AIIP.DOWNLOAD · INSTALL · v1.10

Install the protocol.

The fastest path from zero to a running, owned AI operating system. AIIP routes Claude, a local ~70B, ChatGPT, and Gemini through one orchestrated stack on your machine. Four lanes. Ten phases. Yours.

GET THE BLUEPRINT $97 START THE INSTALL

Blueprint credits 100% toward the $497 install within 14 days.

aiip · install · session ~ /Users/operator

INSTALL PHASES

MODEL LANES

HOURS TO RUNNING

API COST CAPABLE

PRE-FLIGHT

SYSTEM REQUIREMENTS

Six things, then you're cleared for install.

No exotic hardware. If you can open a terminal and you have a free afternoon, you can take a clean machine to a fully operational multi-agent workstation.

Machine

Apple Silicon / Linux / WSL2

M-series Mac recommended · x86_64 Linux · Windows via WSL2

Disk

~50GB free

local models, runtimes, and the memory vault

Model keys

your chosen lanes

Claude · ChatGPT · Gemini keys — local Ollama needs none

Notes

Obsidian

the persistent-memory vault that wires context

Time

2–3 focused hours

out-of-the-box to a running, owned stack

Decide

1–2 agents

which agents you'll stand up first

FOUR LANES · ROUTE, DON'T REPLACE

① Claude Deep reasoning

② Ollama ~70B Local · private

③ ChatGPT Breadth

④ Gemini Multimodal

// every credential lives in your OS keychain or env vars — never in source.

The ten-phase AIIP install visualized as a luminous emerald pathway ascending through a dark data cathedral

THE MAP

Ten verified phases. Foundation to operations.

// one luminous path · each step reproducible

THE INSTALL MAP

Ten phases. Each verified.

The install runs as a sequence — foundation to operations. Every step is verified, documented, and reproducible, so the stack you end with is one you can rebuild from scratch.

A single 4–6 hour guided session takes a brand-new Apple Silicon MacBook from out-of-the-box to a fully operational multi-agent workstation. The phases below are what gets installed — every step verified, documented, reproducible.

Foundation

Xcode CLT, Homebrew, Oh My Zsh, Git, GitHub auth

Dev Environment

Node.js v22+ (v24 rec.), pnpm, Python (pyenv), Rust

III

Agent CLI Layer

OpenCode (free) + Claude Code (subscription)

Local LLM Stack

Ollama runtime + Llama, DeepSeek, Hermes, Qwen

OpenClaw Framework

Multi-agent orchestration + 4-lane model routing

Telegram Control Surface DIFFERENTIATOR

BotFather token + Telegram ↔ agent bridge

VII

Self-Improving Agent Loop DIFFERENTIATOR

Hermes Agent + Aider / Cline pair-programming

VIII

Second Brain

Notion + Obsidian, PAPERBRAIN template

Business Showcase

Portfolio + service page deployed live

Backup & Resilience

iCloud Drive + Obsidian sync + Time Machine

Phases VI & VII are what no other install framework foregrounds: Telegram-anywhere control + agents that improve the codebase they run on.

Three glowing emerald agent constructs in formation — operator, scout, and courier — connected by light threads

THE FLEET

Operator. Scout. Courier.

// stand up three — in order, each earning its place

YOUR FIRST 3 AGENTS

Operator → Scout → Courier.

Don't deploy a fleet on day one. Stand up three, in order — a daily driver, a watcher, and a remote. Each one isolated. Each one earning its place before the next.

01 Start here

The Operator

CLI daily driver / router

Your command line and dispatcher. Routes every request across the four model lanes from the terminal.

02 Then

The Scout

Headless monitor

Runs unattended. Watches, researches, and reports — no window required, no hand on the wheel.

03 Finally

The Courier

Telegram control surface

Control the whole stack from your phone. Commands in, results back — anywhere you are.

A matte-black workstation on a dark desk emitting emerald light and a holographic data lattice — sovereign local AI hardware

THE PAYOFF

Your hardware. Your models. Your stack.

// local-first · $0-API capable · yours forever

WHAT YOU GET

A stack you own — not a tab you rent.

Own the stack

One install you own outright. No rented brain, no vendor that can switch you off, no monthly meter.

Local-first

Runs on your hardware. Local inference, local memory — the cloud is an option, never a requirement.

$0-API capable

Built to run paid-key-free on a local ~70B. Route to cloud lanes only when you decide it's worth it.

Reproducible

Scripted, verified, documented. Rebuild on a fresh machine to prove the whole thing reproduces.

GET INSTALLED

PICK YOUR PATH

Start with the map. Or have it installed for you.

STEP 00

THE BLUEPRINT

$97

one-time

The architecture + the full 10-phase install map, as a PDF. Credits 100% toward the install within 14 days.

GET BLUEPRINT $97

DO IT YOURSELF

SELF-SERVE INSTALL

$497

one-time

All 10 phases scripted, plus 30 days of direct support for config drift, dependency conflicts, and breakage.

SELF-SERVE $497

WE INSTALL IT

DONE-FOR-YOU

Book

15-min call

End-to-end install, run with you. We take a clean machine to a live, owned operator stack.

BOOK A STRATEGY CALL

// want the full picture — layers, ladder, and white paper? aiinstallprotocol.com ↗

DON'T WANT TO INSTALL IT YOURSELF?

Have us build the whole operation.

We don't demo it — we run on it. The same stack runs the entire PaperChaseWebb operation, forged across 60.5M+ tokens of real operator work. Custom buildouts from $25K to $285K+.

BOOK A STRATEGY CALL SEE THE BUILDOUT LADDER →

Zero to a running stack.
Install the protocol.

GET THE BLUEPRINT $97 BOOK DONE-FOR-YOU

LOCAL-FIRST · $0-API CAPABLE · YOURS FOREVER