Manual

How I work

I've shared versions of this document with every team I've led. This is the public edition — for anyone considering working with me in any capacity.

Two modes

I operate in two registers

I operate in two modes, and knowing which one I'm in is the key to working with me.

The operator

Fast, decisive, direct, pushing toward outcomes. This is the muscle I built across a decade of scaling products in high-growth, high-pressure environments. When the problem is messy and the stakes are real, I drive toward clarity and make calls.

The coach

I'm ICF-certified, and that training fundamentally changed how I lead. I learned that the most impactful thing a leader can do is sometimes not solve the problem — it's to ask the question that helps someone else find their own answer.

Knowing when to drive and when to hold space is the hardest judgment call in leadership. I get it wrong sometimes. But I'm always choosing deliberately.

Principles

What I believe

Think big, start small, move fast

Momentum teaches you more than planning. But speed without direction is chaos — so the "think big" part comes first for a reason.

Solve the problem, not the metric

Why are we doing this, and will it contribute to the outcome we actually want? If the dashboard looks good but the user problem is still there, we've failed.

The best idea wins

Not the most senior person's idea. Not the loudest voice. Bring clear reasoning and I'll change my mind. I expect the same willingness from everyone.

Owners, not renters

I set context, remove obstacles, and hold you to a high bar. You own the work. Propose ideas, challenge decisions, and don't wait for permission to do the right thing.

Always write a memo

Verbal agreement is fragile. A written argument forces clarity, creates a record, and respects other people's time. If you want to convince me of something, write it down.

Strengths

What I bring

Operator mode

I cut through ambiguity, make decisions when others are stuck, and hold a high bar. I'm at my best when someone needs to create structure out of noise.

Coach mode

I listen. I ask the question you've been avoiding. I won't hand you the answer — I'll help you find it, because that's how the insight sticks. The goal is your growth, not your dependence on me.

Both modes

I'm direct. I say what I think. I genuinely care about the people I work with doing their best work.

Growth edges

What I'm still working on

My pace isn't always comfortable. When things don't move at the speed I expect, my directness can feel like pressure. I work on this — not by becoming less direct, but by being more intentional about timing and delivery.

I hold strong opinions. I'm genuinely open to changing them, but you need to bring reasoning, not just pushback. Data, logic, or a perspective I hadn't considered — those work.

Stalling without a clear reason frustrates me. The best way to keep me engaged is to show me the bigger picture is on track, even if today's pace feels slow.

Collaboration

How to work with me

Bring your thinking, not just your questions

Come with a point of view and want to pressure-test it. The fastest way to earn my trust is to show me you've thought hard about the problem.

Be direct

I don't decode subtext well, and I don't expect you to decode mine. If something isn't working, say so.

Close the loop

I expect follow-through — in action and in communication. The thing I find hardest to work with is silence where there should be a signal.

Conviction

Building in the AI era

I build products where last quarter's best practice is already obsolete. That kind of uncertainty is different from normal ambiguity — the ground itself is shifting faster than your roadmap can account for.

Hold your conviction on the problem loosely. Hold your conviction on the user tightly. The technology will keep changing. The model will get better. But the human need you're solving for is stable. Build your judgment around that anchor.

I co-authored three academic papers on AI safety — not to be an academic, but because building AI products in high-stakes domains forced rigorous thinking about questions most product leaders hand-wave. When should an AI intervene? When should it defer to a human? How do you design trust into a probabilistic system? These aren't theoretical for me. They're shipping decisions.

Signals

How to read me

Energised

Generating ideas faster than anyone can write them down, pulling people into conversations, compressing timelines.

Stressed

More task-focused, shorter in communication, narrowly locked onto what needs to happen next. If you notice this, help me prioritise — don't add more.

Disappointed

Quieter than usual. More pointed in feedback. If I seem to disengage rather than push back, something needs to be addressed directly.