Build your next deck with AI.
Describe the strategy you want to play. Yu-Gi-AI assembles a real, format-legal deck and refines it with you, card by card.
Studio preview
Watch the studio in action.
A short product walkthrough lives here. For closed beta, this is the simplest way to show the real flow: start from an idea, build a deck, refine the list, and approve concrete assistant edits inside the studio.
- Real deck generation from a plain-language strategy prompt
- Deck refinement loop with concrete, card-level reasoning
- Studio workflow, benchmark-backed model choices, and product direction
How it works
Talk to it. Watch it build. Push back.
A conversational loop built around real decks, not generic chat. You stay in control; the assistant does the grinding.
Describe
Tell the assistant what you want to play — an archetype, a combo line, or just a vibe. It reads your existing list as context.
Build
Yu-Gi-AI assembles a real, format-legal deck with a concrete engine, ratios, and side-deck picks you can try immediately.
Refine
Iterate on structure, swap packages, and stress-test lines. Every suggestion is reasoned, not pulled from a pile of staples.
Also on the blog
We measure the models we use.
When we say a model is the best fit for a job, we benchmark it against the alternatives on real decks and post the numbers.
Benchmarking frontier models on Yu-Gi-Oh deck building
Seven frontier LLMs on a real card pool, graded on strategist creativity, tactician deck quality, and real dollar cost. Vol. 1 of the Yu-Gi-AI Harness Benchmark.
Ready to build?
Open the studio, drop in a deck or an idea, and let the assistant take the first pass.
Open Studio