The 3rd cohort of Advanced Tech & AI Program is starting 20th September. You can learn more and apply to the program here
You can also check what PMs like about the program here
The future is already here, it's just unevenly distributed. In AI product management, this couldn't be more true. The question isn't whether product teams will need the AI skills - it's whether they acquire them before or after the competition does.
With this in mind, I conducted an 80-minute session of the topic ‘Complete Roadmap to AI Product Management in 2025’ last week. In this post, I share key themes covered and the link to the session recording.
Key Takeaways
• The bar for PMs has risen dramatically: 80% of AI PM is still traditional PM, but the bar in these traditional skills have risen up 10x. Discovery that took 3 months now needs to happen in 7-14 days.
• Engineers are ahead: As per Andrew Ng, Software engineers are embracing AI faster than PMs. The demand for AI-savvy PMs is moving up while traditional PM roles are declining.
• Building AI products beats understanding and learning about AI: Level 3+ proficiency (hands-on building) is now essential for AI PMs. You need to prototype and build AI products, not just understand concepts.
• AI adoption is mainstream: ChatGPT hit 700M weekly active users → Customer expectations have been shifting as more adopt new AI products.
Evaluate Where You Stand: The 5 Levels of AI Proficiency
Level 1: Heard the terms, can't explain them
Level 2: Can explain AI concepts, speak comfortably with engineers
Level 3: Can build prototypes and write advanced prompts
Level 4: Shipped AI features to production
Level 5: Building AI strategy for a product line
The UT-FRAME Method for AI Products
Learn and build AI products in this order:
Use case: Find biggest user frustrations AI can uniquely solve
Technology: Choose between traditional ML vs GenAI
Foundation: Test prompts and models with basic evals
Reasoning & Resources: Add RAG, fine-tuning if needed
Agentic AI: Evaluate if autonomous decision-making adds value
MVP: Get functional version to alpha users
Evaluation: Run extensive evals before production
Critical AI PM Skills for 2025
AI knowledge isn’t enough, here is an exhaustive list of areas to master:
AI for PMs:
AI prototyping with no-code tools
Deep research using advanced prompts
28+ prompt engineering techniques beyond basics
Finding the Right Use Case:
ROI calculation with API cost projections
Traditional AI vs GenAI decision framework
Edge case and hallucination management
AI Product Design Patterns:
New UI language (gradients, modern fonts, purple/blue themes)
UX patterns: Human-in-the-loop, graceful degradation
Trust and safety integration
AI Product Strategy:
RICE-AI framework (adding AI capability & stakes)
Build vs buy vs fine-tune decisions (99% don't need fine-tuning)
Creating moats through switching costs (memory, workflows)
AI Knowledge
Foundation models vs traditional models
RAG, reasoning, fine-tuning, etc.
Advanced prompt techniques
Agentic AI
Evals
Resources
Essential Reading - Tech and AI
Tech Simplified - System design for PMs (2-day read)
ByteByteGo Newsletter - Engineering concepts
Andrew Ng's courses on Coursera
Research papers
Product and design thinking
Thinking in Systems by Donella Meadows
Alchemy
Tools
Claude Pro ($20/month) - for serious PMs and builders as a thinking partner
Lovable/Cursor - AI prototyping
$5-10 in API credits for testing
Here is the session recording
This would be all!
Thanks,
Deepak