README.md

Software — Bringing the Box to Life

Hardware is half of a Personal AI Computer. This is the other half: getting the OS, drivers, and BIOS right, verifying the GPUs, and then serving open models locally so your intelligence runs on your machine and answers to no one.

Status: Setup and testing are documented below. The local model-serving guides (Ollama / vLLM / llama.cpp, picking open models, connecting an agent) are in progress — see run-models.md.

Steps

1. Setup — BIOS tuning for multi-GPU, NVIDIA drivers, and OS install.
2. Testing — verify every GPU is seen and stable before you rely on it.
3. Run models — serve open models locally and point an agent at them. (In progress.)

Per-build specifics

The steps here are common to all three builds. Exact BIOS menu paths and board manuals differ per motherboard — see each build's own docs/ for board-specific detail:

• 2× RTX 5090 — docs
• 4× RTX PRO 6000 — docs
• 8× GPU — docs