A hands-on guide to modern AI

Master Claude
from zero to

AI Learn takes you the whole way: what AI actually is, how to install and set up Claude Desktop & Claude Code on any system, which Claude model to pick for every job, how to tune effort from low all the way to Ultracode — and how to plan, code, and test real software with AI at your side.

0
Claude models explained
0
Effort levels — low → Ultracode
0
Tokens of context window
0
Max output tokens
Foundations

What is AI — and what is Claude?

Before the tools, the ideas. Three concepts unlock everything else on this site.

🧠

Large Language Models

An LLM is a neural network trained on enormous amounts of text. It learns patterns of language, code, and reasoning — then predicts the most useful continuation of whatever you give it. That simple loop powers chat, coding, analysis, and agents.

Claude, by Anthropic

Claude is Anthropic's family of frontier AI models — built with a focus on being helpful, honest, and safe. It reads up to 1 million tokens of context, reasons step-by-step with adaptive thinking, and works with text, images, PDFs, and code.

🤖

Agents, not just chat

Modern AI doesn't only answer — it acts. Claude Code turns Claude into an agent that reads your files, runs commands, writes code, tests it, and iterates until the job is done. You direct; it executes.

The core loop

How working with Claude actually works

Whether you're chatting in Claude Desktop or running an agent in Claude Code, every session follows the same five-beat rhythm.

01

You set the goal

Describe what you want — in plain language. Good prompts state the goal, the constraints, and what "done" looks like. The clearer the destination, the straighter the path.

02

Claude thinks

With adaptive thinking, Claude decides how deeply to reason about your request — a quick answer for simple questions, extended step-by-step reasoning for hard ones. You control the depth with effort levels.

03

Claude acts

In Claude Code, thinking becomes doing: reading files, searching your codebase, editing code, running shell commands. Each action's result feeds back into the reasoning loop.

04

Claude verifies

Great AI workflows close the loop — running tests, checking output, re-reading changed files. Chapter 5 shows you how to make verification automatic instead of optional.

05

You review & iterate

You stay the director. Review the result, refine the instruction, and go again. Each iteration compounds — and with Ultracode, whole teams of agents iterate in parallel.

Quick answers

Before you dive in

Do I need to know how to code?
No. Chapters 1–3 assume nothing — you'll get real value from Claude Desktop alone. Chapters 4–5 are about coding with AI, and they're actually a great way to learn to code, because Claude explains everything it does.
Is Claude free to use?
Claude.ai and Claude Desktop have a free tier. Paid plans (Pro/Max) unlock more usage, better models, and Claude Code. API access is pay-per-token — the Models chapter covers exact pricing per model.
What's the difference between Claude Desktop and Claude Code?
Claude Desktop is the chat app — conversations, files, images, projects. Claude Code is the agent — it lives in your terminal (or IDE), works directly on your files, runs commands, and completes multi-step engineering tasks autonomously. You'll install both in Chapter 1.
What exactly is "Ultracode"?
Ultracode is Claude Code's multi-agent orchestration mode — instead of one AI working alone, a coordinator spins up whole workflows of parallel agents that research, build, review, and adversarially verify each other's work. It sits at the very top of the effort ladder. Chapter 3 explains when it's worth it.

Ready to begin?

Fifteen minutes from now you'll have Claude installed and your first real conversation running.

Start Chapter 1 — Getting Started