When do you really need
an AI strategist?

In which situations does it pay to hire an AI strategist?

Hiring an AI strategist pays off in four situations: an expensive decision that's hard to reverse, an internal deadlock, an investment that has to be explained to a board or regulator, and a first-time project under high time pressure. For technical and narrow questions, small reversible decisions, an internal lead with the mandate, or a decision that's already been made, outside help adds little value. For most organisations, a well-defined engagement or a part-time role of a few days a month fits best.

What does an AI strategist add when choosing a small, monthly-cancellable tool?

Little: for a small, reversible decision, an AI strategist is outside help that adds little value, because the time spent weighing it up costs more than the experiment itself. The advice is to simply try the tool and save your analytical effort for decisions you can't easily undo.

Is it worth hiring an AI strategist to reinforce a decision that's already been made?

No. Hiring an AI strategist to reinforce a chosen direction is wasted effort, because nothing new comes out of it and every dissenting finding gets smoothed away. An independent case for a board or regulator is different from buying legitimacy, because there the outcome is still genuinely open.

What size of engagement for an AI strategist usually fits?

For most organisations, a well-defined AI strategist engagement of two to six weeks fits best, or a part-time role of a few days a month. An engagement that runs longer without interim decision points is rarely in the client's interest.

Source: article by Dave Willemstein. Updated: October 2026.

First: when you're better off doing it yourself

There are four situations where outside help adds little value, and where I refer elsewhere or step back myself.

1. The question is technical and narrow

"Which model works best for our document classification?" is a question for your own engineers or a specialist builder. A strategist who answers that is repeating what your team already knows, at a higher rate.

2. The decision is small and reversible

A hundred-euro-a-month tool for one team, cancellable monthly? Just try it. The time spent weighing it up costs more than the experiment itself. Save your analytical effort for decisions you can't easily undo.

3. You already have someone who can do this and has the mandate

If there's an internal lead with experience, budget and access to leadership, an outside adviser is usually an extra layer rather than an accelerator. Give that person room instead of a sparring partner they have to bring up to speed.

4. The decision has already been made

This is the most awkward one. Sometimes advice gets bought in to reinforce a chosen direction towards the board or shareholders. That's wasted effort: for the client, because nothing new comes out of it, and for the adviser, because every dissenting finding gets smoothed away. If I recognise this pattern, I say so. Sometimes a real question turns out to be hiding underneath after all.

When outside help does pay off

1. The decision is expensive and hard to reverse

A multi-year platform contract, an architecture choice that ten applications will depend on, a build-or-buy decision for something that touches your core process. Here, the cost of a wrong choice is a multiple of the advisory investment.

A rule of thumb: if the wrong choice costs more than ten times the advisory budget, and you're not sure, a second opinion is a rational form of insurance.

2. There's an internal deadlock

Two departments with different preferences, both with good arguments, for months. Here, an outsider's value isn't primarily their knowledge, it's their lack of a stake. Someone who'll be gone again next month can reach a conclusion that would be politically loaded internally.

This only works if the brief and the criteria are accepted by both sides in advance. Otherwise you shift the conflict onto the quality of the advice instead.

3. You have to explain something to a board or regulator

If you need to defend an investment to people who aren't involved in it day to day, an independent case helps. Not because your analysis is worse, but because an external check shifts the discussion from "do we believe this person" to "does this reasoning hold up."

Note that this is different from buying legitimacy. The difference: here, the outcome is still genuinely open.

4. You've never done this before and time is tight

The first time an organisation runs an AI selection, sets up an AI governance structure, or prioritises an AI portfolio, it takes two to three times longer than the second time. Someone who's done it before buys you out of that learning curve. That's a legitimate reason, provided you keep the knowledge afterwards.

What to ask before you hire anyone

• What's the smallest engagement we can start with? A good adviser offers a limited-scope way in. Anyone who proposes a six-month programme straight away is selling capacity.
• When are you done? There should be an answer that isn't "it depends."
• What are we left with once you're gone? Documents are the minimum. Frameworks you can reuse yourself are better.
• Which part of this engagement could we do ourselves? The answer to this says more about someone's integrity than any reference.
• What interests do you have? Partnerships, commissions, implementation work on the side. Ask explicitly, and ask for it in writing.

The size that usually fits

For most organisations, a well-defined engagement of two to six weeks fits best, or a part-time role of a few days a month. Engagements that run longer without interim decision points are rarely in the client's interest.

And if you're not sure whether you need this: try writing the question yourself first, on one page, with the options and the criteria. If you manage that, there's a good chance you can answer it yourself too. If you can't, you know exactly what a conversation should be about. And that's a good starting point, whether or not you end up hiring someone.