Foundry
An agentic delivery platform, open-sourced in alpha.
Multi-tenant platform that turns plans and conversations into working deliverables — decomposing requirements, provisioning AI sandboxes, and shipping code to repos.
A four-process distributed system. Traverse it.
Click any process to inspect it, or run a request through the whole pipeline and watch the audited log stream — intake through autonomous code generation and human review.
Foundry closes the gap between idea and implementation. Feed it plans and conversations and it decomposes them into structured requirements, reasons about implementation, provisions AI sandboxes, and ships code to repositories — every query indexed, every mutation auth-gated, every agent action auditable.
Why it exists
Most AI coding tools treat code generation as an isolated act. Real delivery requires structured context: which requirements map to which workstreams, what the design constraints are, which repos own which capabilities. Delivery knowledge is the moat — the platform that structures it wins. Foundry is a solo-founder build, open-sourced not because it is finished but because the problem is too large for closed development.
Architecture
A four-process distributed system: a browser and desktop client connected over WebSocket to a reactive Convex backend, with AI inference and sandboxed execution handled by Cloudflare edge workers. Agent tasks run in isolated Docker containers behind a ten-stage provisioning state machine, each with its own git worktree, an immutable audit trail, and a real-time log stream.
Reference engagement
The reference client, Burlington Medical, runs the full pipeline — 118 requirements, 8 skills, 7 workstreams — from intake through autonomous code generation and human review.
The question isn’t whether this transformation happens — it’s what the delivery platform looks like when it does.