Stop shipping AI regressions

LaunchGate runs evals on every change and blocks deployments when quality drops below your threshold.

Get Started FreeRead the Docs
import { LaunchGate } from "@launchgate/sdk";

const lg = new LaunchGate({ apiKey: process.env.LAUNCHGATE_API_KEY });

const result = await lg.run("rag-faithfulness", {
  input: { context, query },
  output: aiResponse,
});

if (result.status === "cleared") {
  // All systems go — deploy with confidence
}

How it works

1

Define what good looks like

Create an eval suite with test cases and scoring criteria.

2

Run evals on every change

Trigger from your CI/CD pipeline, SDK, or CLI.

3

Gate deployment on results

Pass rate below threshold? Deployment blocked automatically.

4

Ship with confidence

Every release is validated. No more guessing if quality held.

Everything you need to gate AI quality

Quality gates that actually block

Define pass thresholds per suite. If evals fail, the deployment stops. No more shipping regressions.

5 scorer types built in

Exact match, regex, JSON schema, contains, and LLM-as-judge. Mix deterministic checks with AI evaluation.

GitHub Action in 2 minutes

Add one YAML file to your repo. LaunchGate posts results directly to your PR with pass/fail status.

Track quality over time

See pass rates trending up or down. Catch regressions before they reach users.

SDK and CLI

npm install @launchgate/sdk and run evals from code. Or use the CLI in any CI/CD pipeline.

Bring your own keys

Use your OpenAI, Anthropic, Google, or Azure keys for LLM-as-judge scoring. Encrypted at rest.

Ready for launch?

Start with 500 free eval runs per month. No credit card required.

Get Started Free