LinkedIn

AI is making me a better product leader.

I’ve built a team of AI agents that takes the volume work off my desk. My own time goes to the calls only a person should make: what’s worth doing, and what’s worth fixing.

Me and my agentic team

Tjeerd

Sets the direction and makes the calls

TjeerdSets the direction and makes the calls

    • Does

      Follows markets, competitors and news in my field.

      Hands me

      Short briefs with sources.

      I decide

      What it means for us.

  • Ozzy

    Chief of staff

    Does
    Reads my calendar, mail and tasks every morning.
    Hands me
    A morning brief with my priorities and what’s waiting on me.
    I decide
    What I spend my day on.
  • Buzz

    Market analyst

    Does
    Follows markets, competitors and news in my field.
    Hands me
    Short briefs with sources.
    I decide
    What it means for us.
  • Wizz

    Audience expert

    Does
    Knows the target customer inside out.
    Hands me
    What users say in feedback and reviews, and the questions worth testing.
    I decide
    Who we build for.
  • Strategy Board

    Multiple agents

    Does
    A round table of agents that stress-⁠test a plan and argue the other side.
    Hands me
    Counter-⁠arguments and options I hadn’t considered.
    I decide
    The decision.
  • Bizz

    Business case analyst

    Does
    Builds business cases and checks the numbers.
    Hands me
    A first model and the assumptions to challenge.
    I decide
    Whether the case holds.
  • Jozz

    Storyteller

    Does
    Turns an argument into a narrative for a deck or a talk.
    Hands me
    A storyline and a first draft of the deck.
    I decide
    The argument.
  • Razz

    Marketing team

    Does
    Positioning and launch messaging: the work of a small agency.
    Hands me
    Positioning options and first drafts.
    I decide
    Which story we tell, and to whom.
  • Zazz

    Designer

    Does
    Brand identity, and the look of my playbooks and decks.
    Hands me
    Designs ready for review.
    I decide
    What looks and feels right.
  • Gizz

    Engineering partner

    Does
    Turns product intent into clear problems and trade-⁠offs engineers can work with.
    Hands me
    A problem statement and the trade-⁠offs to discuss.
    I decide
    What we build first.

Underneath it all

Second brain

One knowledge system with my context and past decisions, so every agent starts from the same understanding.

Background work

Agents keep researching and drafting while I’m in meetings, the way a good team keeps moving when its manager is busy.

How I run them

  • Every agent starts from a written brief with my context and principles, the way a new hire starts from a handbook.
  • Facts are grounded or cut. Every claim needs a source, and “I couldn’t confirm it” is a good answer.
  • Agents draft, I decide. Nothing goes out unreviewed.
  • At an employer, company data stays in the company’s approved tools. What I bring is the way of working.
  • I pick the best AI model for each job and can switch without rebuilding the team.
  • Agents that don’t earn their place get retired, like a feature nobody uses.

What a brief looks like

Buzz · Market analyst

brief.md
Role
A personal market analyst with the precision of a senior strategist, helping me master my market: the players, the dynamics, the trends and the opportunities.
Expertise
Dissecting competitors, spotting trends before they go mainstream, finding white spaces nobody serves, and assessing who could disrupt the market.
Way of working
Clarifying questions first, insight over lists of data, what it means for me specifically, brutal honesty, and speculation clearly labelled as speculation.
Fixed formats
A competitor analysis ends with a positioning map, the white spaces and a threat level per competitor. A trend analysis gives every trend a time horizon, an impact rating and the “so what”.
Context
My product, my market and what sets us apart.

A typical day

  1. 07:30

    The team: Ozzy’s morning brief: my priorities and what’s waiting on me.

    My call: What gets my attention first.

  2. 08:30

    The team: Buzz flags a competitor’s launch from overnight, with sources.

    My call: Whether it changes anything for us. Usually it doesn’t.

  3. 10:00

    The team: I write a one-⁠page plan, and the Strategy Board attacks it from the customer’s side and the CFO’s.

    My call: Which objections are real, and what I change.

  4. 12:00

    The team: While I’m in meetings, the team keeps researching and drafting.

    My call: Nothing yet. That’s the point.

  5. 14:00

    The team: Jozz turns the revised plan into a storyline for the leadership meeting, and Zazz gives the deck its look.

    My call: The argument, and the order it’s told in.

  6. 16:00

    The team: Razz drafts three ways to position the launch.

    My call: Which one we go with.

  7. 18:00

    The team: Ozzy closes the day: what moved, and what’s open for tomorrow.

    My call: When to stop.

In practice at Decathlon

Customer feedback is scattered, and shaping it into something a team can act on takes hours nobody has. I saw a place for AI to take that work.

In the second half of 2025 I started a programme at Decathlon to build AI fluency across my product and engineering teams. The aim was to give them back the hours that go into gathering and sorting information, so they could spend those hours on what to build and why.

We started with research. A daily brief now pulls customer feedback from app stores, surveys and other channels and summarises it from a development perspective. Developers start with what customers said. Product managers see problems sooner and start exploring solutions earlier.

Find the work that eats your team’s attention and give AI a role in it. Then build the team’s ability to use AI well.

What I believe

Execution is cheap now. A team can build almost anything, because the old limit, too few hands and too few hours, is gone. Removing that limit is also a trap: when everything is possible, nothing forces a choice, and products fill up with features nobody asked for.

Taste has to do the job that scarcity used to do. Taste is made of understanding people: how their days unfold, where the hassles are, and what relief and delight feel like. Code was always in service of that understanding. AI thins the layer between an idea and the product, which puts teams closer to the people they build for. Without understanding people, cheap execution ships more of what nobody needed.

There is a part of the work I don’t hand to an agent. An agent can condense a year of app reviews into one page. It has never had a bad day in the middle of a hassle. Judging which hassle matters takes time spent with real people, and no brief replaces that time.

I bet on curation. A product with a clear point of view gains more over the years than one that does everything for everyone, and the products people keep coming back to sit where usefulness and enjoyment meet.

Read my principles