How it works
Choose a research direction → collect evidence → identify problems → score opportunities → validate → generate a research brief.
1. Choose a research direction
You start a radar: a market, an audience, and the kinds of problems you want to notice. You can build it from ready-made choices in plain language, or write your own wording if you prefer. Nothing is guessed for you — the choices you make are the choices that are used.
2. Collect evidence
You can save evidence by hand: a complaint, a forum post, a review, a conversation note. You can also add public RSS or Atom feeds, and the workspace will collect new items from them. RSS and Atom are currently the only automated external source type. Everything you collect is stored as it was written.
3. Identify problems
Repeated complaints are grouped into one problem, so you can see how often something comes up and from how many independent sources. AI can suggest groupings, but it never creates, merges or splits a problem on its own — you approve every change.
4. Score opportunities
Each problem gets a transparent 0–100 opportunity score. Every component, its weight and where its value came from — measured evidence, your own judgement, or AI assistance — is shown. If the evidence changes, the score is marked out of date rather than silently rewritten.
5. Validate
Validation is deliberately manual. You record what supports the problem and what contradicts it, plan small experiments, and move a problem forward only with a reason. Contradictory evidence is kept and stays visible.
6. Generate a research brief
When you are ready, the workspace produces a research brief built only from the evidence you actually collected, with quoted excerpts checked against the original records.
What you should keep in mind
- You remain responsible for every research decision and conclusion.
- AI-assisted analysis can be wrong. It is labelled as derived, never as evidence.
- Contradictory evidence remains visible instead of being smoothed over.
- Automatic collection and automatic classification are optional and off by default.
- RSS and Atom feeds are the only automated external source type today.