Basically, the tool reframes conversations and tries to find the smallest moves the user COULD have made to have surfaced unknown assumptions for whatever problem they were solving.
The goal is to show the value of using AI as a reflective surface to search for "unknown unknowns" or "things YOU didn't know you needed to know".
-------------------------
The basic premise is this:
For any given AI conversation, the quality of the output (for the specified goal) depends on the context from which decisions can be made.
You can learn to surface assumptions (the AI's AND your own) and get better output from informed decisions. Best of all, when operating in this way, you have to actively engage the problem and strategy to accomplish this, so you don't lose understanding or creative agency.
When you iterate during problem definition or strategy planning, it's cheaper and more efficient, too.
For example:
[# of Planning Phase Iterations] x 1,000 tokens [# of Strategy Phase Iterations] x 3,000 tokens [# of Execution Output Phase Iterations] x 5,000,000 tokens
Usually when we are solving problems with this method, we one-shot the output even with local models like Qwen 3.6/3.8, spending FAR less time in the expensive execution phase.
---------------------------
Anyway, please have a look! Totally free, no signup, you just need to export conversations in specific formats and upload them.
This was made by myself and a few interns in a barn while doing indie game development ... and costs us money, so please don't abuse it. <3