How do you run discovery for a new software project today?

I'm launching Discovery Phase AI here on October 9 and would love to hear how others handle the discovery phase before a build starts.

In my experience it looks like this: a kickoff call, a few decks and emails, then days of turning them into stakeholders, goals, requirements, user flows and a rough roadmap, usually spread across docs and slides. And when one decision changes, the other documents quietly go out of date.

That is the problem I built Discovery Phase AI for. You upload what you already have (a call transcript, a deck, notes), the AI drafts ten linked discovery stages, you review every draft before it is saved, and you export one report as PDF, Word, PowerPoint or Markdown.

A few questions for you:

1. What does your discovery output look like today: one document, a Notion space, slides, something else?

2. Which part takes the longest: gathering input, writing it up, or keeping it consistent as things change?

3. Would you trust an AI first draft of requirements, or only of the background and summary parts?

If you want to see what the output looks like without signing up, here is a full example report for an invented project: https://discoveryphaseai.com/example-report?utm_source=producthunt&utm_medium=referral&utm_campaign=launch

Honest feedback, including "I would never use this because...", is very welcome.