Tackling the AI market: more than just a product

Photo credit: Pavel Tancibok via Unsplash

My Dad used to enjoy buying things. Or, to be more accurate, he enjoyed the process of getting ready to buy things.

If we needed a new washing machine, or a microwave, or, best of all, a car, he would break out his graph paper and pencil, and a big pile of Which? magazines and other literature. He would make a list of criteria, and he would go through all of the candidate products, scoring them out of ten, adding the scores up, and adjusting them if he wasn’t convinced by the answer. And then we would tour the shops, interrogating the sales staff to a level of detail they were not always prepared for.

We didn’t always end up buying something. Sometimes the prospective purchase was beyond our means, and sometimes nothing quite hit that perfect combination of features. But my Dad enjoyed the process nevertheless, often more than he enjoyed the ownership of another appliance or gadget.

This was back in the seventies and eighties, when consumer information came in printed magazines, analysis was conducted on paper, and goods were bought from physical shops. It might seem quaint and outdated today, the process of a hobbyist rather than a serious buyer.

But, even though the tools have changed, much technology purchasing still follows a remarkably similar process. Business sponsors want a particular capability (and have often been convinced that they need a particular set of tools). The technology team invites them to express their needs as a set of requirements. The procurement team asks for those requirements to be converted into assessment criteria. The RFP is launched, the bids are assessed and the product is purchased.

Unfortunately, while this approach may have served to intrigue and entertain my Dad when he was buying household goods in the last century, it is a mistake when buying technology in 2026. And it is even more of a mistake when buying AI products and services.

This is because technology (and AI) products stopped resembling washing machines and microwaves at least ten years ago. They are no longer static assets with features fixed at the time of purchase: they are shifting capabilities fed by a continuous pipeline of change. It used to be possible to pretend that they were fixed assets, by installing them on-premise and failing to apply upgrades, but the dominance of software-as-a-service (or all of the other variants of something-as-a-service) has made that impossible.

That means that, when you make a technology or AI purchase, you are, whether you like it or not, entering into a relationship with a partner who could either serve you well or serve you badly in the coming years. They could continue to invest in their product, adding features and expanding capabilities. They could optimise their own operation, reducing their prices while increasing quality. Or they could abandon their product, change their terms, or hike their prices. And, even if you grow to trust your partner, they could be acquired by a company with different goals and values.

This means that, if you aspire to become an enterprise AI leader (or the leader of an enterprise which is dependent on technology - that is, all enterprises), you can’t carry on buying technology as if you are buying washing machines, and you can’t leave it to your Chief Information Officer or Chief Procurement Officer (even if you have made them integral parts of your leadership team).

Instead, you must learn to make conscious choices about the companies you are going to partner with. Do you understand their business models? Are they making a profit, or are they still burning through their investors’ capital? What will happen when that capital runs out? Are their prices likely to remain stable? Is what you get for those prices likely to remain stable? What assurances can they give you about the quality of their service or the privacy of their data? Are those assurances good intentions, or are they actually written into your contract? Do they have the competency and capability to make good on those assurances?

You may end up making a list of criteria similar to those made by my Dad. (If you are feeling paranoid, you may even want to write them down on paper, rather than asking your favourite chatbot to come up with them). But those criteria shouldn’t just capture formal features such as parameters, performance, accuracy and pricing: they should capture features that don’t appear in consumer reviews, such as trust, purpose, vision and leadership.

Being an enterprise AI leader means betting the productivity, integrity, stability, security and quality of your enterprise on the tools you pick and the companies which provide them: you had better put the work in to understand them and their motivations.