AI, made visual and interactive.

Learn how AI actually works by playing with it. Every lesson and lab runs the real algorithm in your browser: drag the data, change the settings, and watch the model learn.

Start learning free Try a lab

47 lessons, 25 labs and hundreds of practice questions. Nothing to install. See the roadmap

Live: a neural network learning in your browser. Tap to add points.

epoch 0

loss 0.690

acc 50%

Place

A 2-8-8-1 network, tanh activations, trained with Adam. Shift-click, or pick a colour below the canvas, to place the other class.

How it works

1Read one idea
Short sections, plain words, a picture for every concept, and the maths only when it helps.
2Play with it
Change the inputs and watch real K-means, backpropagation, attention or A* respond. Simplifications are labelled.
3Make it stick
Quick checks as you read, then five-minute practice sessions that bring back what you miss.

Linked from university courses and classrooms.Using it to teach?

Pick a path

5 paths that alternate reading with doing. Start with Foundations if you are new; jump to any other if you already know the basics.

See every step

Foundations
22 steps
Start here. What learning from data means, how networks train, and how to judge the result.
1AI Fundamentalslesson 2The Maths Behind MLlesson 3Machine Learninglesson 4Logistic Regressionlesson 5k-Nearest Neighbourslesson 6Decision Tree Builderlab 7Forests and Boostinglesson 8Boosting Lablab 9Support Vector Machineslesson 10Naive Bayeslesson 11Evaluating Modelslesson 12Confusion Matrix Explorerlab 13Overfitting and Regularisationlesson 14Neural Networks and Deep Learninglesson 15Backpropagationlesson 16Optimisers: SGD, Momentum and Adamlesson 17Gradient Descent Lablab 18Neural Network Playgroundlab 19Unsupervised Learninglesson 20Clustering Visualizerlab 21AI Ethics and Safetylesson 22Fairness Lablab
Start Foundations
Seeing
6 steps
How machines read images, from a single convolution to detecting every object in a scene.
1Computer Visionlesson 2Image Filter Playgroundlab 3Convolutional Neural Networkslesson 4Object Detectionlesson 5Object Detection Lablab 6Image Segmentationlesson
Start Seeing
Language
12 steps
From counting words to the transformers inside every modern chatbot.
1Natural Language Processinglesson 2Text Playgroundlab 3Tokenizationlesson 4Tokenizer Lablab 5Recurrent Networks Lablab 6Machine Translationlesson 7Transformers and Attentionlesson 8Attention Visualizerlab 9Transformer Explorerlab 10Generative AIlesson 11Diffusion Lablab 12Speech and Audio AIlesson
Start Language
Deciding
8 steps
How AI plans routes, learns strategies from reward, recommends what you see next and acts in the physical world.
1Search and Optimizationlesson 2Pathfinding Visualizerlab 3Travelling Salesmanlab 4Reinforcement Learninglesson 5Reinforcement Learning Arenalab 6Recommender Systemslesson 7Recommender Lablab 8Robots and World Modelslesson
Start Deciding
Frontier
24 steps
The ideas behind the AI in the news: LLMs, retrieval, agents, decision models like Jev, and multimodal systems.
1Large Language Modelslesson 2Next-Token Sampling Lablab 3Prompt and Context Engineeringlesson 4Scaling Laws and Pretraininglesson 5Post-Training and Alignmentlesson 6Reward Model Lablab 7Reasoning Modelslesson 8Opus 5.5, Fable 5.1 and Looped Transformerslesson 9GPT-6.1 Sol, Gemini 4 Argon and DeepSeek V4.1 Flashlesson 10Claude Haiku 5.5, Muse Wonder and the price of a tokenlesson 11Muse Spark Open, parallel agents and a frontier model on your desklesson 12GPT-6.1 Terra, Ultrafast and agents that use a computerlesson 13GPT-6 Rosalind, robot brains and AI that leaves the chatlesson 14Mixture of Expertslesson 15Mixture of Experts Lablab 16Embeddings and Retrievallesson 17Embedding Explorerlab 18AI Agents and Tool Uselesson 19Decision Models and Jevlesson 20Calibration Lablab 21Multimodal AIlesson 22Fine-Tuning and Efficient Modelslesson 23Model Compression Lablab 24Interpretabilitylesson
Start Frontier

Five minutes a day. It sticks.

Practice turns every lesson into quick hands-on questions: drag a slider, tap a plot, sort the steps. Build a streak, earn stars on every skill, and the ones you miss come back until they don’t.