A Y Combinator-backed law firm recently shipped a production MCP server. The product: submit a contract, get attorney-reviewed redlines delivered back through the same AI environment where you’re working. No tab switching, no external upload, no waiting for a separate workflow to complete. The review comes back into the conversation.

The interesting thing about this isn’t the product. It’s what building it required them to believe.

They had to believe that professionals — attorneys, their clients — would trust AI-mediated document workflows for consequential work. They had to believe that MCP-native delivery was a better user experience than existing web-based tools. They had to believe the market was large enough and the bottleneck clear enough to warrant building infrastructure around it.

YC backed those beliefs. That’s notable.

The document review via MCP model is now validated at the highest level of startup signal available. Not by a blog post. Not by a survey. By a funded company making the exact same bet on the exact same user behavior in an adjacent vertical.

The adjacent part matters. A law firm with MCP-delivered contract review and an acquisition firm with MCP-delivered lease abstraction are not competitors. They’re parallel applications of the same insight: professionals doing document-intensive work benefit from keeping that work inside their AI environment rather than exporting it to separate tools.

Adjacent validation is usually better than direct validation. Direct validation means someone already built it — which often means the window has closed. Adjacent validation means the core insight is proven but the specific application is still open.

The window is open. The YC signal says: this category works. +++