Skip to content
Tozca
§ case

Die Forward

A roguelite where defeated players' possessions become loot for the next player. Solo-shipped in about three weeks using AI as the production pipeline.

  • Game
  • · AI pipeline
  • · Solo ship
§ body

The problem

Can a single product leader ship a complete, playable game in three weeks by treating AI not as a feature but as the production pipeline itself?

The approach

Every non-creative step was AI: code generation, asset production, iteration on game feel, playtest summarization. The human job was the decisions — what the game is, what it feels like, what stays and what gets cut.

The game: a roguelite with a persistent economic loop between deaths. When you die, your gear becomes loot in another player’s run. Your run is fed by someone else’s failure.

Outcome

AI-as-pipeline changes what a single person can carry from idea to shipped. It doesn’t replace taste, product sense, or the decision of what to make. It does replace most of the time between those decisions and the result.

§ note

Have an AI question worth thinking through?

Tell us the shape of the problem. Thirty minutes is usually enough to clarify the next move.