Overview
Not every product needs AI. But every product team should know where AI would create genuine value vs. where it's a distraction. This skill helps you find real AI opportunities — not "add a chatbot" but "here's where AI fundamentally improves the user experience."
Before You Start
Ask the user:
- What's the product? — What it does, who uses it, core workflows.
- Current pain points — Where do users struggle, wait, or do repetitive work?
- Data available — What user data, content, or behavioral data exists?
- Team capability — Do you have ML engineers, or would this require hiring/buying?
- Strategic context — Is AI a strategic priority, or exploring early?
Opportunity Identification Framework
Step 1: Map User Workflows
For each core workflow, identify where users:
| Workflow Step | User Action | Time Spent | Friction | Data Available? | |-------------|-------------|-----------|----------|----------------| | [step] | [what they do] | [minutes] | [High/Med/Low] | [Yes/No — what data] |
Step 2: AI Opportunity Filter
For each high-friction step, evaluate AI fit:
Good AI candidates:
- Tasks that are repetitive but require judgment (classification, prioritization, summarization)
- Processes with lots of data but little insight (pattern detection, anomaly detection)
- Personalization at scale (recommendations, content curation)
- Content generation with human review (drafts, suggestions, completions)
- Prediction based on historical patterns (forecasting, risk scoring)
Bad AI candidates (Nate B Jones principle: don't force it):
- Tasks requiring perfect accuracy with no tolerance for errors
- Workflows where users need to trust and verify every output
- Problems that are simpler to solve with traditional software
- Features where AI adds complexity without proportional value
- Areas with insufficient or biased training data
Step 3: Evaluate Each Opportunity
| Opportunity | AI Approach | User Value | Data Ready? | Build/Buy | Effort | Risk | |------------|-----------|-----------|------------|-----------|--------|------| | [opportunity] | [technique] | [benefit] | [Y/N] | [build/buy/partner] | [S/M/L] | [H/M/L] |
Step 4: Value Assessment
For each opportunity, answer:
User value: Does this save time, reduce errors, or enable something previously impossible? Business value: Does this increase revenue, reduce cost, or create competitive moat? Feasibility: Can we build this with available data, team, and infrastructure? Risk: What happens when the AI is wrong? How bad is a false positive/negative? Trust: Will users trust AI in this context? How do we build that trust?
Step 5: Implementation Strategy
For top opportunities:
Start with augmentation, not automation. AI suggests, human decides. This builds trust and generates training data for better models.
Build the feedback loop. Every AI feature needs a way for users to correct it. These corrections become training data that makes the model better over time.
Plan for failure gracefully. What's the user experience when the AI is wrong? If there's no good fallback, the feature isn't ready.
Output
# AI Opportunity Assessment — [Product Name]
## Executive Summary
[2-3 sentences: top opportunities, strategic recommendation, readiness assessment]
## Opportunity Map
| Priority | Opportunity | Value | Feasibility | Recommended Approach |
|----------|-----------|-------|-------------|---------------------|
| 1 | [opportunity] | High | High | [approach] |
| 2 | [opportunity] | High | Medium | [approach] |
| 3 | [opportunity] | Medium | High | [approach] |
## Detailed Analysis
[Per-opportunity deep dive]
## What NOT to Build with AI
[Opportunities that were evaluated and rejected, with reasoning]
## Prerequisites
- Data requirements: [what's needed]
- Team requirements: [hire/upskill/partner]
- Infrastructure: [what's needed]
## Recommended Roadmap
- **Phase 1 (0-3 months):** [Quick win AI feature — augmentation mode]
- **Phase 2 (3-6 months):** [Higher complexity — with feedback loop data from Phase 1]
- **Phase 3 (6-12 months):** [Full automation where trust has been established]
Save as AI-OPPORTUNITIES-[product-name].md.