Scott Farrell · Chief Scientist, LeverageAI

Ideas explain code.
Code tests ideas.

I study how AI changes what people can do, what companies are worth and how work should be organised — and build the systems that test the thinking.

7 disciplines· 263 ebooks· 28 years· Systems running today
Read the practice → Request the first meeting
The question the work follows

What changes when intelligence becomes more available — and how should people and companies reorganise to benefit?

The work follows that question from philosophy to code — and back. Experiments expose weak theory; better theory changes what gets built.

The shape of the practice

Seven disciplines. One arch.

Value is carried across the span — from why it matters to what provably works. Not a pyramid: nothing here is a base you finish first. Every stone is load-bearing, and the load only crosses when the arch is closed.

A seven-stone arch carrying a block labelled AI value: philosophy, economics, strategy, a gold governance keystone, knowledge, responsibility, engineering — standing on “why it matters” and “what works”, with thrust arrows running both ways
Take one stone out and the load finds the gap. That gap is where intent is lost between disciplines.
Governance is the keystone — not the last stone laid, the one that lets capable action stand.
More AI on top is carried by a complete arch. On a broken one it carries nothing, however capable the model.
Each stone opens in full — choose one
01 Philosophy What is AI changing? Read → 02 Economics What becomes cheaper — and what becomes more valuable? Read → 03 Strategy What should be pursued? Read → 04 Governance Under what conditions may the system act? Read → 05 Knowledge What must the system understand about this organisation? Read → 06 Responsibility What continues after the conversation ends? Read → 07 Engineering Where should intelligence sit? Read →
The research loop: observe, frame, challenge, build, inspect, revise
How it works

It runs in both directions

An inquiry can start anywhere — a research finding, a commercial assumption, an agent behaving badly. The job is to find the level that explains it.

A request for a better prompt → missing knowledge
Repeated escalation → missing authority, or a missing tool
A technically successful pilot → a weak unit of sale
An attractive strategy → nobody owns the outcome once it starts

The symptom → the mechanism

Across the company

The object of design is not an AI application.
It is how the business works as a whole.

Marketing, sales, operations, service, HR and finance are not six AI use-case lists. They are six views of one promise. Change the promise, and every view changes.

Six functions around the commercial promise; customers, suppliers and partners on the outer ring
One change, worked From selling hours → to a maintained decision service
Marketing
explains a different benefit
Sales
secures a different commitment
Operations
sustains it
Governance
bounds it
HR
reshapes the roles
Finance
measures a different unit
The machinery

The books are part of the machinery

Not content marketing beside the technical work. An ebook explains the argument, a doctrine states the rule, a playbook makes it a procedure, a wiki serves it during work — and code tests it.

Ebook → doctrine → playbook → wiki → code
Part of the practice Representative titles What it contributes
Human thinking & consequences The Reshape · The AI Learning Flywheel · HR Governance Seat Problem framing, augmentation, learning, the human terms of adoption
Economics & strategic choice The Terminal Value Doctrine · The Re-Roll · The Cognition Dimension Ladder Value migration, commercial redesign, defensible search
Knowledge & organisational understanding Executable Worldview · The Wiki Is the Kernel · The Wiki Playbook Source-connected institutional knowledge that informs live work
Authority & persistent work Two Leashes · The Business Runtime Bounded action, continuing responsibility, explicit completion
Engineering practice The Generative Pendulum · Designing Loops, Not Prompts Agentic / deterministic boundaries, feedback, verification, reuse

Thirteen titles shown. The library holds 263, on two wikis — the ideas and the builds, inspectable side by side.

Receipts
28 years

enterprise delivery — insurance, banking, energy, Defence

Built & sold

an 11-person consulting firm — these economics are lived, not theorised

Taught twice

postgrad lecturer — MGSM 2001, CSU / IT Masters 2018–20

263 ebooks

doctrines, playbooks and field guides — a published, queryable body of work

Two wikis

IP and dev — the ideas and the builds, cross-linked and source-grounded

Running today

production AI systems in live businesses — governed, audited, measured

Where this meets your company

The Review is this practice, applied to your company in four weeks.

We arrive with a thesis about your market and the machinery to test it. Your numbers decide whether it survives. Forty-five minutes to start.

Request the first meeting The Future Value Review →