Run AI on the computers
in your office.
The Mac Studios on your designers' desks, the GPU box in the corner, the workstation idle overnight — Grid pools them into one private AI network your whole team shares: chat, images, video, and code — all on your own hardware.
$curl -fsSLhttps://grid.autonomous.ai/install.sh| bash
All the apps you love. Now local.
Keep the coding agents and tools you already use, exactly as you use them. Grid speaks the OpenAI API — point OpenClaw, Hermes, Codex, or any OpenAI-compatible app at your grid and they answer from your own hardware, not someone's cloud.
Your work is your edge. Don't feed it to a frontier lab.
Every prompt to a hosted model hands over your code, your clients, your data — to store, learn from, and build on. Grid runs on machines you own, so your edge stays yours: nothing stored, nothing trained on, no one turning your work into their product.
Everyone else is racking up a token bill. Yours is $0.
Token bills are exploding — one company ran up a reported $500M in a month, and Meta capped its own staff after 70+ trillion tokens. Grid has no meter: it runs on computers you already own, so your only cost is electricity — the bill stays $0.
.webp)
One machine isn't a grid.
Your whole team is.
Ollama, vLLM, and LM Studio each power one machine. Grid is the orchestrator on top — everyone joins and adds their box, so your grid gets bigger and faster as the team grows. One shared endpoint, all of it.
Stop renting AI.
Run it in-house.
Built in the open.
The core is MIT-licensed and OpenAI-compatible — read every line, run it yourself, and follow development on GitHub.