I
II
III
IV
V
VI
VII
VIII
Limit of survey
The Blank Marginthe first question
Trig Pointread what exists
Base Campsmallest slice that runs
The Ridgewhat's expensive to change
The Long Climbthe build
False Summitslooks done, isn't
The Summitlaunch
The Far Siderunning it
Bay of Assumptions
Strait of Second Thoughts
Sea of Scope
Uncharted

You are hereStage I, the first question

I

Roy van KaathovenSurvey of a projectSheet 1

Where do you even start?

How I take a project from the first question to running in production.

Eight stages, drawn as the ground they feel like. Follow the red line. Each stop says what happens, what you get and what I need from you.

Walk the routeContact me

Heights show how hard the going usually feels. Nothing on this sheet was measured.

Stage IThe Blank Margin

Where do you even start?

With the problem. Before anyone says a word about stacks or tools, I want to know who has this problem, what it costs them today, and what changes when it goes away.

What happens
A conversation. You tell me what's worrying you, or what decision you're trying to make.
What you get
A straight answer on whether I can help, and what the first step would be.
What I need from you
The problem in your own words. You don't need a spec.

Stage IITrig Point

What are we actually dealing with?

A surveyor fixes a few points they can trust and measures everything else from those. If a system already exists, I read the code and the infrastructure alongside the diagrams, because the design and the implementation usually explain each other. Then I separate what will really bite from what only looks untidy.

What happens
I read what exists and list the constraints and the failure modes that matter, ranked by likelihood and blast radius.
What you get
A clear written assessment and a session to walk through it: what to fix now, what can wait, what to leave alone.
What I need from you
Access to the code and infrastructure, and the questions you most want answered.
This can stand alone as a fixed-scope review

Stage IIIBase Camp

What is the first thing that actually works?

The smallest slice that runs end to end, in the real environment. It's small enough to throw away and real enough to learn from. Everything later gets carried up from here.

What happens
One path through the product, from the first screen to the data and back, deployed where it will eventually live.
What you get
Something you can click, and a short list of what it taught us.
What I need from you
Someone who can say yes or no quickly, and one real user or workflow to aim at.
The live-streaming platform began as a prototype I wrote in 2019

Stage IVThe Ridge

Which decisions will be expensive to change?

Up here you can see both valleys. Most decisions are cheap to reverse and a few are not, so those few get the time. I keep designs simple and reversible, and I work on the architecture and the team together because they mirror each other.

What happens
I decide the things that are hard to undo, and I write down the trade-offs behind each one.
What you get
Decisions you can read, explained, and grounded in the codebase I'm working in.
What I need from you
Your view on what the business can't afford to get wrong.
How this works as interim CTO

Stage VThe Long Climb

How does it actually get built?

I own the design, the architecture and every decision. AI coding agents do much of the implementation in parallel branches against a written plan. That only works if the codebase is built for it: a design document per system, and a test suite and typecheck that every change has to pass.

What happens
Small, AI-augmented teams work through the plan while I direct the agents, check their output and decide what's good.
What you get
A dependable cadence of working software, instead of ad-hoc shipping.
What I need from you
Regular, honest reactions to what's on the screen. You're the one who knows the users.
Tiny Tactics was built this way, first commit to live in about 2 weeks

Stage VIFalse Summits

It looks done. Is it?

Usually not yet. Some generated work is confidently, plausibly wrong, which is the worst kind because it looks finished. Turning a pile of raw material into something shippable is the job: selecting, judging, killing the bad stuff, fixing the almost-right stuff.

What happens
Deterministic gates for non-deterministic output, tests above all, and QA in a real browser.
What you get
An honest list of what is done and what only looks done.
What I need from you
Real data, people willing to try to break it, and patience when I say it isn't ready.
More on this in Designing the factory

Stage VIIThe Summit

What does launch take?

Getting out the door is the quieter half of the job. You depend on a lot of moving parts you don't control, and each one has to be integrated, tested and signed off before it can go live.

What happens
I pull the moving parts into a single launch: our side, the outside teams, the payment providers in the middle.
What you get
A product that is live, with real people on it, and a view of what they're doing.
What I need from you
Someone who owns each outside party, and a date you can defend.
I led a gaming platform's launch this way

Stage VIIIThe Far Side

And after launch?

Now you live here. A live product has no quiet maintenance window, so I invest early in making problems visible before users have to report them: errors, analytics, health checks, alerting and clear ownership.

What happens
We run it, watch what people actually do, and let that decide the next trip up.
What you get
A system your team can run. If you want me to stay, that can be anything from a few days a month to a fuller interim role.
What I need from you
A decision on who owns it from here.
Six years of this on the live-streaming platform

Legend

This sheet has no time scale. No two projects walk it at the same pace, some stages take an afternoon and some repeat. Tell me where you are and I'll tell you what I'd expect from there.

Symbols

  • Route, first question to production
  • Route over open water
  • Index contour, every fifth line
  • Named stage
  • Limit of survey
  • Unsurveyed, in pencil
  • Elevation tints, easy going to hard

With sound on in the header, the contours move to Godspeed, a track of mine.

Gazetteer

  1. IThe Blank MarginWhere do you even start?
  2. IITrig PointWhat are we actually dealing with?
  3. IIIBase CampWhat is the first thing that actually works?
  4. IVThe RidgeWhich decisions will be expensive to change?
  5. VThe Long ClimbHow does it actually get built?
  6. VIFalse SummitsIt looks done. Is it?
  7. VIIThe SummitWhat does launch take?
  8. VIIIThe Far SideAnd after launch?

Sheet 1The Blank Margin

Your project next?

You've seen the whole route, and you're back at its start: a question and a lot of blank paper. That's the right place to be. Tell me what the problem is and we'll work out the first mile.

Tell me about your project

Sheet 1. Printed in five inks, more or less in register.