How a task runs
- Click the task on your home page or in the sidebar.
- Have the conversation. The AI runs an interview tailored to that task’s goal. Answer honestly. Skim answers = skim document, that’s just how it is.
- Research phase (sometimes). Some tasks fire off background research while you’re talking. You’ll see chunks streaming in. Don’t panic, it’s working.
- Document generation. When you’re done, the app produces a polished document.
- Done. The document is saved. The task is marked complete. You can re-read it any time, edit it later, the works.
Prerequisites
Some tasks are locked behind others. If task B says “Complete Brand Story first”, that’s because B literally cannot do its job without the document A produces. We’re not gatekeeping for the bit. The AI needs the upstream answer to do downstream work well, full stop. Trust the locks. They’re for you. The home page shows a lock icon on tasks that aren’t ready yet. Finish the prereqs and they unlock automatically. No refresh needed.Can I redo a task?
Yes. Open the task and hit Restart. Heads up: this nukes the existing document for that task. Anything downstream that referenced it will use the new version going forward. If you just want to tweak something rather than redo the whole interview, edit the generated document directly in your Library. Same outcome, way less effort.Where the document goes
Two places, simultaneously:- Library tab: your full document library. Searchable. Groupable. Downloadable.
- Auto-injected into projects: when a project lists a task as a prerequisite, that task’s generated document gets fed to the AI as context every time you chat in that project. You don’t have to attach it manually. It just shows up. That’s the magic part.