OS·WholeTech ↩ WholeTech network
The AI infrastructure handbook

Set up your AI stack on any device you own.

Claude Code, Codex, Gemini, Tailscale, Ollama — the five tools every modern AI setup runs on. This site shows you exactly how to install all of them, one device at a time, in plain language with copy-paste commands.

Built for anyone who keeps hearing the names but hasn't had one clean place that walks through the install. Pick your device below and follow the same six steps every time.

8 device & cloud guides 5 tools per device Same 6 steps everywhere Maintained by the WholeTech network
Start here

Pick the device you're setting up

Each guide is self-contained — open the one for the machine in front of you and work top to bottom. Got a lot of one kind? You only learn it once, then repeat.

The pattern

Every guide follows the same six steps

Once you've done it on one device, the rest feel familiar. You're always doing these six things, in this order:

Step 1

Terminal & package manager

Get a command line open and a way to install software from one (winget, Homebrew, apt). Everything else rides on this.

Step 2

Claude Code

Anthropic's coding agent — the one you're talking to right now. Lives in your terminal and does real work on your files.

Step 3

Codex

OpenAI's command-line coding agent. A second opinion / second engine that works the same way as Claude Code.

Step 4

Gemini CLI

Google's command-line AI. Big free allowance and a huge context window — good as a third tool in the rotation.

Step 5

Tailscale

A private network that lets all your devices see each other securely, anywhere. The glue that makes 15 machines feel like one.

Step 6

Ollama

Runs AI models locally on your own hardware — free, private, offline. Share it across the house over Tailscale.

Plain English

What each of these actually is

If the names blur together, read this once. No jargon — just what the tool is, who makes it, and why you'd want it.

ToolWho makes itIn one sentenceWhy you want it
Claude CodeAnthropicAn AI coding assistant that lives in your terminal and edits real files.Builds and fixes websites, scripts, and documents by itself. The most capable of the three.
CodexOpenAIOpenAI's version of the same idea — a command-line coding agent.A strong second engine; handy when you want a different model's take.
Gemini CLIGoogleGoogle's command-line AI, with a very large free tier.Generous free usage and a huge memory for long documents.
TailscaleTailscale Inc.A "VPN" that privately connects all your devices to each other.Reach any machine from any other — phone to PC, laptop to NAS — securely, from anywhere.
OllamaOllama (open source)Runs AI models on your own computer instead of the cloud.Free, private, works offline. No subscription, nothing leaves your house.
💡The mental model: Claude Code, Codex, and Gemini are the workers (they do tasks). Tailscale is the roads between your machines. Ollama is a worker you own outright that never sends your data anywhere. You don't need all five on every device — the guides tell you which ones make sense where.

🗺️ Want the whole picture? Read the tool map

Every AI tool in widespread use — chat assistants, coding agents, AI editors, local models, MCP — organized into four simple layers, with how they interrelate and how to future-proof yourself as new tools (like Anthropic's Mythos) arrive. The one page to read before everything else.