AI strategy consulting
AI strategy consulting at Fulkerson Advisors is the work of deciding which AI initiatives a company should fund, in what order, and on what evidence, before the money is committed. It is delivered as the Diagnosis: three weeks, one written verdict.
We work with executives who need an answer for a board that has stopped accepting enthusiasm as a plan. The job is to sort the initiatives into the ones that lead somewhere, the ones that are wishful thinking, and the ones to stop before they use up another quarter.
How it works
A founding partner leads the work personally and sits with your executive team and the people who run the operation. When a question needs deeper industry knowledge than we carry, we bring in a specialist from our network rather than guess. The deliverable is a written verdict, the cost and payoff behind each initiative, and the baseline needed to judge the spend later. We do not produce a deck.
When to call
Most executives come to us in one of four moments. The board has asked for an AI position by the next quarterly review. A first wave of pilots ran for a year and produced demos but nothing the CFO can point to. A competitor has made a visible AI commitment and the question is whether to match it. Or someone inside has proposed a build-or-buy decision that the executive team is not yet equipped to judge. A written point of view with numbers attached resolves these situations, and another vendor pitch does not.
How we do it
We work in a fixed order. We understand the business before we plan, plan before we commit money, and commit before we scale. We start by mapping where AI could plausibly earn money or save it in your business, then test each candidate on four questions: whether the data exists, whether the system's answers can be checked, whether the people will use it, and whether the payoff covers the cost. The result is a ranked list of the initiatives worth funding now, a watchlist to revisit in six months, and the bets we recommend against, with reasons. When the answer is to buy, we read what the vendor's terms let it do with your data. In our study of nearly 3,700 software products, 87% of terms said nothing about training AI on customers' data.
Proof
Strategy work is only credible if the recommendations survived contact with the real operation. The files below are engagements where the plan fed directly into systems we then helped build and measure. Each one started with the same question: of everything we could do with AI, what is worth doing first?
We don't write AI plans for organizations that have not decided whether they want one. We don't produce hundred-page decks to justify a vendor someone already chose. We don't run hackathons or ideation workshops disconnected from the business. And we don't take engagements where the sponsor will not commit to a funding decision at the end. Without a funding decision at the end, the plan sits in a drawer and the same question comes back the next quarter.
A strategy is a decision, and decisions need an owner afterward. When several initiatives are running and nobody is accountable for the whole, the fractional Chief AI Officer engagement is the next step.
Asset Management
Equity index backtesting for a $55B US pension fund
Strategy work that identified, scoped, and validated the backtesting environment the fund now uses to evaluate strategies for a multi-billion-dollar portfolio.
Enterprise Software
GenAI strategy for an enterprise software startup
Build-versus-buy roadmap that stopped two proposed pilots and accelerated funding for the one initiative tied to renewal economics.
Retail (Caribbean conglomerate)
Demand forecasting Center of Excellence
Sequenced strategy from single-category pilot to enterprise-wide forecasting capability, with the operating model and governance to sustain it.
Start with a value map weighted by addressable spend and decision frequency rather than by technical excitement. The candidates worth funding have a measurable economic outcome, data you already control, and an executive willing to own the operational change. Everything else belongs on the watchlist or in the do-not-fund pile. Our first deliverable is usually that ranked list, with the financial logic attached.
It depends on whether the capability is a source of competitive differentiation or table stakes. Buy commodity capabilities (transcription, generic copilots, off-the-shelf agents). Build the layer that encodes your proprietary data, workflows, or institutional judgment. The mistake is applying a blanket posture across the portfolio. We make the call initiative by initiative and write down the reasoning so the next executive can audit it.
Each pilot should earn the right to the next by clearing explicit gating criteria: business impact validated, integration path mapped, change-management owner identified. Pilots that cannot articulate their next stage at kickoff almost never reach it. We design the sequence so the second initiative reuses the data infrastructure, governance, and operating model built for the first, compounding rather than restarting.
Boards get comfortable with numbers. The strategy needs to land with an explicit funding ask, a ranked portfolio showing what gets funded and what doesn't, a financial model tying spend to outcomes, and a governance structure for killing initiatives that miss their gates. We build the materials for that conversation and, when useful, sit with the executive sponsor through the read-out.
Three weeks, as the Diagnosis. Longer only if the scope spans several business units or includes a build-versus-buy decision against a crowded vendor market, and we say so before we start. We scope tightly and prefer to ship a defensible answer to a narrower question than a sprawling answer to everything.
Often, but not always. Roughly two-thirds of our strategy engagements continue into the implementation practice. The other third hand off to internal teams or a chosen vendor with our roadmap as the contract of work. We don't gate the strategy work on a downstream commitment, and we'll tell you when an initiative no longer needs us.
