Yesterday I attended Marty’s talk: “New Standard of PM in the Age of AI” hosted by Udacity.

As Steve Messer put it, “his views on AI were suprisingly comforting”, I would add reasonable as well. Marty emphasized on the three different product management job descriptions (PO, Feature Team, Empowered Team). He skipped fueling the hype about tools and models, or bloated expectations about 10x everything. So nice! I had similar takeaways after attempting the Product Management Festival 2025 last year. Here are my notes:

What kind of PM faces challenges due to AI?

  1. Traditional Product Owners: Pure backlog management is being/has been automated.
  2. Feature Team PMs (Project Model): A focus solely on output and command-and-control can be handled more cost-effectively by agents now.

Where does the future lie? – Where it has always been, according to Marty:

  • Empowered team PMs (“Solution Builders”):
    • Focus on outcomes and “business” results.
    • Solving puzzles (Discovery), followed by coding (Engineering) - Coding is faster than ever today thanks to AI.
  • Double & Triple Threats:
    • Those who combine skills from 2–3 disciplines (Product Sense + Design Sense + Architecture/Data Sense) remain valuable.
    • But it’s about the “senses,” not the “skills.”

A colleague asked about the difference between skills and senses. Here is my butchered explanation, Marty does it better in his book “Inspired”:

  • PM view:
    • Skill = How to build (SCRUM etc.)
    • Sense = What to build (Value / Strategy / Trade-Offs)
  • Engineering view:
    • Skill = Fluent in a coding language, “How to build this efficient”
    • Sense = Knowledge and reflection on different technical patterns, architecture and system approaches (What tradeoffs to make)

What does all this mean for our daily work, now?

  • Discovery and “build to learn” are more important than ever: Thanks to AI, prototyping is now almost “free.”
  • The hard work for us is deciding what’s worth actually building
  • Don’t outsource thinking: AI makes it easy to work through processes, but it must not be used as a substitute for our own thinking (trade-offs, decisions)!

Marty’s To-Do List for AI Time: