Raw thoughts go in. The model interviews you about the gaps, then hands back an execution plan, a budget, and a task list.
About the project
Ideas do not survive contact with criticism, including mine. Most of them get shot down before they have enough shape to defend themselves. I built a place to hold them until they do.
Raw thoughts go in and get tagged. When one shows promise, you promote it to a brainstorm, and the model runs an interview to pull out the details you have not thought about yet. From there it compiles an execution strategy, an expense tracker, and a task list. I prototyped the workflow in Notion first, then rebuilt it on Lovable with a Supabase database. Both meant a monthly subscription for something I use a few times a week, so I stripped the cloud database out and wired the front end straight into Google Sheets and Drive, with a data layer that handles offline edits.
So far I have only run it backwards. I fed it several finished projects, LithoLab among them, to see whether the plans it produced matched what I actually did. They held up well enough that I trust it. I have not yet taken a brand-new idea through it from start to finish. That is the real test.
- React
- TypeScript
- LLM Integration
- Google Drive API