Run AI privately on your own hardware — no cloud, no data leaks, no subscriptions.
Instant PDF download. No subscription.
Every prompt you send to a cloud AI is a data point. It gets logged, retained, and potentially used to train the next model. For most tasks, that tradeoff is fine. For sensitive client work, proprietary code, and confidential business data, it isn't.
This guide gives you the setup, the tools, and the workflows to run capable AI locally — so you decide what leaves your machine, and what doesn't.
Minimum specs by model size, the Apple Silicon advantage, CPU vs GPU performance numbers, and three fully-costed hardware tiers — from using what you have to a purpose-built local AI machine.
Step-by-step installation for Mac, Windows, and Linux. Pull and run your first model in under 10 minutes. Run both tools as local API servers compatible with any OpenAI-compatible client.
8 model profiles — Llama 3 8B, Mistral 7B, Phi-3 Mini, Gemma 2, CodeGemma, DeepSeek Coder, Qwen2, and Dolphin Mixtral — each with RAM requirement, use case, honest assessment, and pull command.
6 fully-documented workflows: private document analysis, local coding assistant (VS Code + Continue), offline research, private email drafting, basic RAG setup, and OpenAI-compatible API integration.
Every workflow includes: Tools, Setup Steps, Prompt Pattern, and Privacy Notes.
If you use AI tools daily and have ever hesitated before pasting sensitive client data, confidential documents, or proprietary code into a chat window — this guide shows you how to run the same capabilities privately on your own hardware.
In the Marines, every piece of equipment came with a manual. Darnell Baker applies that same logic to building and running businesses — documenting the systems that solo founders actually need to operate at scale.
Oakland, CA | LinkedIn | FounderFieldManuals