HOW I WORK

I take three or four engagements a quarter. Here's what each one looks like.

I work with product leaders at Series B → pre-IPO companies who need an operator in the room, not a deck in the inbox.

The work

Who I work with.

If one of these sounds like where you are right now, it's usually a good place to start the conversation.

If you're leading AI inside a company

You're the person on the hook for AI working. Pilots have shipped, the next bet is bigger, and you'd rather have someone in your corner who's actually shipped the kind of thing you're being asked to ship — than another advisor who's only ever advised on it. Pilot purgatory isn't a strategy failure. It's an execution gap.

What we'd work on

Roadmap pressure-testing, eval and cost-model work, embedded review on the calls that matter most.

If you're a founder building an AI product

You're somewhere in the run-up — pre-launch, pre-raise, or mid-pivot. The product has to land and the runway is finite. Token economics will surprise you at production volume if you haven't modeled them. You'd rather find the wrong assumption before you spend a quarter building on it than after.

What we'd work on

Strategy pressure-test, architecture review, the calls that de-risk the build.

If you're an investor working with portfolio companies

You need an honest read on where AI capability is real and where it isn't — across one company or across a portfolio. Fast enough to fold into the value creation plan (VCP), deep enough to act on.

What we'd work on

Diligence reads, gap-to-value reports, and a value creation plan built on the actual AI capability in the room — not the pitch.

None of these quite fit? Tell me what's going on anyway — I'll be honest about whether I'm the right person for it.

Three shapes

Pick the shape that fits the work.

When boards demand AI outcomes on a quarterly basis, the full-time Chief AI Officer hire creates a timing problem: 90-day onboarding, a six-month strategy phase, then the Q4 board deck. The companies shipping meaningful AI this year mostly have one thing in common — a practitioner who could start building in week one. These three engagement shapes are designed for that. Each one starts with the problem, not the discovery phase.

Advisory retainer

Monthly · 3-month minimum

You have an internal AI lead. They're good. They're alone. You need someone in their corner who's done this at scale, on call for the calls that matter.

  • Weekly working session with the AI lead
  • Async architecture and review on demand
  • Standing seat in the leadership AI sync

Embedded sprint

Fixed-scope · 90 days · Quarterly slot

You have a specific bet — a product launch, a cost crisis, an eval rebuild — and you need it shipped, not advised on. I embed, I take operating accountability, I leave when it's running.

  • Diagnose → prioritize → ship inside 90 days
  • Code, evals, and architecture decisions, not just recommendations
  • Production-ready handoff with the team that owns it next

Workshop & training

Two-day on-site · One team at a time

Your engineers are smart and stuck in the same loop — reinventing the harness, the eval framework, the cost model. They need a reset from someone who's already shipped the stack they're building.

  • Reference architecture for the agentic harness
  • Eval and cost-model patterns they can take into the next sprint
  • Decision log so leadership knows what the team agreed to

How engagements work

Three things that are true of every engagement.

These aren't caveats — they're the model. Each one exists because the alternative produces worse outcomes for the people doing the work.

Engagements end with running systems, not slide decks.

The exit condition is a working system in your environment — with the operating principles documented so the team that owns it next can maintain it. Deliverable documents are a byproduct, not the product.

Every engagement is scoped to an outcome.

Hourly advisory turns into a meeting habit with no clear finish line. Here, we agree what done looks like before we start. The bill reflects the outcome, not the time on the calendar.

Two or three active clients at a time. That's it.

If I'm embedded in your work, I'm not scattered across ten others. The selectivity is how the engagement actually works. It's also why most available slots are filled from referrals before they're posted.

Receipts, not résumés

Anonymized field reports from active engagements. Real numbers. Real dates. Real lessons.

Read the field notes →

Tell me what you're trying to ship.

Selective by necessity — I work with a few teams at a time. We figure out together if there's real fit.