A developer asked for API access. I shipped it the same week. Here's what changed.

One of my early users was a developer. He had an AI agent connected to 30+ services. He wasn't asking about the consumer app. He asked: "What would make this useful inside my workflow?"

I had no API. I built one today.

Here's what I actually shipped — three endpoints, auth-gated, rate-limited at 100 req/hour:

POST /api/v1/doors — drop a situation, get a mirror (brutal diagnostic of what's keeping the person stuck) + three structured paths with typed fields + situation_id

POST /api/v1/choice — record which door was chosen, returns choice_id for the data chain

POST /api/v1/interrogate — raise an objection on the chosen door, get a refined path that is harder than the original door, not easier

What I didn't expect: the API changes the data moat thesis entirely.

Every call from any developer's app, any agent, any n8n workflow writes to the same Supabase tables as the consumer app. The dataset compounds from multiple surfaces simultaneously. A developer integrating ZYVV into their workflow isn't a distribution cost. They're a data source.

The real value isn't the interface. It's the accumulating record of what stuck humans do when handed three genuinely competing paths.

API access is open. If you're building something and want a key — drop your email in the comments or DM me.