The thesis For founders & boards

The org chart
was a
workaround.

For all of history, thinking was human and attention was scarce, so we built departments, approval chains, and review meetings around the shortage. That constraint has lifted. This is the argument for what comes next, and the programme I run to get organisations there.

The coordination machine, redesigned: same people, arranged around intelligence.

The premise

Every organisation ever built
is a coordination machine.

Org charts, department structures, approval chains, review meetings: infrastructure for getting the right human judgment to happen in the right order at the right time. Those workflows were never efficient. They were the best available answer to a hard constraint. And for the first time, that constraint is gone.

It reads and writes at volume.

A new class of systems performs cognitive work directly, at the volume your organisation actually produces: every order, every ledger entry, every complaint, every report.

It works continuously.

No working hours, no leave, no attention that drifts. The night is no longer dead time.

It acts within mandates.

Defined actions, executed without being asked, with every step logged and reviewable. Not a better tool for the machine's operators: a new kind of worker inside the machine.

The trap

Most companies will get this wrong. A chatbot here, a document tool there: it feels like progress, and it is the expensive path. Automating an existing structure makes the wrong shape run faster: the same bottlenecks, the same blind spots, at higher speed. The pilot succeeds; the organisation does not change.

The advantage is never in the effort applied to a structure. It is in the structure.

An illustration, not a prediction

While the company slept,
the company worked.

  1. 22:00

    The day closes. Orders, ledgers, and messages stop arriving. The agents' shift begins.

  2. 01:30

    Orders reconciled against stock and dispatch plans. Exceptions flagged, each with its reason attached.

  3. 04:00

    Receivables reviewed. Follow-up notes drafted for the three accounts that need attention.

  4. 06:40

    The morning brief assembles itself: production, sales, cash, and the two decisions that genuinely need leadership.

  5. 09:00

    The managing director walks in to a company three hours into its day. Every overnight action logged and reviewable.

Nothing was approved by a machine. Everything was prepared by one.

Four strains

The knowledge walks out.
The coordination tax stays.

  1. 01

    The invisible knowledge problem

    Decades of judgment, supplier history, pricing instinct, and process memory live in the heads of a few people, and leave in their resignation letters. Not a training problem: an organisational vulnerability, and in almost every company, entirely unmeasured.

  2. 02

    The coordination tax

    Approvals, reconciliations, reminders, status meetings, sign-offs: message traffic ricocheting between departments with high motion and low forward progress. And it grows faster than revenue does.

  3. 03

    The leadership bottleneck

    Decisions queue behind a small number of senior people, so the organisation moves at the speed of its most burdened calendar. The people hired for judgment spend their days on assembly.

  4. 04

    The paper organisation

    Processes exist impeccably on paper while the real work runs on tribal knowledge: known to the people who do it, invisible to everyone else. The gap between the documented organisation and the real one is where operational risk lives.

These strains are treated as facts of life. They are design consequences of the old constraint, and they are now solvable.

Three layers · six outcomes

Same people. Same business.
Different structure.

Today, everything routes through people: every question, every exception, every approval climbs to the same few heads, and the centre is always the bottleneck. In the AI-native organisation, an intelligence layer carries that load continuously. What reaches leadership is the exception and the decision, not the assembly work. The centre becomes judgment, not traffic.

The redesign, drawn once: judgment stays with people; the carrying moves to the layer.
3

The leadership interface

Sits where a chief of staff sits: between leadership and every department. Briefs, answers, exceptions, and the day's genuine decisions, in one place.

2

Departmental AI agents

Autonomous agents with defined mandates, one per function: finance, operations, management. Each watches its domain, flags what matters, and executes defined actions. Agents draft and flag; your people confirm.

1

The knowledge foundation

SOPs, workflows, and decision logic captured as active, queryable, versioned knowledge. The institutional memory every layer above draws on.

Connects to what you already run: ERP Accounting CRM Documents Email & messaging

What good looks like:

What this is not

Not a replacement of your systems.

The intelligence layer sits above the ERP, accounting, CRM, and document stores you already run. It connects them and makes those investments work harder.

Not headcount reduction by another name.

The purpose is to amplify the judgment of the people you have. Machines prepare; people decide. Every draft, summary, and flag is reviewed by a human before it becomes an action of the company.

Not surveillance.

The system tracks processes and promises, never per-person performance. It watches whether Tuesday's commitment was kept, not who works late. A design principle, not a configuration option.

Not a leap of faith with your data.

Nothing touches core systems until security, privacy, and compliance are explicitly scoped. Everything built is specific to your company, access-governed, and logged.

The machine prepares the decision. The person makes it, and owns it.

Three tracks, one programme

Diagnose. Transform.
Build.

The work of my practice, run as one programme. The diagnostic tells us where to act; the transformation track carries your people; the build track delivers the system. Every engagement is short, focused, embedded, and accountable: diagnose, rebuild the operating cadence, install the systems, exit when the team owns it.

Track A · The mirror

Diagnose

A four-part organisational diagnostic: the technology stack, leadership readiness, and how the company actually works, converted into a sequenced roadmap the board can govern. Valuable even if you stop there: you will know your own organisation better than you ever have.

You receive

  • A stack health report and a prioritised gap list
  • A frank leadership readiness map
  • The knowledge, workflow, and decision-logic inventory, and it belongs to you
  • A phased roadmap with owners and measurable definitions of success

Track B · The people

Transform

Transformation fails in the middle of the organisation, not at the top. A bespoke management programme makes the change durable, sequenced so the organisation feels time coming back before anything new is asked of it.

You receive

  • A change programme with a named executive sponsor and champions
  • Hands-on leadership sessions: each senior leader ships one real automation
  • An operating cadence redesigned around a shared, current picture

Track C · The system

Build

The three-layer intelligence system, built inside your company by a dedicated AI engineering team. In phases, each paying for the next, starting with quick wins that visibly return time to named people.

You receive

  • Quick wins inside the first 60 days
  • The knowledge foundation, live and queryable
  • Departmental agents in the functions that hurt most
  • The leadership interface: the morning brief becomes real

And it is instrumented, so you see the return rather than feel it: a baseline in the diagnostic, then the same four dials tracked throughout: hours returned to senior people, cycle time on the chosen workflows, follow-ups completed without human chasing, and the share of critical knowledge captured out of heads. No dashboard theatre. Cost follows scope: pricing is a focused conversation after the diagnostic is agreed.

One rule governs all of it: every intervention must remove more work than it adds.

The window is open as you read this.
It is the kind that closes.

Latecomers will be able to buy the tools. They will not be able to buy the experience of having built with them from the beginning. The way most businesses are run today was designed for a world that no longer exists. Yours can be one of the organisations rebuilt for the world that does.

The next step is a ninety-minute working session with your leadership team. Not a sales call: a conversation about where your organisation is and what it could become.