AI Feature Definition Prompt
A structured prompt for defining AI-powered features in product specs — specifying the model behavior, confidence thresholds, fallback logic, and human-in-the-loop checkpoints required.
Syntax
product-owner-ai
Write a product spec for this AI feature:
Feature: [describe what the AI should do]
Include:
1. Input: What data the model receives
2. Output: What the model produces
3. Confidence threshold: When to show the result vs. ask for human review
4. Fallback: What happens when confidence is too low or model fails
5. Human-in-the-loop: What decisions require human approval
6. Feedback loop: How the model improves over time
7. Acceptance criteria for launchExample
product-owner-ai
Feature: AI-suggested invoice due date
Input: Client payment history, invoice amount, client tier
Output: Recommended due date with confidence score
Threshold: Show suggestion if confidence ≥ 80%; prompt user to choose if <80%
Fallback: Default to 30 days if model unavailable or confidence <50%
Human-in-the-loop: User can override suggestion; override is logged as training signal
Feedback loop: Actual payment dates vs. suggested dates fed back monthly
Launch criteria: Suggestion accepted by user ≥ 60% of the time in A/B test