← All work  ·  Case 09 of 09 · Stopped (a decision, not a feature)

Enterprise Healthcare GRC SaaSProduct OwnerSolo POC2026 · Stopped pre-launch

The report-generation POC I stopped on purpose.

A solo AI prototype for automated compliance report generation, built the same way as the quote automation system. Nearly demo-ready. A customer contract clause ruled out AI-generated content. Stopped in the same meeting.

Solo
Built end to end
with Claude Code
0
Customer contracts
breached
Same meeting
From surfacing the blocker
to stopping the build
01The setup

Same pattern as quote automation. A different product line.

I set out to automate report generation for another product in our Enterprise Healthcare GRC portfolio, the same way I'd built the quote automation prototype: solo, with Claude Code, no engineering team, from problem to a working system. The build went well. Document ingestion, structuring, and draft report generation were all working end to end, ready to demo.

02What surfaced

The contract clause was older than the project.

I brought the finished POC to a demo with our director. That's where I learned that customer contracts for this product line included a clause, negotiated well before I ever touched the project, that explicitly ruled out using AI to generate report content. If a customer discovered AI had touched their compliance report, that wasn't an awkward conversation, it was real legal exposure.

The uncomfortable finding: the build worked. It would have shipped. It could not be used without creating exactly the kind of risk this product exists to help customers avoid.
03The stop

Stopped in the same meeting.

I stopped work on the POC on the spot, before the demo was even over. No trying to scope it down, no partial version, no "what if we just used AI for the first draft." A contract clause written around AI involvement doesn't have a smaller-AI compromise. The only correct move was to stop completely, immediately, and say so plainly.

"A contract clause doesn't care how good the demo is."
04Lessons

What this taught me about speed.

Speed cuts both ways. Building fast and solo with Claude Code means going from idea to a working POC before a slower, team-based process would even finish scoping it. It also makes it easier to skip the discovery step that slower process would have forced. That's the trade I'm watching for now: the same speed that makes solo AI-assisted builds valuable is what let this one get all the way to a demo before the constraint surfaced.

Contract terms are a discovery input, not a legal-review afterthought. The clause existed before I opened Claude Code. I just hadn't asked the question that would have surfaced it.

A working demo isn't the same as a viable product. This one worked perfectly and still couldn't ship. Those are two separate bars, and only one of them is about the technology.

Let's build what actually ships.