7 Ways to Ease AI Adoption for Your Enterprise Customers

Pip Coburn was a well-known global technology strategist at UBS, a financial services firm with presence around the world, during the technology boom of the late 1990s and through the early 2000s. From his experience listening to scores of startup executive teams tout how they were going to change the world, and then monitoring startups that actually succeeded versus the vast majority that failed, Coburn came up with a seemingly simple-sounding theory. He called it the Change Function. His theory postulates that people - either as individual consumers of technology or as enterprise decision makers and users of it - adopt a new solution only if the pain of their current situation exceeds their perceived pain of adopting a new solution.

Change = function (level of current crisis, perceived pain of adoption of potential solution)

Over the last couple of years, many smart observers have written insightful articles about the pain (or challenge) of AI adoption in the enterprise and ways to reduce it. However, most of their advice, though useful and well-intended, seems to be focused on what companies intending to integrate AI into their operations need to either do better or differently. From my experience working with several companies that have successfully integrated AI into various aspects of their operations, here are seven ways in which AI vendors can proactively help ease AI adoption for their enterprise customers:

1. Verify that your prospective customer is operationally ready for AI: It is important that an AI vendor does independent diligence on a prospective customer’s operational readiness for AI. This includes, at a minimum, verifying that there is a real business problem that needs to be solved, the relevant data is available, there is a planned budget and timeline to conduct a pilot, and executive sponsorship exists for potential deployment at scale if the pilot is successful. In the absence of any one of these crucial factors, the stark reality is that the AI initiative will unfortunately have a very low chance of success.
2. Build strong industry and problem-specific expertise: Numerous customers have lamented to me over the past few years about the incredible amount of time their teams had spent to educate their previous AI vendor(s) on their industry, the specific problem(s) and its nuances. I have even heard of many a situation where an AI vendor delivered a successful pilot, but getting there took so much of the customer team's bandwidth that the customer concluded it could not afford to take any more time to get the AI pilot into production.
3. Clearly understand customer workflows relevant to the problem: For an AI solution to be adopted, it must become an integral part of a customer’s new decision-making process. Understanding a customer’s current workflows relevant to the problem being solved can help the AI vendor focus its change management efforts to ensure that the right technical and organizational integrations are in place. Without these, the customer may not be able to derive any value at all from the AI solution.
4. Don’t ignore the critical importance of user interface & experience (UI/UX): It is Product Management 101, but still useful to re-emphasize here that AI vendors need to have a very clear understanding of the users of their AI solution. For example, if the recommendation/prediction made by an AI solution seems ambiguous or unclear, there is a real risk that it gets ignored or, even worse, misinterpreted by the users. This can potentially undermine the entire initiative. In contrast, a well-thought out UI/UX can add further value and help make the AI solution sticky.
5. Minimize post-deployment “operational hassle” to your customer: While advancements in AI are happening at a fast pace, these systems are still rudimentary and fragile in many ways. This implies that teams at your customer might sometimes have to take on additional tasks (that have to be done by a human) in order to keep the environment of the AI solution within the optimal range of operations. An AI vendor needs to be sensitive to the “operational hassle” its solution imposes on the customer in order to keep it running smoothly. It might even be advisable to de-prioritize these high “operational hassle” use cases in favor of others that have a lower “operational hassle” associated with them, especially if the customer is just starting its AI transformation journey.
6. Educate your customer on benefits of human-in-the-loop AI solutions: Customers often (wrongly) expect a state of perfect nirvana (e.g., 100% accuracy/precision) when they apply AI to solve a business problem. However, human-in-the-loop solutions can enable AI vendors to avoid spending a disproportionately large amount of time and resources upfront in resolving the long tail of edge cases. Over time, the human involvement can reduce as the AI solution gradually gets better at handling more edge cases. It is the AI vendor’s responsibility to help the customer realize how much better they already are with a human-in-the-loop AI solution compared to before.
7. Proactively help align customer’s internal stakeholders: Since we are still in the early stages of enterprise AI adoption, customers may not yet understand the different internal stakeholders, who need to be actively involved for successful deployment of an AI solution. It is the AI vendor’s responsibility to help identify customer's internal stakeholders and start aligning their (sometimes divergent) interests as early as the pilot phase itself. This is a non-trivial task. For example, in one company I have worked with, we had to align their teams in automation, quality, production, ergonomics, human relations, communications, legal, and procurement before we got the initial green-light for global deployment of our AI solution.

CB Insights' research indicates that 4300+ AI startups have raised ~$83B in funding since 2014. As a consequence, there are often several AI startups trying to solve similar problems. For example, a manufacturing company COO recently shared that he had been pitched by a couple dozen startups over the past nine months, and all of them were focused on predictive maintenance of factory assets. In highly competitive environments like these, the seven ways described above can enable an AI vendor to clearly differentiate itself from the pack and help its enterprise customers ease the pain of AI adoption. If you think about it, you'll see that each of the ways described above contributes to reducing the time to value of the AI solution for your enterprise customer. Thus, a good proxy for testing if you are indeed helping reduce the pain of AI adoption is measuring the time to value for your enterprise customers. In fact, I'd recommend that executive teams at AI vendors ought to always be asking themselves, “What else should we be doing to reduce the time to value for our enterprise customers?”