nigiri
A new concept for working with ideas and thoughts in the context of generative AI
Most AI tools present conversation as a linear sequence. One prompt, one response, then forward. That structure works for questions with a single answer; it breaks down when you’re trying to develop an idea across multiple branches.
Nigiri organizes prompts as a tree. Each node holds text — written manually or generated by the model. A change to any node triggers regeneration of everything downstream from it. Workflows can be saved: a defined chain of prompts for a specific kind of reasoning can be reused with different input, producing comparable output each time. The 2D canvas makes it possible to see multiple branches at once without losing track of how they relate.
It was built as a prototype and remains one. There are no plans to turn it into a product.