The Lab

An independent AI software lab, built around one idea.

PruneLabs builds AI tools that help developers reduce complexity before it spreads through the codebase. Run by one person. Joined by agents.

Why this lab exists

Most AI coding tools optimise for output volume. They generate more code, more files, more abstractions. The lab is built on the opposite belief: software gets better when it gets smaller.

How it works

Products are designed in pairs with the agents that ship them. Each tool answers one question, runs in one workflow, and is held to the same simplicity standard as the code it produces.

What it ships

Two connected tools for cleaner code, PruneAgent and PruneOS, plus Graft, a newer direction that connects your product to the tools around it. Each one stands alone, held to the same simplicity standard.

AI coding should not mean more code by default. It should mean better, smaller, cleaner software.

PruneLabs, on building

What the lab is shipping next.

Tools at different stages. Each one earns its place before it ships.

  1. PruneAgentCreateIn development. Chat surface and small-edit defaults landing first. Private alpha next.See PruneAgent
  2. PruneOSConstrainIn development. Prune-plan output, deletion-first heuristics, and PR preparation are being finalised before release.See the PruneOS workflow
  3. GraftConnectIn design. The integration delivery model and the verification harness that keeps connectors green are being designed now.See Graft