Release Autopilot

Releases that run themselves.

Agents shipping to production was unthinkable. Think again.

Classify every change.

Each PR gets a risk score that results in a flag and release plan recommendation. Engineers can review and approve to automatically add flags and kick off releases once code is deployed.

Flag the risky ones.

Feature flag code is automatically added to PRs (or via Agent Skills). The flag resources are auto-created in LaunchDarkly, assigned to the correct maintainers, and assigned an appropriate release plan.

Run the release.

Once code is deployed, signals automatically kick off release automation. Build custom, multi-environment release plans that gradually ramp through audience segments. Ensure that your best release practices happen by default.

Roll back and fix.

Measure every change directly. If anything goes wrong, get an alert and have releases rolled back immediately. The regression analysis provides the context for an agent to auto-fix the issue.

Clean up the code.

Automated checks ensure that flags are ready to be cleaned up safely. PRs are auto-generated and assigned to engineers for review. After successful merge, flags are automatically archived.

We now ship continuously and release only when we’re ready. The result: 3.9 times more work delivered to customers, and escaped defects down from 64 per release to virtually zero. Release Autopilot helps us move this fast without asking customers to take on more risk.

Tiama Hanson-Drury

Chief Product and Technology Officer

Make safer releases the standard.

Safeguard every change: Know the risk, limit who sees each change, and stay in control if something goes wrong.

Reduce release toil.

Give engineers their time back to solve the biggest problems, and give your agents the controls to release safer.

Stop losing sleep over releases.

Ship more AI-generated code with a safety net that catches problems early, rolls back what breaks, and helps you fix it.

We fit right in

Works with the tools you already use.

Install and connect

Get started with Release Autopilot.

Keep your repos, your model provider, and your release controls.

Get started Get started

01Add the GitHub app.
Install the GitHub app, connect repos, and set up each repository for auto-flagging and auto-release.
02Bring your own key, or use ours.
Use your own model-provider key to power Release Autopilot. It still handles multi-environment releases, guarded rollouts, rollback, and cleanup while enforcing your release controls. LaunchDarkly inference is an alternative to your own key, where supported.
03Turn it on, repo by repo.
Choose which repos use auto-flagging and auto-release. Set up once and run consistent automation to help ensure best practices by default.

Infrastructure

Trusted in production,
proven at scale.

As software delivery becomes more automated, releases and rollbacks depend on fast, reliable runtime control. The LaunchDarkly runtime network gives you production-grade control at global scale. Performance, resilience, and security are built in, and the infrastructure is managed for you.