5 live games, built with AI

AI-Assisted Game Experiments

Small interactive products built through AI-assisted experimentation, rapid prototyping, and creative problem-solving.

Each game went from a rough idea to a live, shareable link with AI as a working partner. This page shows the games, the tools behind them and the full build process — as evidence of how I experiment, not as a claim to be a game developer.

Explore the Games View Live Projects
Quiet Detective Stone Rush Rune Cricket Falling Stars Table Tennis

Tap any sticker to play.

Why build games?

I don't only use AI to generate content. I use it to build, test and learn from interactive products.

These projects explore how far AI can carry an idea toward a working experience — shaping the concept, spelling out the rules, building them, then iterating until a stranger can open the link and play.

Games are a brutally honest testing ground. If a control feels wrong or a rule is unclear, you know within seconds. That makes them a fast way to practise the loop behind good growth and product work: form a hypothesis, ship something small, watch how it behaves, improve it.

The goal isn't to become a game studio. It's a habit of rapid experimentation with AI as a creative and product-building partner.

The five experiments

Every game is live. Open the build log on any card to see how it was made. This page works on any screen, but check each game's device badge before you play.

Works on:

Showing all 5 games.

Experiment 1

The Quiet Detective

Laptop

A pocket collection of ten short mysteries. Inspect the scene, solve two small puzzles, then name the culprit and explain their trick. No timer and hints on demand — a calm five-minute break.

Built with

  • AI-assisted build
  • HTML, CSS & JavaScript
  • Netlify

What it explores: interactive storytelling and puzzle structure across ten cases, plus a deliberate UX choice — low pressure and hint-friendly instead of a race.

Open the build log
  1. IdeaA mystery game that feels like a breather, not a test: short cases, no clock.
  2. Brief to AITen self-contained cases, each with a scene, the people present, two small puzzles and a final deduction: who did it, and what was their trick.
  3. BuildA case picker, an evidence board that fills in as puzzles are solved, a progress tracker, hints and a show-solution option.
  4. Test & refinePlayed cases through to check every clue actually leads to its answer. Kept to laptop, where the reading-heavy layout works best.
  5. ShipPublished as a live Netlify link.
Play Game → The Quiet Detective (opens in a new tab)
Experiment 2

Highway Stone Rush

Laptop + Phone

Switch lanes to dodge oncoming cars and collect stones. Up to three players take turns across Easy, Moderate and Tough rounds of 30, 45 or 60 seconds, with a shared results table at the end.

Built with

  • AI-assisted build
  • HTML, CSS & JavaScript
  • Keyboard + touch controls
  • Netlify

What it explores: core arcade mechanics, pass-and-play multiplayer, and designing for two input types at once.

Open the build log
  1. IdeaA dodge-and-collect lane game that friends can pass around one device.
  2. Brief to AIThree players, one turn each; difficulty set by round length (30/45/60 s); collect stones, avoid cars; final results table.
  3. BuildLane movement on Arrow keys or A/D for laptop, on-screen left/right buttons for phone, and a live scoreboard tracking stones, cars avoided and turns.
  4. Test & refineWent through a fix-and-retest pass on both laptop and phone before the live version went up.
  5. ShipPublished as a live Netlify link.
Play Game → Highway Stone Rush (opens in a new tab)
Experiment 3

Rune Cricket

Laptop + Phone

The notebook game of book cricket, rebuilt for two players. Five overs each; every ball draws a random number and its last digit decides the runs — land an 8 and you're out.

Built with

  • AI-assisted build
  • HTML, CSS & vanilla JavaScript
  • CSS animation
  • Netlify

What it explores: turning a familiar paper game into precise digital rules — and keeping innings, overs, targets and results readable at every moment.

Open the build log
  1. IdeaRecreate a game almost every Indian school kid played on the back page of a notebook.
  2. Brief to AIThe rules had to be spelled out, because a general AI won't guess them: five overs each, a random number per ball, last digit equals runs, 8 means out, second innings chases the target.
  3. BuildTwo scoreboards, a live match panel, a ball-by-ball log, an animated bat and ball, and a winner screen. No external libraries.
  4. Test & refineChecked the innings switch, the target chase and the match finish on laptop and phone.
  5. ShipPublished as a live Netlify link.
Play Game → Rune Cricket (opens in a new tab)
Experiment 4

Catch the Falling Stars

Laptop + Phone

Move the basket and catch as many stars as you can before three slip past. Pick a round from 15 to 60 seconds and a difficulty, then chase your best score on the high-score board.

Built with

  • ChatGPT
  • HTML, CSS & vanilla JavaScript
  • Browser storage
  • Netlify

What it explores: a tight replay loop — player-set difficulty and session length, a clear fail condition, and saved scores that give people a reason to try again.

Open the build log
  1. IdeaA simple catch game where the fun comes from beating your own score.
  2. Brief to AIWorked with ChatGPT on the rules: basket control, round lengths from 15 to 60 seconds, three difficulty levels, out after three misses.
  3. BuildTouch and pointer controls so it plays on phone, speed and spawn rate that scale with difficulty, and a high-score board saved in the browser so it survives a reload.
  4. Test & refinePlayed across levels and devices to keep score, misses, time and level in sync.
  5. ShipPublished as a live Netlify link.
Play Game → Catch the Falling Stars (opens in a new tab)
Experiment 5

Table Tennis Arena Pro

Laptop

A head-to-head table tennis match against a computer opponent. Steer your paddle with the arrow keys and choose Easy, Medium or Hard — first to five points wins.

Built with

  • ChatGPT
  • HTML, CSS & JavaScript
  • Netlify

What it explores: real-time movement and collision, a computer opponent tuned across three difficulty levels, and a short win condition that keeps matches snappy.

Open the build log
  1. IdeaA quick one-player sports match you can finish in a couple of minutes.
  2. Brief to AIWorked with ChatGPT on the rules: you versus the computer, arrow-key paddle, three difficulty levels, first to five.
  3. BuildBall and paddle movement, a computer opponent whose skill changes with difficulty, and a live scoreboard.
  4. Test & refineControl relies on arrow keys, so it's labelled laptop-only rather than claiming phone support.
  5. ShipPublished as a live Netlify link.
Play Game → Table Tennis Arena Pro (opens in a new tab)

The toolkit

A deliberately small stack: one AI partner, plain web technology, free hosting and real devices for testing.

ChatGPT

AI build partner

Turned written game rules into working logic, generated the first playable builds and handled debugging and fixes during iteration.

Used on: Catch the Falling Stars, Table Tennis Arena Pro

HTML, CSS & JavaScript

What the games are made of

Every game is a plain browser game: nothing to install, nothing to download, it just opens from a link.

Used on: all five games

CSS animation

Motion and feedback

Brings the bat, ball and match moments to life without any animation library.

Used on: Rune Cricket

Browser storage

Saving progress

Keeps the high-score board after the page is closed or reloaded, with no backend or accounts needed.

Used on: Catch the Falling Stars

Netlify

Publishing

Turns each finished build into a live public link that anyone can open and share.

Used on: all five games and this page

Laptop + phone

Real-device testing

Every game is played on actual devices. That's where the compatibility badges come from — they're tested, not assumed.

Used on: all five games

The process, step by step

Every game on this page followed the same five stages. For each one: what happens, where AI helps, and which decisions stay with me.

  1. Idea

    Pick a concept worth exploring — usually a format people already know, like book cricket or table tennis — and decide who it's for and how long a session should last.

    Where AI helps

    Brainstorming variations and checking whether a concept is small enough to build and play in one sitting.

    My part

    Choosing the idea and cutting scope hard, so the first version can actually be finished.

  2. AI assistance

    Write a clear brief: rules, controls, scoring, win and lose conditions, difficulty levels and target devices.

    Where AI helps

    Translating the written brief into game logic and a first working build.

    My part

    Supplying the rules an AI wouldn't guess — like "8 = out" or "no timer" — because vague prompts give generic games.

  3. Prototype

    Get a first playable version running in the browser as fast as possible.

    Where AI helps

    Generating the HTML, CSS and JavaScript, the layout, the scoreboards and the animations.

    My part

    Playing it immediately and judging the honest question: is this clear, and is it any fun?

  4. Iteration

    Play-test on laptop and phone, list what breaks or feels wrong, feed it back, test again.

    Where AI helps

    Debugging, targeted fixes and adjusting difficulty, speed and controls.

    My part

    Deciding what's a bug and what's a design choice — and labelling device support truthfully.

  5. Playable product

    Publish the finished build as a live link anyone can open and share.

    Where AI helps

    Last checks on the build before it goes out.

    My part

    Deploying on Netlify, setting the compatibility label and putting it in front of real players.

What these experiments demonstrate

Behaviours shown through the projects themselves — not formal qualifications.

AI-assisted prototyping

Getting from a rough concept to a working, published build without a traditional development team.

Seen in: all five games

Rapid experimentation

Shipping small, testing quickly and treating every build as something to learn from.

Seen in: five formats, all live

Product thinking

Deciding what the experience is for, then shaping rules, modes and win conditions around it.

Seen in: difficulty levels, round lengths, scoring

UX awareness

Designing for real people and real devices, from pressure-free play to touch controls.

Seen in: The Quiet Detective, Highway Stone Rush

Interactive storytelling

Structuring clues and reveals so a player can reason their way to the answer.

Seen in: The Quiet Detective

Clear AI instruction

Writing briefs precise enough that AI builds the right thing — including rules it could never guess.

Seen in: Rune Cricket

AI as a building partner, not just a content tool

My day-to-day work is growth marketing: SEO, paid media, lifecycle automation and reporting, with AI running through all of it. These games take the same habit somewhere less familiar — using AI across the whole path from a question to a finished thing people can use.

AI speeds up the exploring, drafting, building and fixing. The judgement about what to build, what feels right and what to cut stays with me. That combination — moving fast with AI while staying responsible for the result — is exactly what carries over into product and growth marketing.

  • Ideation
  • Exploration
  • Prototyping
  • Iteration
  • Execution

Explore the experiments

Play the Games →

Interested in AI-native experimentation, growth or product marketing? Let's talk.