The AI Business Case Review starts from the operating model you run today, not from the technology, and in three weeks tells you which of your problems AI can actually solve, what each is worth, what it costs to build, run and govern, and how to spend the budget and when to release it. Where it finds a case, I build the operating model to run it, and where the engagement warrants it I take the seat on a fractional basis. It is written by someone who has run that operating model across the whole fund cycle for twenty-five years, from the buyer’s side of the table, with no payment from vendors.
Every firm in this market is under pressure to do something with AI. The pressure produces motion, the motion produces spending, and most of that spending ends up with whoever sold the tool rather than with the firm that bought it. The value of an AI system does not sit in the algorithm. It sits in a better prediction that drives a decision someone in your firm owns and an action your firm can actually take. Strip out the decision and the action and what is left is a licence fee.
The usual AI strategy starts with the six families of technology and asks where each could be applied. It reads well and it funds nothing, because it was never anchored in how the firm works. This review starts at the other end, with the operating model you already run: the functions, the processes, the service providers, the controls and the people, and what each of them costs you today in time, error, cost and exposure. Only then does it ask which of those problems a machine could solve, and the answer is narrower and more useful than a technology tour. A task has to be data-driven, well defined and repetitive before AI is the right answer, and the firm has to hold the data and the skills to absorb it, which is where most cases quietly fail. A task being automatable does not make automating it worthwhile; the case has to cover the full cost of building, running and governing the system.
Not whether to use AI. It is already in the firm, through your vendors, your administrators and your own staff, and under the Senior Managers regime the people around the table are personally accountable for it whether or not they chose it. What the board is deciding is money: how much the firm should be willing to spend given what is genuinely available to it, how that spend is spread between the cases that pay back inside a year and the one deliberate bet that may not, on what evidence each tranche is released, who owns the portfolio, and how the board will know, a year from now, whether it worked. That is a budget decision with a governance discipline around it, and it is the part every AI strategy on the market leaves to the client.
The regulatory position is built into the case rather than bolted on. The FCA has said it will not write AI-specific rules and will hold firms to the frameworks they already answer to, so there is no commencement date to wait for. Under the EU AI Act the duties that fall on a firm using AI on its clients, investors or staff were never deferred. Each opportunity in the register carries its own regulatory exposure, who the deployer is and what that obliges, so the board is choosing between cases it can defend.
The clock starts when your files arrive, not when the engagement is signed. What you cannot produce is itself the first finding.
A scoping call, then a one-page list of what I need and where it usually lives: the strategy and risk appetite statements, the organisation and process maps you have, the outsourcing and service provider register, the systems list, cost and headcount by function, recent board and risk papers, any AI pilots already running and what they have cost, and the inventory from a Vendor AI Oversight Review where one exists. Your team gathers it; the three weeks begin when it lands. A half-day session with the executive team is booked in for week one.
The executive session surfaces the firm’s most important problems and ranks them by what they cost today, with a named owner against each, because a problem nobody owns is a problem nobody will fund a solution to. Each is then mapped to the AI that actually fits it, across generative, predictive, agentic, language, vision and rule-based automation with judgement added, and put through a single screen: is the task data-driven, well defined and repetitive, and does the prediction connect to a decision someone owns and an action the firm can take. Alongside it, a plain reading of the data and the skills the firm holds, and whether each case is incremental, improving a process you already run, or disruptive, doing something you do not do today.
Each survivor is tied to the strategy and the risk appetite, and anything outside them is set down however attractive it looks alone. The build-or-buy call is made honestly, including where buying a generic tool would simply hand the advantage to the provider. The regulatory exposure of each case is stated. Then the leaders are valued and costed through three lenses: the return the opportunity creates against the full cost of building, running and governing it, with build cost from a transparent day-rate model; the risk-adjusted return of doing it under proper governance, the loss or penalty avoided weighted by its likelihood; and the cost of inaction, what the firm forgoes for each quarter it waits. Each lens is kept separate. Nothing is summed into a single inflated total.
The rated register becomes an investment portfolio. The envelope: what the firm should be willing to spend given what is on the register. The allocation: the split between incremental cases and the reserved share for the one deliberate bet. The release mechanics: funding staged against pilot results, with the kill criteria agreed before the first pound moves. The ownership: one owner for the portfolio and the quarterly view the board sees, with the numbers rerun as they change. And the sequence the budget funds, in order, with an owner against each step. All of it written for a board, not a technologist, in a paper the accountable executive tables as their own.
A rated register of every candidate with an owner on every line, including the pilots already running and whether to fund, stop or redirect them. The business case on the leaders, and the indicative ROI model itself, which you keep and rerun as your numbers change. The cost of inaction per quarter. The build-or-buy call and the regulatory exposure for each case. The budget envelope, the allocation, the release gates and the kill criteria. The sequence, with owners. A vendor-neutral annex mapping tool categories to each use case. And the board paper.
A four-to-six week build of the AI-ready operating model for the opportunities the board has funded: the functions, processes and outsourced services that change, build, buy or outsource re-cut for AI, the control model that has to change when judgement moves to a system and agents act, the roles and the people to run it, and the hooks into your governance framework so that what you build is governed from the first day rather than retrofitted. This is the operating model work I have done for twenty-five years, applied to the one change most operating models were never designed for.
Where the engagement warrants it, I run the portfolio the review creates, as fractional Chief Operating Officer or AI programme lead: a fixed number of days a month on an agreed minimum term, owning the portfolio, the gates and the board reporting, until the firm can own it.
It is not a governance review of the AI already in the firm; that is the Vendor AI Oversight Review, and the two are designed to run together. It is not vendor selection, though the annex tells you which categories to look at; running a selection is the separate Technology Selection service. It is not implementation, though the build follows. It is not legal advice. And no payment is taken from vendors, so I do not advise a firm and the vendor it is considering on the same purchase.
A three-week, fixed-scope review that starts from the operating model you run today, ranks the problems worth solving, maps each to the AI that fits it, values and costs the leaders through three lenses, and sets the budget, the release gates and the sequence. It ends in a rated register, an ROI model you keep, and a board paper.
No. Pilots started without a case are the most common reason a board loses confidence in AI spend. The review puts the pilots already running into the register alongside the candidates nobody considered, and tells you which to fund, which to stop and which to redirect.
No. The review is delivered at board and senior management level and does not assume one. Where a case depends on data or skills the firm does not hold, that is a finding, and it is costed.
The Oversight Review governs the AI already in the firm. This review decides what the firm should build next and what it is worth. Where an Oversight inventory exists it seeds this review, and the register this review produces feeds the governance framework.
The strategy is vendor-neutral. The annex maps tool categories to each use case and names a small number of well-known tools only as illustration, and any selection is subject to your own due diligence, which is a control your firm has to own. No payment from any vendor on a purchase I advise on. Referral and advisory relationships with vendors are declared up front, and the full list is available on request.
Twenty-five years running the operating model, across the whole fund cycle.
I have held the seats that run a fund business, Chief Operating Officer, Chief Information Officer and Chief Technology Officer as well as Chief Revenue Officer, at HSBC Alternative Investments, ONYX Capital Group, Backstop Solutions and an FCA-regulated RegTech platform. At HSBC Alternative Investments I signed off the platforms and service providers behind a business whose assets grew from $200 million to $31.6 billion, which is where the build-or-buy, the outsourcing and oversight and the control model on this page were learned, from the inside, under the regulator. I know where the work actually happens in a fund business and what each step costs, which is why this review starts there rather than with the technology. The AI is the edge on that record, not the substitute for it: the method draws on the Oxford Artificial Intelligence Programme at Saïd Business School and the IAPP AI Governance Professional certification, and on Disruptive Innovation in Financial Services, published in 2026. Through the Vendor AI Oversight Review I write the due diligence questions regulated firms put to their vendors; this review makes sure the firm knows what it wanted from them in the first place.
Use the contact form, or message me on LinkedIn.