Go from a vague founder-ask to a testable goal

Every founder has a list of "make X feel better" asks. They're real signals — the founder's pattern recognition is usually right — but they're not specs. This use case is the bridge: from a one-line vibe to a goal that the team can actually build against.

The input

One sentence. Examples from real customers:

• "Make onboarding feel less janky"
• "Power users should get to value faster"
• "The empty state on the dashboard is depressing"
• "We need a better story for teams of 50+"

That's it. No supporting evidence required at the start — gathering it is part of the flow.

The flow

1. Create a new intent. Paste the one-liner into the conversation panel as your opening message.

2. Let the AI push back. Pathmode's Socratic personality challenges vagueness one question at a time:

• "What does 'janky' mean specifically? Slow? Confusing? Broken?"
• "Whose onboarding? First user in a workspace, or the third teammate they invite?"
• "How will you know it's better? What's the leading indicator?"

3. Answer in your own words. Each answer crystallizes another piece of the spec. The right panel updates in real time — you watch the spec form sentence by sentence.

4. Anchor as you go. When you mention specific friction ("the email verification step times out for some users"), the AI suggests creating an evidence item. This grounds the spec in something concrete instead of staying at vibes-level.

5. Stop when the ThinkingQuality score looks healthy. Watch the four dimensions — specificity, testability, coherence, tension. Until testability and tension are both above zero, the spec is still a vibe in disguise.

The output

From a one-liner you get:

• A specific goal — "Reduce time-to-first-value for new users from 14 minutes to under 5 minutes"
• 3–5 outcomes anchored to the friction you described
• Verification criteria — what metrics or events tell you it worked
• Open questions flagged for follow-up research
• Out-of-scope notes — what you explicitly chose not to address yet

That's something the engineering team can build against. It's also something you can defend to your board, your investors, or future-you.

Why this beats "founder writes the PRD"

Founders write fast and skip the parts they think are obvious. The team then spends two weeks reverse-engineering the obvious parts, often wrong. The Socratic dialogue makes the obvious parts explicit — without slowing the founder's velocity, because they're just answering questions.

Try it yourself

1. Pick a one-liner you've been meaning to "do something about"
2. Open Pathmode → product → New intent
3. Paste the one-liner
4. Answer the AI's questions, one at a time
5. Stop when the ThinkingQuality score looks healthy across all four dimensions

Related

• Use case: Audit a draft spec for missing edge cases
• Playbook: Your First Spec in 10 Minutes
• Use case: Anchor every outcome to user evidence