On the AI question every mid-market leader is being asked

There is a conversation happening in every mid-market board meeting, and it goes something like this.

A board member asks: what are we doing about AI?

The leadership team gives an answer that lands somewhere between defensive and aspirational. Usually: we're looking at it, we've identified some use cases, there's a pilot in the commercial team, we're working with a couple of vendors, the strategy will be clearer by the next meeting.

Everyone nods. The topic moves on. Everyone knows, privately, that the answer was thin. The board members know it. The leadership team knows it. The strategy remains unclear, the pilots remain pilots, and six months later the same conversation happens again with slightly different words.

This is not anyone's fault. The question is genuinely hard, and the conditions in which it is being asked make it harder.

This article is about how to think about AI in a way that produces a grown-up position, rather than a series of defensive answers. It's aimed at the people being asked the question, which is most people accountable for mid-market businesses now.

The first move is to notice what the question is actually asking.

'What are we doing about AI?' is not really a question about AI. It's a question about whether the business is keeping up with a meaningful commercial shift, and whether leadership has thought clearly about what that shift means for them specifically. The board member asking is usually worried about two things at once: that the business is being left behind, and that the business is about to spend a lot of money on something it doesn't understand.

A good answer to the question acknowledges both concerns. It demonstrates that the business has thought clearly about where AI genuinely creates value for this specific business, where it doesn't, and what the business is actually doing, not just considering, in the places where it does.

The answer most businesses give fails on this because it treats AI as a single thing to have a strategy about. It isn't. AI is a set of capabilities that do different things, with different levels of maturity, for different kinds of problems. A credible position starts by disaggregating.

There are at least three different AI conversations going on at once, and they get muddled in most board discussions.

The first is about using AI in products. This is the conversation most relevant to businesses that sell software or digital services. It's about whether the product experience should be enhanced by AI, how, and at what cost to build and run. The answer is sometimes yes, sometimes no, and depends heavily on what the product is for.

The second is about using AI in operations. This is about whether internal work can be done better, faster, or cheaper by applying AI to specific processes. This is usually where the most immediate value sits for mid-market businesses, because the operations of most mid-market businesses contain significant amounts of work that is routine, data-heavy, and human-judgement-light. Good candidates. But it requires the business to be honest about which processes are genuinely like that, and which just look that way from a distance.

The third is about using AI to make better decisions. This is the hardest of the three and the one most often overclaimed. AI can help a business see patterns in its data it couldn't see before. It can synthesise information faster than humans can. But turning this into better decisions requires the underlying data to be in good shape and the decisions themselves to be ones that can be informed rather than replaced. Businesses that skip the first step and try to get AI to make decisions directly usually end up with confident but wrong outputs, which is worse than having no AI at all.

A mid-market leader trying to hold a credible position needs to have a view on all three of these, and that view needs to be specific to the business. Generic AI strategies don't work because AI isn't generic. The answer depends on what the business sells, how it operates, and what data it has.

The second move is to be specific about what 'doing something about AI' actually means in each category.

For most mid-market businesses, the honest answer includes a mix of the following:

In products: probably some careful experimentation in areas where AI genuinely enhances the customer experience, with a clear eye on cost and a willingness to not use AI where it doesn't add value. The pressure to 'AI-enable' products can produce worse products than the non-AI versions. Leaders should be willing to resist this pressure when the evidence says to.

In operations: probably some specific, narrow automation of routine work that clearly fits AI's current capabilities. The goal is to free up time and reduce cost in areas where the work is predictable enough for AI to handle reliably. These projects should have specific numbers attached to them. Time saved, cost reduced, errors reduced. If a proposed AI operation project can't be stated in these terms, it's not ready.

In decisions: probably some investment in getting data in shape first, before layering AI on top. Most mid-market businesses have data quality and data integration issues that would undermine any AI-driven decision support system. The right move is usually to fix these first. It's less exciting than 'doing AI' but it's the thing that makes later AI work actually useful.

Underneath all of this is a discipline most businesses haven't yet developed: saying no to AI projects that aren't ready, aren't specific, or aren't solving a real problem. The pressure to say yes to everything is strong. The businesses that get real value from AI are the ones that say yes to fewer things and invest more seriously in each one.

A credible AI position for a mid-market leader is, in short, a position that can tell the board:

Here is where we've decided AI matters for us, and why. Here is what we're doing about it specifically. Here is what we've decided not to do, and why. Here is how we'll know if it's working. And here is what we'll change if it isn't.

A board hearing this answer sees a business that has thought clearly about a hard question. A board hearing 'we're looking into it' sees a business that hasn't.

If the question is being asked in your business, and if the answer so far has been closer to the second than the first, the work is to get to the first.