Legal intelligence, as an API.
Erin is a legal AI API by GLR Labs. Send it legal work - drafting, research, review, citations - and judge it by one thing: the output. One endpoint, OpenAI-compatible, drop-in with any standard SDK.
It invents citations with perfect confidence. It answers Indian questions with American doctrine. It smooths over genuine legal disagreement into false certainty. And it treats a court filing and a client email as the same kind of task. For most industries these are annoyances. In law, each one is malpractice-adjacent.
Erin behaves the way a careful legal professional behaves.
Each response carries a machine-readable citations array - every case, statute, and section marked verified, unverified, or suspect. Nothing invented reaches you unflagged.
Erin detects the jurisdiction of your question and answers within it - India and US aware today. It will not silently answer an Indian limitation question with American doctrine.
Where the law is genuinely unsettled or authorities conflict, Erin tells you so. In legal work, disagreement is signal - confidence is earned, not performed.
A sale deed, a research memo, and a client email are different disciplines. Erin handles each accordingly. How it does so is our work; what you receive is yours to judge.
Full OpenAI API compatibility - streaming and non-streaming, standard SDKs in every language, existing tooling such as LangChain, LlamaIndex, and Vercel AI SDK works unmodified.
Swap the base URL and your application speaks legal. Keep your SDK, keep your development workflow, and test Erin against the legal tasks that matter to your product.
Request test accessErin is a GLR Labs research product, and this test version is the first public step in a longer program: legal AI that ultimately runs where the most sensitive work demands it. We are releasing early, watching how real legal work exercises the system, and improving Erin continuously. Test-tier users shape what Erin becomes.
Erin is infrastructure for legal professionals. The output is built for review.
Send a hard drafting, research, review, or citation task and judge the output.