// Playbook — AI value · conceptual

All in on AI. Still waiting for the value.

A short, conceptual playbook for organisations that have already committed — the licences bought, the pilots running, the copilot on every desk — and can't yet point to the line on the P&L that moved. The problem isn't too little AI. It's too much of it, pointed at nothing in particular. Worked example: Tallowood Bank, seen through its procurement team — switch to the function closest to yours. Reference: Bain & Company's AI value paradox.

11AI use cases
· in procurement

3Pilots
· live at once

1Contract tool
· bought first

0Cost-to-income points
· attributed to AI

// See it in your function

01 —

Meet Tallowood’s procurement team.

Procurement · composite

An AI contract tool. Three supplier lists underneath it.

Procurement at Tallowood Bank — an ASX-listed regional bank built from three mergers — inherited around 1,800 suppliers across three vendor masters. When the board went all in on AI, procurement moved first: an AI contract-review tool, a supplier chatbot, a copilot drafting RFPs.

The tool reviews contracts quickly. It can’t tell that “Smith IT Services” and “Smith IT Pty Ltd” are the same supplier. Only 41% of spend is under contract. And under the prudential standard for operational risk, the bank has to know which of those suppliers are material service providers — today, that list lives in a spreadsheet.

Nobody did anything wrong. The tool was good. The data under it wasn’t ready.

The worked example. Procurement goes from three pilots to two funded bets tied to cost-to-income and third-party risk. Tallowood Bank is a composite — not a real bank — and the figures are illustrative.

1,800Suppliers
on the books

3Vendor masters
· one per merged bank

41%Spend
under contract

12People in
procurement

// And the people you'll meet

// Cast 01

The Transformation Lead

You, the reader

Has to connect procurement’s AI work to a number the market actually tracks — without undoing a year of effort.

// Cast 02

Nadia Petrakis

Head of procurement · function lead

Inherited three supplier lists. Bought an AI contract tool before they were merged.

// Cast 03

Ines Moreau

Chair · the board

Asks which value driver the synergies from the mergers show up in. Procurement is where many of them live.

// Cast 04

Raj Mehta

CIO · owner by default

Integrated the contract tool on time. Nobody asked him about the vendor master.

// Cast 05

Owen Burke

Procurement officer · the frontline

Spends Mondays de-duplicating suppliers by hand. Knows which forty contracts matter.

// Cast 06

Beth Okoro

Head of operational risk · the line

Needs a reliable register of material service providers. Can’t sign off one built on three spreadsheets.

02 —

What all in looks like.

Five symptoms · after Bain

// Bain calls it the AI value paradox: frenetic activity, flat earnings. Roughly four in five CEOs say their AI programme is at best partly delivering. Five symptoms show up again and again — Tallowood has all five — and so does its procurement team.

// 01

The swirl.

Ideas arrive faster than anyone can prioritise them. Effort spreads so thin nothing gets enough attention to matter.

// 02

Minutes, not margin.

Faster emails, quicker summaries. Everyone is slightly more efficient. Enterprise value doesn't move.

// 03

Tools first.

A copilot bought, then layered on the old workflow. The process never got redesigned — it got decorated.

// 04

Pilots forever.

Demos that work and never reach production. The data, the platform or the people to scale them aren't there.

// 05

Owned by IT.

AI handed to technology. No visible owner at the top, so no one is convinced it will last — and it stalls.

03 —

Four shifts.

Value first · tools last

// Not more AI. Different AI. Four shifts, in order — each one only works if the one before it has happened.

From

The use-case register

The cost-to-income ratio.

Procurement’s line in the investor presentation is cost-to-income. At Tallowood: spend under contract, supplier count, material service providers known. Every AI idea is judged against those.

From

Choosing a tool

Redesigning the work.

Owen and two buyers mapped how a purchase actually happens across the three legacy banks — the answer was one vendor master first, AI second.

From

Many pilots

A few bets, in production.

11 use cases became 2 funded bets: supplier de-duplication and a material service provider register. The chatbot stopped. The contract tool was paused until clean data could feed it.

From

Launching

Landing the change.

Beth co-owned the register from day one — risk and cost moved together, not in turn. Spend under contract is now a number the board sees.

The tool was good. The data under it wasn’t ready.

04 —

Three ways the recovery stalls.

Common pitfalls

// Pitfall 01

Cutting without a story.

Pilots are stopped by spreadsheet. The people who built them hear it as a verdict on their judgement — and stop bringing ideas at all.

The fix

Explain the cut against the value plan, in person. Keep what was learned. A good call can still have been the wrong bet for now.

// Pitfall 02

A new swirl, better organised.

The register gets a new front door, a scoring form and a steering group. The volume of ideas stays the same. So does the result.

The fix

Measure how much gets declined, not how much gets captured. Saying no is the operating model working.

// Pitfall 03

Winning at the pilot.

The four bets get a launch and a slide. Production, adoption and the data work underneath quietly lose their funding to the next idea.

The fix

Fund the path to production up front. Call a bet done when the value plan moves — not when the demo works.

05 —

Five questions for Monday.

Start here

// No template required. If the leadership team can't answer these in one meeting, that's the diagnosis.

Which three to five outcomes does our strategy actually depend on?

If the answer is a list of twelve, the swirl has reached the strategy too.

For each live AI initiative — which of those outcomes does it move?

"Productivity" isn't an answer. Name the line.

What would we stop tomorrow if we had to fund only four?

Most teams already know. The question is whether anyone can say it out loud.

Who, by name, owns AI value — not AI technology?

If it's the CIO by default, the sponsorship gap is already showing.

Have the people who do the work redesigned it — or just been given a tool?

Tools on an unchanged workflow buy minutes. Redesign buys margin.

06 —

Using this in practice.

Closer

Fewer bets. Bigger lines.

This playbook is conceptual by design — the shape of the problem and the order of the way out. One organisation, four functions: procurement, IT, People & Culture, finance. The swirl looks different in each — a contract tool here, a chatbot there — but the four shifts are the same. What changes is where the function’s line on the plan sits, and who asks which line? For procurement, it’s the Chair, against cost-to-income. Tallowood is one organisation; yours will differ.

// Where this connects — the full method for diagnosis, prioritisation and landing change lives in the transformation methodology: Getting past the swirl →

If your organisation has gone all in and is still waiting for the value, I'm happy to talk through where it's getting stuck.

// Reference — Brian Kmet, Roger Zhu, Benjamin Cooke & Robert Howgego, Getting Past the AI Value Paradox in Private Equity, Bain & Company, 16 September 2026. The five symptoms and four shifts adapt Bain's obstacles and steps; the worked example and framing are my own.