Choreodle: How My Lean AI Team Built a Wordle for Ballroom and Latin Choreography
It is incredibly rewarding when a side project starts gaining organic traction, especially when you haven’t spent a single dime on marketing. That is exactly what has been happening quietly with my latest build: Choreodle.
If you are a dancer, you know that rhythm, flow, and technical sequencing are everything. Choreodle takes that premise and turns it into a daily puzzle. Think of it as Wordle, but instead of guessing a five-letter word, you are piecing together a five-figure sequence of ballroom or Latin choreography.
Here is a look at how this project came to life, bridging the gap between my love for dance and my obsession with modern, AI-driven software development.
Translating Dance to Code: The Mechanics
The core loop of Choreodle is simple to grasp but deceptively tricky to master, relying entirely on a dancer’s understanding of how figures naturally connect.
Players are presented with five empty slots (numbered exactly 1 through 5) and a pool of figures for a specific dance style. Let’s take a promotional run of a Waltz puzzle as an example, which I recently recorded specifically to demonstrate the game’s mechanics and that classic puzzle-game frustration.
To show what happens when things go wrong, I threw out an initial wild guess: Spin Turn, Progressive Chasse, Backward Lock, Outside Change, Hesitation Change. Total miss. The board returned all gray tiles.
As the promotional playthrough continues into the third and fourth attempts, the board shows the struggle of scrambling the order before the logic finally clicks. The game provides familiar visual feedback to guide the player:
- Green: The figure is correct and locked into the perfect slot in the routine.
- Orange: The figure belongs in the sequence, but it is in the wrong position.
- Gray: The figure isn’t part of today’s choreography at all. Eventually, the board resolves into a classic, flowing Waltz sequence that any seasoned dancer could lead or follow in their sleep: 1. Closed Change -> 2. Natural Turn -> 3. Reverse Turn -> 4. Whisk -> 5. Chasse from PP.
Building the logic to handle these sequence validations, track attempts, and calculate the unused-round bonus score was a fun challenge—and one I didn’t tackle entirely alone.
The Build: Eating My Own Dog Food
In a previous post, I outlined my framework for a Lean AI-Powered SDLC with a Team of Agents. Choreodle was the perfect opportunity to put that theory into rigorous practice.
Instead of hand-coding every piece of the game’s logic and UI, I leaned heavily on my AI setup.
Having an AI team handle the boilerplate and rapid prototyping allowed me to focus on the domain expertise—mapping out accurate, technically sound choreography riddles for styles ranging from Samba to Quickstep.
What’s Next?
Watching the daily user count tick up organically is a great validation of the concept.
Building Choreodle has been a fantastic reminder that when you combine a personal passion with the right tech stack—and a capable team of AI agents—you can ship products that truly resonate with a niche community.
