2
1 Comment

Built a CLM for 3 Years. Realized It Wasn’t Needed. Open-Sourced the Best Part.

For the past 3 years, we have been building a CLM tool with built-in AI features. It included contract review, contract drafting, and data extraction using AI. We spent all our time improving accuracy, responses, and citation support. But we recently realized the full tool is no longer needed, so we are stopping it. Still, we didn’t want our work to go to waste, so we open-sourced the contract review part.

While working on our CLM, we focused on trust in contract review, aiming to build something lawyers and paralegals could actually rely on and verify, not just a “chat with PDF” tool that sounds correct. So we added a citation feature that shows exactly which part of the response matches which clause in the contract. So instead of hallucinating or guessing, it sticks to what’s actually written in the contract.

We also noticed many contracts don’t contain all the details themselves and instead refer to other files like amendments or annexures. So we built support for that too.

So we released LexReviewer. It’s open source and follows one rule: no answer without a direct source link.

Repo is here: https://github.com/LexStack-AI/LexReviewer

Check it out if you’d like, and tell us if there’s a feature you want or something else you’d want us to release.

on February 27, 2026
  1. 1

    Three years is a hard thing to stop. The fact that you open-sourced it rather than just shelving it says something I think, that the citation-grounded approach is genuinely the right idea for legal review, and it deserves to exist somewhere even if the full product didn't find its market.

    I'm curious what the signal was that told you the full CLM wasn't needed. Was it that the market already had something good enough, or that the buying process was too long, or something else? The decision to stop is usually clearer in hindsight than it feels at the time.