Fleet Command
Connect this UI to your always-on Fleet Command host — the durable coordinator for workflows, workers, and scheduling. GPU workers on Vast are separate.
Checking Fleet Command…
Fleet Command is your always-on coordinator host — not a rented Vast GPU worker.
Connect
Prefer auto-detect when Fleet Command is running locally (http://127.0.0.1:8080). For Vercel → always-on host, paste the public tunnel URL after setup.
Install on Ubuntu
Foolproof one-script setup. Downloads a self-contained Fleet Command package from this site — no GitHub username, password, or token prompts.
Neon on Vercel is used automatically when present. macOS installer comes next.
Local models (LM Studio)
On the Fleet Command host, start LM Studio → Developer → Local Server on http://127.0.0.1:1234. Setup writes AISPACEFLEET_LM_STUDIO_BASE_URL and prefers local models. For GPU workforce on other machines, use Offline capital ship.
Manual setup (optional)
Prefer cloning by hand? Use the repo scripts instead of the downloadable installer.
- Clone the repo and run
bash scripts/ubuntu/setup-fleet-command.sh. - Point both Vercel and Fleet Command at the same Neon DATABASE_URL.
- Copy env with
node scripts/fleet/print-vercel-env.mjs, then Connect (or wait for tunnel auto-publish).
Advanced — temporary Vast host (not recommended for always-on)
Fleet Command should run on your own always-on machine. Renting a Vast box only makes sense as a temporary experiment — GPU workers are separate and stay under Capital Ships / Add Remote Worker.
Open temporary Vast host tools →