17-Page Windows Setup Guide

Run Private AI
on Your Windows PC

ChatGPT costs $20+/month and logs everything. This guide shows you how to run models just as capable — free, offline, permanently — right on your own machine.

Get the Guide — $9.99
$9.99 one-time  ·  instant PDF download  ·  no subscription
PowerShell

PS C:\Users\You> ollama run llama3

pulling manifest... done

running on your machine. 0 requests to OpenAI.

100% Private & Offline
17-Page PDF Guide
Instant Download
No Subscription — Ever
Works on Any Windows PC
The Problem

You don't need a subscription
to run powerful AI

The tools exist. They're free and open-source. You just needed someone to write every step down clearly — that's what this guide does.

Paying $20+/month for ChatGPT?

That's $240/year, every year. Local models now match GPT-4 quality on everyday tasks. This guide is a one-time $9.99 — no renewals, ever.

Your conversations aren't private

Every prompt you send to cloud AI is stored and used to train future models. Local AI runs entirely on your machine — zero data leaves.

Setup guides are scattered and vague

Every doc assumes you already know half the answer. This guide is numbered steps, exact commands, nothing left ambiguous — start to finish.

What's Inside

10 sections. Everything you need,
nothing you don't.

Written for Windows users of every skill level — from AI-curious to developers building local API integrations.

System Requirements by RAM Tier

Hardware tables for 8 GB, 16 GB, 32 GB, and 64 GB+ machines with GPU guidance. Know what runs before you download anything.

Model Recommendations per Tier

20+ models compared by quant type, file size, RAM requirements, and best use cases. No guessing which model fits your machine.

LM Studio Complete Walkthrough

Download, install, find models, configure GPU offload layers, and enable the local OpenAI-compatible API — every step documented.

Ollama Complete Walkthrough

Install guide, full CLI reference, API verification, and writing custom Modelfiles for prompt templates and system instructions.

Open WebUI Setup

Full browser-based ChatGPT-style interface for Ollama via Docker and pip — chat with local models through a polished web UI.

Performance Tuning

GPU layer offloading, context window sizing, quantization trade-offs, and CPU thread config. Get 3–10x faster responses from your GPU.

Troubleshooting Guide

10 most common errors with exact causes and working fixes. Stop searching Stack Overflow — every major blocker is documented with the solution.

Quick Reference Cheat Sheet

Commands, file paths, PowerShell API snippets, and keyboard shortcuts on one page — bundled as a separate PDF in the Complete tier.

Who It's For

Built for people who want AI
on their terms

If any of these describe you, this guide is exactly what you need.

The Subscription Escaper

Paying $20+/month for ChatGPT and want a permanent free alternative with no ongoing cost.

The Privacy-Conscious User

You don't want your ideas, code, or conversations stored on someone else's server. Local AI means nothing leaves your machine.

The Curious Windows User

You've heard about local AI but don't know where to start. This takes you from zero to running in a single afternoon.

The Developer

You want a local OpenAI-compatible API for your apps — test prompts, build tools, ship without rate limits or per-token cost.

What You Get

Two files. Every answer included.

Professionally formatted PDFs, ready immediately after purchase.

  • Main guide PDF — 17 pages, 10 sections, professionally formatted
  • Local AI Cheat Sheet PDF — standalone quick reference (Complete tier)
  • Model comparison tables — 20+ models across all RAM tiers
  • Quantization guide — Q4, Q5, Q8 explained with real RAM figures
  • Troubleshooting table — 10 errors, 10 exact fixes
  • No DRM — download and keep forever, no account required
17
Pages
10
Sections
20+
Models
10
Errors Fixed
2
PDF Files
$0
Ongoing Cost
Requirements

If you have a Windows PC,
you're already ready

No Linux, no coding experience, no expensive hardware required. GPU speeds things up — but the guide covers CPU-only from start to finish.

Windows 10 or 11 (64-bit)
8 GB RAM minimum
GPU optional — works CPU-only
No coding experience needed
4–20 GB free disk space for models
Pricing

One guide. Two tiers. No subscription.

Pick the version that fits you — both are instant downloads.

Guide Only
$ 9.99
One-time payment
  • 17-page setup guide PDF
  • LM Studio + Ollama + Open WebUI
  • Model comparison tables
  • Troubleshooting table
  • Cheat Sheet PDF not included
Get the Guide
Best Value
Guide + Cheat Sheet
$ 12.49
One-time payment
  • 17-page setup guide PDF
  • LM Studio + Ollama + Open WebUI
  • Model comparison tables
  • Troubleshooting table
  • + Local AI Cheat Sheet PDF
Get the Bundle

Secure checkout via Gumroad. Questions? indigonynja@gmail.com — I reply within 24 hours.

FAQ

Common questions

Do I need a powerful or expensive PC?
No. The guide starts at 8 GB RAM and covers exactly which models work at every tier. A GPU speeds things up 3–10x, but the guide covers CPU-only operation fully. Even an older laptop can run useful AI models.
Will this work on Windows 10?
Yes. The guide covers both Windows 10 and Windows 11 (64-bit). All three tools — LM Studio, Ollama, and Open WebUI — support Windows 10.
Do I need coding experience?
No. Every step is written in plain language with numbered instructions. Some sections include optional PowerShell snippets for developers, but you can skip those entirely and still complete the setup.
What AI models are covered?
20+ models including Llama 3, Mistral, Phi-3, Gemma, and others — with specific recommendations by RAM tier. You'll know exactly which model to download before you start.
What's the difference between the two tiers?
Both include the full 17-page setup guide. The Complete tier ($12.49) also includes the Local AI Cheat Sheet — a separate PDF with commands, file paths, PowerShell snippets, and shortcuts on one printable page.
Is there a refund policy?
Yes. Gumroad handles all purchases and supports refunds. If the guide doesn't deliver what's described, email me and I'll sort it out.