Spellbook

I find it relatively natural to take an ambiguous idea, turn it around from different angles, ask myself a long sequence of questions, and eventually arrive at something coherent. A lot of people have good ideas but find that process difficult, especially when the final result has to become a serious specification, strategy, plan, or document rather than just an interesting conversation with an AI.

Spellbook is my attempt to turn that way of thinking into a product. Instead of asking one enormous question and getting one enormous generated answer, it takes you through a workflow. It asks the next useful question at the right point, exposes things you have not considered, researches what needs to be researched, and makes you resolve contradictions before they quietly become assumptions.

The important thing is that Spellbook should never replace the person's idea with the model's idea. The decisions remain yours. The product gives you a process for exploring the problem thoroughly enough that you understand why you are choosing one direction over another. By the end, you should know what you want to build, how it should work, what alternatives you rejected, and what questions still remain.

The output can then become whatever useful document the work requires: a product specification, a strategy, a research report, a business plan, or something else entirely. Over time, I also want Spellbook to preserve the context behind those decisions so future work does not constantly begin from zero. The goal is simply to make meticulous, first-principles thinking easier to do.