Discerning companies and startups with true AI from those who are still pretenders is a key skill in the evolving world of IT due diligence and tech dealmaking. Photographer: Kiyoshi Ota/Bloomberg

Claiming some level of fluency with AI methods has become commonplace for startups and tech companies of all stripes, whether they’re peddling SaaS, hardware, ad-tech or the latest social network. But the claims often fall short of reality.

As AI becomes a larger piece of IP in tech deals going forward, whether the buyer is a private equity firm or a strategic acquirer, the need for technology due diligence around AI methods and models continually increases.

I’ve seen a significant uptick just during the last several months from clients seeking to understand how good a target acquisition’s AI is compared with competitors and what best practices in the larger AI space look like.

Gaining visibility into a company’s AI competency can be paramount to feeling good about an acquisition bid and having the confidence to remain active in a competitive sales process.

The task here wouldn’t be so challenging if there weren’t so many different things represented as AI within the market. Some companies can get very liberal in what they describe as AI.

Hiring a couple of data scientists and setting up a data warehouse is more than many companies manage to do, but it’s a long way from knitting machine learning and AI into a product or using it to bring true insights to a business. There is a large gap between companies who have achieved some mastery of AI techniques leading to incremental productivity and the rest who have not.

Parsing those differences, being able to tell whose AI is likely good and legitimately implemented from those whose isn’t can be hard during a single conversation or product demo.

But there are tells. There exist some key questions that can be asked that can very quickly give a clearer picture of how well and how far a company’s efforts have gone.