Turn a design rationale into outcomes engineers can verify

A mockup shows what. The rationale — why this flow, why this order, why we removed the step — is why, and it never makes it into the file. So when an agent or an engineer builds from the mockup, they optimize for what's easy to code, and your deliberate choices get quietly undone. Authoring the rationale as outcomes makes the intent durable: the reasoning ships alongside the pixels.

The input

The thinking behind a design you've already explored:

• The decision ("we collapsed signup into one screen")
• The reason ("every extra screen dropped completion ~10% in the last test")
• The constraint you're respecting ("must still work on a 320px viewport")

A figma link or screenshot helps, but the words matter more.

The flow

1. Open a new intent and state the decision. Paste the rationale into the conversation as your opening message — decision and reasoning together.

2. Let the Socratic AI separate decision from preference. It pushes on the soft spots: "Is one screen the goal, or is completion rate the goal and one screen the means?" That distinction is the difference between an outcome an engineer can verify and a layout they'll feel free to change.

3. Anchor the reasoning to evidence. When you cite the test result, the AI offers to attach it as an evidence item. Now the choice is backed by data, not by taste alone — and it'll survive someone asking "why not two screens?"

4. Land outcomes with verification criteria. "Signup completion stays above the current baseline at 320px width" is an outcome a build can be checked against. The intent of the design is now executable, not just illustrated.

The output

A spec that protects the design decisions instead of leaving them to chance:

• Outcomes, not layouts — what must stay true, expressed so it can be verified
• Rationale anchored to evidence, so choices are defensible months later
• Constraints written down ("works at 320px"), so they don't get dropped in build
• Taste turned into judgment — a decision the team can act on, not just admire

Why this beats handing over a mockup alone

A mockup with no rationale is a suggestion. Anyone downstream — agent or engineer — is free to reinterpret it, and they will, toward whatever's convenient. Authoring the why as verifiable outcomes makes the design intent load-bearing: the build is measured against your reasoning, not against a guess at it.

Try it yourself

1. Pick a design decision you've had to defend more than once
2. Open Pathmode → product → New intent and paste the decision + reasoning
3. Answer the AI until each choice is an outcome you could verify
4. Anchor the strongest reason to an evidence item

Related

• Use case: Anchor every outcome to user evidence
• Use case: Catch the experience gaps in a spec before build
• Playbook: The Six-Part IntentSpec