Your own private AI, powered by llmcli and llama-server.

The Private AI Setup – H100 is a complete, production‑ready stack that runs entirely on your own hardware — no internet connection required for inference. It combines the llmcli command‑line agent (tool‑calling, session memory, prompt scripting) with the lightning‑fast llama-server inference engine, all secured by a hardened OpenBSD host. Perfect for organisations that cannot risk sending internal data to third‑party APIs.

What you get

• llmcli agent — full‑featured CLI agent: multi‑turn conversations, tool calling, state management, and prompt injection
• llama-server — the canonical llama.cpp inference server, CPU‑ and GPU‑friendly, zero cloud dependency
• Hardened OpenBSD host configuration — our production pf.conf ruleset, relayd TLS termination, and minimal attack surface
• Installation playbook — step‑by‑step from bare metal to a working llmcli session in under two hours

Why self‑hosted?

• Absolute data sovereignty — documents, emails, and conversations never leave your network.
• GDPR‑ready — no data processor agreement, no US‑based API, no third‑party risk.
• Air‑gap capable — law firms, clinics, and accountancy practices can run it completely offline.
• One‑time cost — no per‑token billing, no monthly API subscription.

Technical summary

After purchase, you’ll receive a secure download link and a setup guide that takes you from a fresh server to chatting with your own llmcli agent in under two hours.