AI PM Lab

Goal: Build a AI PM product in small, shippable slices.

Current MVP Idea

Insight2Spec — Convert raw user feedback (tickets, reviews, support chats) into:

1. clustered themes
2. opportunity statements
3. draft PRD sections
4. experiment ideas

Why this product

• Extremely PM-relevant (discovery, prioritization, PRD quality)
• Strong LLM fit (summarization, clustering, synthesis)
• Teaches practical AI product engineering (prompting, evals, latency/cost/quality tradeoffs)

MVP Scope (v0)

• Upload/copy feedback text
• Generate top themes + confidence
• Generate 3 opportunity statements
• Generate draft PRD outline
• Save run history as JSON

Stack

• Python 3.12
• FastAPI backend
• OpenRouter for model calls
• Simple local JSON storage for now

Learning outcomes for Asif

• Prompt design and iteration
• Structured output validation
• Failure handling and guardrails
• Evaluation loops for quality
• Shipping incrementally with measurable outcomes

Run

pip install -r requirements.txt
uvicorn main:app --reload