Lesson 5 of 6
Narrow the intent
7 min read
You ship a shiny chat box: "Ask me anything." Users type three words, get a mediocre answer, and never come back. What went wrong?
The blank-box trap
A blank chat box hands the user the hardest part — figuring out what to ask — and hands the model no guardrails. So it does everything a little and nothing well. Narrowing the intent — from an open box to a specific, guided task — is the single biggest lever on how reliable and useful the feature feels.
Reliability isn't only the model — it's the scope. The narrower the job, the more of it the model can actually nail.
Scope a job, not a genie
Give the feature one clear job. Replace "ask me anything" with suggested prompts, then guided fields, then a single-purpose task — "summarize this ticket," "draft a reply in this tone." Each turn of the screw removes ambiguity, and reliability climbs with it. You can always widen later; you rarely recover a bad first impression.
A narrow scope is also evaluable: a single clear job has a right answer you can test against. "Ask me anything" has no test set — and no way to know if it's getting better.
The shape of it
- —A blank box asks the user to do the hard part and gives the model no guardrails.
- —Narrowing intent — prompts, fields, one clear task — raises reliability.
- —A narrow job is testable; "ask me anything" can't be measured or improved.
Your open-ended AI box gets lukewarm use. What's the highest-leverage fix?
Continue in the app
Take the whole Discovering & Scoping AI Opportunities course — tracked
Get your personalized path, progress and streaks in the app — this lesson and every next one, in order.