← AI 🌀

Markov Chain Sentence Generator

Feed it a paragraph, it 'writes' new sentences by statistically remixing word pairs -- the real trick behind old-school text generators.

🔴 HardDifficulty

4-6 hoursTime

Python, JavaScript, TypeScriptLanguages

What you're building

You're building a real Markov chain text generator -- the same core algorithm that powered goofy 'AI' text generators for years before large language models existed. It reads a training text, learns which words tend to follow which other words, then generates brand-new (often hilarious, sometimes eerily coherent) sentences by randomly walking that word-chain. This one is genuinely a statistical model, just a very small and simple one.

What you'll unlock 🔓

Probabilistic algorithmsHash mapsString tokenization

You'll need

Comfortable with JavaScript
Objects and arrays
A browser

🗺️ Build roadmap

01 Build the training-text input and generate button → 02 Tokenize the training text into words → 03 Build a word-pair transition map → 04 Implement the random walk generator → 05 Handle sentence-ending logic → 06 Add a chain-order toggle (1-word vs 2-word memory) → 07 Add a 'visualize the chain' debug panel

🔒

Log in to start building

Your first project is free -- create an account to unlock it.