# Prilog bug fixing: AI bug fixing for production errors

Prilog bug fixing: AI bug fixing from production signals to code, root causes, tested patches, and PRs for human review. Try your first 3 fixes free.

Prilog bug fixing is AI bug fixing grounded in your production logs, traces, GitHub issues, and connected codebase. Prilog investigates the cause, prepares a patch, tests the change, and opens a pull request for human review.

## From a signal to a pull request

1. **Signal:** connect an observability source, send OpenTelemetry logs and traces, upload a selected log file, or import a supported public GitHub issue.
2. **Root cause:** investigate the failure against the connected repository and the available runtime evidence.
3. **Patch:** prepare a code change for the affected behavior.
4. **Tests:** validate the proposed change with repository checks. Inspect the checks actually run and any gaps; passing checks alone do not prove a bug is fixed.
5. **Pull request:** publish the proposed change under your project settings so a person can review the diff, CI results, and remaining risks.

## You control review and merging

Prilog never merges on its own. A human reviews every PR. Project settings control analysis, review before publication, or automatic PR creation after a policy pass. Automatic PR creation does not mean automatic merging. Read the [automation controls](/docs/automation/).

## Pricing and getting started

Pay **$8 per publishable fix**: a generated code change prepared for review and publication as a pull request. Your **first 3 fixes are free**, with **no credit card**. Codebase analysis and OpenTelemetry monitoring are free. See [current pricing](https://prilog.ai/#pricing) for the live terms.

Start with the [Quickstart](/docs/quickstart/), explore the [developer docs](/docs/), and inspect [public results](/results/), including upstream PRs and the separately disclosed benchmark. Public contributions do not establish elapsed fix time or imply customer endorsements.

## Evaluate the workflow

Compare [Sentry Seer](/compare/prilog-vs-sentry-seer/), [Datadog Bits Code](/compare/prilog-vs-datadog-bits-code/), [Resolve AI](/compare/prilog-vs-resolve-ai/), [Superlog](/compare/prilog-vs-superlog/), or [manual coding-agent workflows](/compare/prilog-vs-coding-agents/). Use equivalent evidence and record tests, reviewer effort, extra edits, and unsupported cases.