MODE / 01
Build.
Make it real. Get a signal.
Prototype with AI, experiment and put something in front of a customer. Learn from what happens.
A field guide for leaders in the AI eraBy Peter van Sabben
The future
won’t rebuild
itself.
AI changes what you can build. Rebuilding changes how your organisation thinks, decides and grows.
01 Build with intent02 Judge with evidence03 Align for growth
01 /The identity
Building got easier.
Knowing what to
rebuild didn’t.
The next challenge is changing the company while it’s still running.
You have customers, a team and a business that works. You also have assumptions built for a world that is changing. Adding AI tools is only the beginning.
A Rebuilder retests the business model, the strategy and the way people work together. On purpose. Before the old way stops working.
Read the founding essayRebuilder / noun
A leader who moves deliberately between building, judgement and alignment, and turns that rhythm into an organisation’s ability to grow.
Three modes.
One constant rhythm.
Different work calls for a different posture. The skill is knowing when to switch.
MODE / 02
Judge.
Decide what’s worth doing.
Challenge assumptions. Write predictions before you build. Use evidence and taste to keep, kill or scale.
MODE / 03
Align.
Build a shared reality.
Invite dissent. Make decisions and their reasoning clear, so humans and agents can act together.
The capabilities
Nine capabilities.
Fuelling the three modes.
Three capabilities fuel each mode. Five core leadership mindsets underpin them all.
01 — 03
Builder Mode
Execution is cheap, signal is not. You build to find out, not to ship.
01AI Prototyping at Speed
Build the thing that settles the question in hours instead of writing the deck that describes it. Speed matters because it shortens time-to-signal, not because shipping is the point.
02Experimentation
Design the cheapest test that produces a clean signal, and prefer signals someone paid for or acted on over signals someone reported. And hold your own agents to the same bar. An eval is just an experiment with a pass mark.
03Distribution
Build the way it reaches people into the product itself. When anyone can build anything, supply stops being the constraint and attention becomes it.
04 — 06
Judgement Mode
Options are cheap, so selection is the bottleneck. Including the largest selection of all: what this company is for.
04Taste
Tell a strong idea from a merely plausible one before you spend a week on it, by comparing against examples rather than trusting your gut. At the top of the company the same act has a different name: the payoff function—choosing what this business is deep in.
05Decision Architecture
Record what was decided, why, what good looks like, what’s out of bounds, what you rejected, and why you made the last exception. That last part is what your systems of record never hold, and it is the part an agent needs. At company scale, this capability is what builds the world model.
06Calibrated Judgement
Write the prediction before you build, score it afterwards, and know how often you’re actually right.
07 — 09
Alignment Mode
The picture has to hold when you’re not in the room, and people have to believe it. Trust is what makes a decision travel.
07Inviting Dissent
Make disagreement cheap enough that you hear it while it’s still useful. Psychological safety is the result; this is the behaviour that produces it.
08Decision Briefing
Write a decision down so it can be executed without you: what was decided, why, what good looks like, what’s out of bounds.
09Reading the Room
Tell real agreement from polite agreement, in time to do something about it.
Explore a capability to see what it means in practice.
The shared foundationCore leadership
Core leadership
mindsets.
How a Rebuilder shows up, whichever mode the work calls for.
Growth Mindset
Cultivate curiosity and rapid learning cycles that normalise experimentation and adaptation.
Systems Thinking
See whole systems and causal loops; design interventions that tackle root causes.
Strategic & Transformational Thinking
Set a bold long-term direction and sequence decisive bets toward lasting change.
Change Management
Plan and sequence reforms, build coalitions, move from pilot to sustained adoption.
Energy, Resilience & Health
Sleep, train and recover properly. Energy and clarity are unfair advantages once tools are equal.
02 /The operating practice
The rebuild has
no finish line.
But you can tell whether the way you work has changed. Start with three questions.
Explore the six testsCan people decide without you?
Give the people closest to the work authority over reversible decisions, inside written limits.
Do you know how often you’re right?
Record predictions before the work begins. Score them afterwards. Make learning visible.
Is killing something a normal event?
Stopping a weak bet should be a routine decision, with a process and a named owner.
Agents can act. Humans remain accountable. Put a name, a test and a way back beside every agent in production.
03 /The practical playbook
Start at the edges.
Rebuild inward.
First, test whether a rebuild is needed. Then use a digital twin to test workflows, scale what works and gradually migrate the core.
Nine frameworks to get started
01Where value is leaving
Look at what you charge for and where your costs sit. Locate the exposure in your own company.
What is the unit on our invoice: hours, seats, outcomes, or access to something physical?
02Business model vs organisation
Score the two mismatches separately: charging for what got cheap, and spending too much on coordination. Having both is the case for a rebuild.
Take one core workflow: how many days does it take, and how many minutes is anyone actually working on it?
03The Moat Test
Test whether your advantage survives when time and effort get cheap. A useful test must also be able to say: don’t rebuild.
If a competitor had our AI tools tomorrow, why would customers still choose us?
04Archetype & route
Identify which mode disappears under pressure. Strengthen it, then choose a route that fits your capacity and the constraints of the running business.
In the last bad quarter, what did we do more of: decide alone, start things, or review things?
05The mandate
Agree the protected resources, the expected dip, the review rhythm and the stopping conditions with the board. Put both sides’ commitments in writing.
Who signs the one-way doors, by name, and who is allowed to call the kill?
06The lab & the portfolio
Run ten small probes from one connected bundle of workflows. Every probe starts with a prediction and a condition that will end it.
For each probe, what did we predict, and can we show it was written before the build?
07The evidence review
Hold a 90-minute Wednesday gate. Review predictions, evidence, ownership and costs. Record one verdict: kill, keep or scale.
Did anyone pay for this or act on it, or were we only told that it worked?
08Organisation redesign
Classify each workflow step: plain software, an agent, or human judgement. Name business, technical, risk and human approval owners.
After this bundle migrates, does it still pay anyone to sit still?
09The roadmap
Move through four phases: Establish, Prove, Migrate and Normalise. Advance on evidence at each gate, with leadership one beat ahead.
Which gate would we fail today, and what is the one artefact that is missing?
Establish → Prove → Migrate → Normalise
Progress through evidence. Repeat as the world changes.
The thinking behind
the rebuilding.
Three essays by Peter van Sabben. A starting point that keeps evolving.
01The Rebuilder Playbook:
Introducing the Rebuilder
A new leadership identity
for the AI era.
The Rebuild Has No Finish Line
What you can check in
the way your company works.
The Rebuilder Playbook:
How to start?
Nine practical frameworks.
From diagnosis to action.
Created by Peter van Sabben.Built on practice. Open to challenge.
Build. Judge. Align.
Then do it again.
Follow the ideas, questions and practical work behind the Rebuilder.
Follow on Substack