page describes the candidate diff and test evidence. This page describes the wider incident-response policy. Agent-based operational actions remain in development; check current availability.

Frequently asked questions