Most open source projects die in the same place: between the people who love them and the people who pay.
We’re launching Open Mercato Sandboxes. Here’s the founder’s view of why this matters beyond another product release.
Geoffrey Moore called that place the chasm. Innovators buy technology. The early majority buys a whole product: tools, people, training, support. Ship only the core and you stay on the wrong side.
We spent 12 months at Open Mercato learning this the expensive way. The codebase is over 1,4M lines, MIT licensed, 140+ contributors. Not one line written by a human. And none of it, by itself, gets a CIO to sign.
What moved enterprise conversations was the rest of the stack:
- Open source core plus an agent harness: specs, skills, conventions. Claude Code and Codex build features, not architecture.
- Enterprise modules. The big one is Agent Orchestrator: the agent proposes, a human approves. No large company lets an agent near production without that gate.
- A partner program. Certified agencies that help large clients right away.
- Cezar, an open source orchestrator for coding agents. Queue 30 tasks, set one limit, close the laptop.
- AI Engineering training at scale. From code down to what actually blocks a rollout.
- Sandboxes. All of the above in one place. Describe the app in plain language, the agent builds it inside the Open Mercato architecture. Works for a CFO and for a CTO.
Sandboxes are the capstone. That's why this is our biggest launch of the year, not a feature.
The founder lesson: code was the cheapest layer. The expensive layers were the ones agents can't build. Partners, training, trust.
Founders ask me when to add partners or training. My answer: when you want the early majority. Earlier is a cost. Later is too late.
