gAI Lab

We build in the open.

The Lab is where our engineering team releases the tools we build for our own AI work. Free, open source, and used by developers well beyond our portfolio.

Project 01Open source · npm

context-doctor

See where your AI agent's tokens go, and stop paying for the ones that do nothing.

AI agents re-send their whole context on every message: old tool output, pasted files, long instructions. context-doctor measures that waste on your own sessions, in dollars, then removes it with an every-prompt hook, smarter compaction, a cache-safe proxy, RAG chunk packing and an MCP server.

Try it: what your recent sessions cost

npx context-doctor savings

Why a venture firm ships open source

We use what we build, so we know what it takes to build.

  • For founders

    The engineers who wrote this build your product.

    When you co-build with us, the same team ships your stack. Inference cost and agent reliability are problems we solve in code, and the fixes go into your product from day one.

    Build with us
  • For investors

    We see AI costs from the inside.

    Running agents at scale ourselves keeps our view of AI margins grounded in real usage, not vendor decks. It shapes how we underwrite the companies we build and back.

    How we invest