Three clear ways to work together.
The pressure to do something with AI rarely arrives with a plan. Here are three, each with a defined shape, length and result, so you know what you are buying, and I know what I am promising. Every one starts the same way: a 30-minute call and a one-page proposal. If none fits exactly, say so; the shapes bend.
Thirty minutes. You leave with what I would do first, whether or not we work together.
Opportunity Sprint
Choose the one workflow worth doing.
- Duration
- Two weeks
- Format
- Fixed scope, fixed fee
Starts with a 30-minute call and a one-page proposal. No commitment before the number.
Leadership teams with real budget, real pressure from above, and a long list of AI ideas, but no confident first move.
- 1A ranked map of AI opportunities across your workflows
- 2One chosen first build, scoped to ship in weeks rather than quarters
- 3A plain-language plan: data, risks, owners, cost
- 4A go / no-go recommendation you can act on the same day
In two weeks we look at how your operation actually works, find where AI removes real drag, and separate the few high-leverage opportunities from the expensive distractions. You end with one workflow to build first, and a plan your board can read in ten minutes.
A 30-minute call to understand your situation, then a one-page proposal within two working days.
First Useful Version
Ship the system, not the slide deck.
- Duration
- Six to ten weeks
- Format
- Embedded with your team
Starts with a 30-minute call and a one-page proposal. No commitment before the number.
Teams with one promising workflow, or a stalled pilot, that needs turning into something operations will accept and people will trust.
- 1A working system in your environment, used by real people
- 2Evals written as contracts: named failure modes and acceptable rates
- 3Approval points and hand-offs your operations team can live with
- 4Documentation and a handover, so ownership stays with you
I design and build the first production-shaped version of an AI workflow or agent, with your team and inside your environment. Visible inputs, clear stop points, and evals that define what good looks like before anyone depends on it. Not a prototype: the first version real people use.
Usually after a sprint, but not always. If you already know what to build, we start with a scoping session.
Operating Rhythm
Make it stick without me.
- Duration
- Monthly, from three months
- Format
- Advisory retainer
Starts with a 30-minute call and a one-page proposal. No commitment before the number.
Organisations where AI is already in daily use and needs structure: shared patterns, approval rules, and judgment that travels beyond one enthusiast.
- 1Monthly working sessions with the teams doing the work
- 2Reviews of the prompts, agents and workflows already in use
- 3Shared patterns and guardrails written for your organisation
- 4A direct line for the decisions that cannot wait a month
Monthly working sessions with the people who actually operate the systems: reviewing the prompts and agents already in use, setting approval rules, and teaching nontechnical teams to get consistent results, and to recognise when not to trust them.
A conversation about where AI already lives in your organisation, and where it is quietly causing problems.
From first call to handover. No theatre in between.
- 01
A 30-minute conversation
Bring one real situation. I will tell you honestly whether I can help, and what I would do first.
- 02
A one-page proposal
Scope, timeline, fee, and what you will have at the end. No forty-slide deck, no discovery theatre.
- 03
We do the work
Short cycles, visible progress, and decisions made with the people who will own the system.
- 04
You keep it
Documentation, patterns, and enough teaching that your team keeps moving after I leave.
What people ask before the first call.
- How is this different from hiring a consultancy?
- A consultancy sells you a programme and a bench. I am one senior person who has shipped and run AI products of my own, works directly with your team, and leaves when the system is running. You get the decision and the system, not the deck.
- Why not just run a vendor pilot?
- A vendor pilot is built to prove the vendor's tool. I am independent, so we start from your workflow and choose the tools afterwards, including the honest answer when a smaller model, or no model at all, is the right one.
- Do you work with nontechnical teams?
- Mostly. The people who benefit most from this work are operators and leaders, not engineers. I translate in both directions.
- How do you price?
- Fixed scope, fixed fee for sprints and builds; a monthly retainer for the operating rhythm. You will see the number in the proposal, not after.
- Where do you work?
- I am based in Amsterdam and work on-site across the Netherlands and Europe, and remotely elsewhere.
Something else? Email yvoderooij@gmail.com.
Not sure which one? Bring the situation. We’ll pick the shape together.
Replies within one working day. Or email yvoderooij@gmail.com.