The Exponential Agency · Workbook edition

The AI Takeoff
Playbook.

For non-technical CEOs of non-tech businesses. From first principles. Applies everywhere. No BS.
Nine moves Three phases 17 weeks May 2026
02 · The Answer
The constraint

The constraint on AI value in your business is organisational design, not technology.

We have studied firms at the frontier of getting value from AI and distilled what they do into nine moves, in order. Sector, size and type are irrelevant — the dependencies are the same.

Each move exists because a first principle requires it. Skip one and you miss the dependency that makes the next one work. You can run this yourself, with your team.

03 · The Schematic

Nine moves. Three phases. Two loops.

Move
Phase band
Recursive loop
PHASE 1 — SEE THE WORK MOVES 1 · 2 · 3 PHASE 2 — BUILD THE LOOP MOVES 4 · 5 · 6 · 7 PHASE 3 — REDRAW THE DEAL MOVES 8 · 9 CHECK · MOVE 03 writable target? REBUILD THE NEXT WORKFLOW RESTOCK THE TEAM 1 Tell the workforce what is coming 2 Map your cost workflows, not tools 3 Pick the first workflow 4 Rebuild the workflow 5 Universal access in parallel 6 Re-rate every major workflow 7 Right Person. Right Role. 8 Redesign the structure 9 Three populations three deals
17 weeks first cycle · then it repeats
Compounding is asymmetric — six weeks lost is not six weeks recoverable
05 · Move 01
Move
01
Phase 1 · See the work
Principle

Trust precedes behaviour change. People who do not know what is true cannot act on it.

Tell the workforce what is coming.

What you do
Stand up in front of the people who work for you. Not email. Not a memo. Not delegated to HR.
Name the threat and the retrain commitment in the same sentence. Be specific by function.
Pair the threat with a commitment you will fund — and a date by which you will fund it.
What you are testing for

Whether the room leans in or hides. Expect roughly 10% leaners, 20% leaners-out, 70% wait-and-see. Note the surprises — they are your first signal.

06 · Move 02
Move
02
Phase 1 · See the work
Principle

Workflows are the unit of value. Tools are not. Each workflow has three costs: thinking, coordinating, doing.

Map your cost.

What you do · Two-hour leadership session, four to six people in the room
Which workflows produce most of our revenue, cost, or customer outcomes?
In each, where is the time going — thinking, coordinating, or doing?
Of those, which have a target you could write down and a machine could check?
Output

One page. Workflows ranked by value at stake, with the dominant cost named in each. If the room cannot say where the time goes, you have just discovered the first thing to fix — and it is not an AI problem.

07 · Move 03
Move
03
Phase 1 · See the work
Principle

Workflow economics are governed by volume, stakes, and variability. AI lowers the cost of synthesis, coordination, and verifiable-target judgement.

Pick the first workflow.

Pick exactly one. It must meet all four conditions.
High volume. Many repetitions per week or month.
Recoverable downside. If the loop is wrong, you can fix it.
Writable target. What good looks like, written down by Friday.
Accessible data. Already inside the business, not scattered or uncollected.
Lock it

Lock the scope, the team, and a six-week clock. Resist the temptation to pick three. The benefit of going narrow is faster signal.

08 · Bridge to Phase 2
Before Phase 2 starts

Choose your technical build resource — pick before Move 4 begins.

Option A · Default

No-code AI tool, used directly by the workflow team.

Best for. A rebuild that fits inside one tool with prompts, basic logic, and a knowledge layer.

Tools. Claude Cowork, ChatGPT for Business, Zapier, n8n, Lindy, Gumloop, Make.

Cost$200 / user / month
Option B

Six-week consultant build.

Best for. Workflows that touch multiple systems, need light custom code, or have data a no-code tool cannot reach.

Avoid. Traditional agencies and SIs that default to discovery phases.

Cost$30k – $80k / six weeks
Option C

Internal hire — applied AI engineer.

Best for. Businesses planning to rebuild more than three workflows in 18 months. Maths only works at scale.

Skill. Has shipped at least one production AI workflow. Hybrid build + eval.

Cost$150k – $220k base, NZ
The default for your first rebuild is OPTION A. Most CEOs over-engineer the build resource. The first workflow is a learning exercise — if Option A fails, you now know exactly which lane the second workflow needs.
10 · Move 04
Move
04
Phase 2 · Build the loop
Principle

AI substitutes for synthesis, coordination and judgement where the target is verifiable. First attempt usually fails. Same executive succeeds with narrower scope.

Rebuild the workflow.

Six weeks · workflow team + technical build + executive sponsor
Write the target down formally. What good looks like. How you score it. The threshold above which the loop runs alone.
Specify three lanes. AI alone · human exception · human approval every output.
Run the recursive loop against the target. Iterate to convergence.
If the loop has not converged at week four, kill the scope. Do not kill the programme.
Worked example · invoice exception handling

Flag invoices where line-item variance from PO > 5%. Score: human reviews 50 random outputs / week. Threshold: loop runs alone above 90% accuracy across two consecutive weeks.

11 · Move 05
Move
05
Phase 2 · Build the loop
Principle

Permission and capability are both required for the work to change. They cannot be sequenced.

Run universal access in parallel.

Same week Move 04 begins · everyone in the business
Every employee gets a general-purpose AI tool. No forms. No approval gates. No usage caps.
Decide three things in writing first: which tools, the data policy, the approved-use list — signed by you and IT together. One hour.
Pair the access with 15-minute daily AI challenges for the first two weeks. The point is muscle, not output.
Replace formal L&D with peer learning — practitioners teaching practitioners.
What you are testing for

Self-selection. Curiosity is not equally distributed across the org chart. Watch who uses the tools without being told. The list is information no performance review can give you.

13 · Move 06
Move
06
Phase 2 · Build the loop
Principle

The cost map is recursive. Capability moves up the curve quarterly.

Re-rate every major workflow.

Half-day · CEO + workflow leaders + target writers
Take the cost map from Move 02 and finish the job. Apply the four conditions from Move 03 to every workflow.
Sort into three buckets: rebuild now · human-in-loop interim · leave alone for now.
Use what you learned in Move 04 to recalibrate. Workflows that failed the writable-target test six weeks ago may pass it now.
What you are testing for

Whether you have enough working evidence from Move 04 to make confident calls. If not, run a second Move 04 on a different workflow before scaling.

14 · Move 07 · Loop origin
Move · loops back to Move 04
07
Phase 2 · Build the loop
Principle

AI cannot write the target. AI works best in recursive loops against verifiable targets. The remaining human-only work upstream is target writing.

Right Person. Right Role.

One-hour decision · ongoing development · CEO + HR
Identify the people who, during Move 04, translated an ambiguous outcome into a checkable rule. The skill is rare and currently invisible on most org charts.
Name the role. Promote them visibly. Pay them. The pre-AI analogue is a Turnaround Executive or Transformation Officer.
Build the next workflow rebuild around them — this is where the loop returns to Move 04 with a stronger team.
Build a development pipeline. Teaching looks like frontier research — show-and-tell, paper reviews, demos — not classrooms.
What you are testing for

Whether you can resist promoting purely from existing seniority. Their old-org-chart seniority is irrelevant — their ability to translate ambiguous outcomes into checkable rules is the qualification.

15 · Move 08
Move
08
Phase 3 · Redraw the deal
Principle

Firms exist because internal coordination is cheaper than market transaction. AI lowers internal coordination cost. The structure designed at the old cost is leaking value.

Redesign the structure.

Trigger · three or four workflows rebuilt
Spans of control widen. Layers thin. Functions merge.
The new shape is an hourglass: accountable humans on top · AI orchestration in the middle · lean human delivery + AI agents at the base.
Do this now, not before. Designing on paper before the rebuilds gives you an org that will not survive contact with the work.
This is a political event, not a workshop. Run consultation in parallel with the design. Communicate the rationale (data from Moves 4–6) before the structure.
What you are testing for

Whether the redesign is forced by what you have learned, not by what you read in an article. If you cannot point to specific coordination cost that has collapsed, you are guessing.

17 · Move 09
Move
09
Phase 3 · Redraw the deal
Principle

Trust precedes behaviour change, applied at the population level. Different motivations require different deals.

Three populations, three deals.

Eight to twelve weeks of universal access has now sorted your workforce
Leaners (10–20%) need protection. AI does not reduce work — it intensifies it. Define what you authorise them to stop doing. Burnout risk in 6–9 months.
Watchers (60–70%) need proof in three currencies: visible promotion of leaners, peer-led demonstration in their function, retrain budget actually spent on them.
Refusers (20%) need honesty about the timeline. Pretending it is longer than it is collapses trust with watchers and loses the leaners first.
What you are testing for

Whether you can write three different deals at once. Most leaders write one deal for everyone. That is the most common reason transformations fail at the human layer.

Where you are now

Then it repeats.

A workforce that knows where it stands, sorted toward the work that compounds. The cycle restarts at Move 02 with a richer cost map and a stronger team — led by the target writers you named in Move 07.
theexponential.agency · Designed for an exponential world