Learn · Free to use today

Learn AI. Ship apps. Write code.

Three shelves, one rule: everything here is free to use and everything here is real. Concepts explain the ideas in plain words, guides walk you through doing it, and build notes show how this site itself gets made. The way in is always the same: see it, understand it, try it, use it.

Start with concepts The free app stack

Concepts

Understand the ideas

Eight ideas behind modern AI, explained the way you would explain them to a smart friend: no jargon, no hype, no math degree required. Read them in order: each one builds on the last. Every concept has two doors: the interactive version in the lab, and the longer written version in the blog — the technical shelf for readers who want the details.

Your words get sliced into tokens, then the model writes an answer.

Concept 1How an LLM Works

Watch how a large language model turns a prompt into text: tokenize, read, predict, repeat.

Written version: What is an LLM?

Concept 2Tokens: The Pieces AI Reads

An LLM never reads whole words: it reads tokens. See how text gets sliced before the model reads it.

Written version: What are tokens?

Concept 3RAG: Answers Grounded in Your Documents

Retrieval-augmented generation finds the right passages in your documents, then lets the model answer from them.

Written version: What is RAG?

Concept 4Embeddings: Meaning as Numbers

How do computers compare the meaning of two sentences? By turning them into lists of numbers.

Written version: What are embeddings?

Concept 5AI Agent vs Chatbot

A chatbot replies; an agent works in a loop toward a goal. Watch the decide-act-observe loop.

Written version: AI agent vs chatbot

Concept 6AI or Plain Logic? When Not to Use AI

An LLM adds uncertainty, cost, and failure modes. A decision guide for when plain code wins.

Written version: When should you NOT use AI?

Concept 7Why AI Hallucinates

A language model predicts text, not truth. The four reasons AI invents convincing-but-wrong details.

Written version: Why AI hallucinates

Concept 8Prompt Anatomy: Writing Clear Prompts

A clear prompt has four parts: task, context, constraints, and format. Watch a vague prompt rebuilt piece by piece: the capstone of the path.

Written version: How to write a clear prompt

Every concept is a simplified educational explanation: the shape of the idea, labeled as such. Prefer watching to reading? The AI Lab animates the LLM, RAG, and agent flows step by step.

Guides

Follow the steps

Step-by-step guides for doing real things, not just understanding them. Start with the stack this very site runs on: zero dollars, honest limits, and the API-key pattern done right.

Push your code and it flows to the world — on free tiers.

Guide

The free app stack

GitHub for code, a free hosting tier for deploys, free public APIs with no key, and the API-key pattern done right: with PlainLogic as the worked example. What you can build for $0, the real limits, and the signs you've outgrown free. No hype, no affiliate links.

New guides land here when they are finished — never as stubs.

Build notes

How this site gets built

Honest notes from building PlainLogic itself: what the site runs on, and what each lab demonstrates under the hood. No manufactured consulting history — just the real thing, as it ships.

Real data goes in, the lab works it, a result comes out. That is the lesson.

Live proof

This site is the demo

PlainLogic deploys from a free GitHub repo to a free hosting tier on every push, and its live data comes from free public APIs. The free app stack guide points at the real pieces of this site: not a hypothetical example.

Code → push → live. That is the whole deploy pipeline.

Build note · Live APIs

Situation Room

How the flagship console works: live weather, water, earthquake, and space-weather data through a server-side proxy that keeps API keys out of the browser.

Build note · Networking

Network Lab

How the internet moves data — packets, routes, retries, latency — animated with labeled states for success, retry, and failure, like real networking code.

Build note · ETL

Data Pipeline Lab

Extract, transform, load on real-shaped data: how each stage fires, and where the honest ETL pain — schemas, dirty rows — actually lives.

Build note · Workflows

Automation Lab

Chains of steps that run on their own: triggers, conditions, and the failure handling real automation needs — an animated diagram with labeled states.

Build note · AI systems

AI Lab

The LLM, RAG, and agent flows as step-by-step animations — every one labeled a simplified educational visualization, never pretending to be model internals.

Build note · Operations

Server Room

Health checks, load, failover, and monitoring: the server-room loop that spots trouble before users do, visualized as it happens.

Honest answers

What Learn is (and is not)

Is this a course? Do I get a certificate?

No. Learn is a collection of guides and working labs, not a course platform. There are no accounts, no progress tracking, no quizzes, and no certificates: just pages you can read and labs you can run, free.

Where should a total beginner start?

With the Concepts shelf, in order: Concept 1 explains the one idea everything else builds on. If you want to build things instead, start with the free app stack guide: it shows what a real deployed app is made of before you write a line of code.

Are the simplified explanations honest?

The guides teach the shape of each idea and say so where they simplify. The AI Lab goes further and labels every animation a "simplified educational visualization." The goal is understanding that survives contact with the real thing, not the appearance of expertise.

Get PlainLogic

Useful AI and technology, explained simply. New tools, practical guides, prompt techniques, and the occasional update worth knowing about.

New useful tools when they ship
Plain-English AI guides
Prompt techniques that work
Important AI updates
New interactive labs

Want better AI results today? Start with the Prompt Playbook — the four-part prompting framework, free on this page.