Making AI work requires careful budgets

For many companies, AI is more essential every day—but integrating the technology into an organization’s stack requires a level of budgetary ninjitsu that’s not always straightforward for IT teams.

Demands on AI can come from anywhere: leadership, staff, users, and company boards. What they all have in common is a desire to see more of the technology within the organization’s day-to-day, integrated as smoothly as possible.

IT teams and IT leaders tasked with this integration are in a sticky spot. On the one hand, they’re facing reasonable demands to deploy the technology; on the other, spend is constrained, and teams are asked to do more with less.

Determining factors. Basel Kayyali, who co-leads McKinsey’s technology practice, told IT Brew at this year’s MIT Sloan CIO Symposium in Boston that it’s critical for IT teams to define what they want out of AI transformation.

“What we’ve seen, over and over again, is ‘Let one thousand flowers bloom’ results in pilot purgatory,” Kayyali said. “You’re better off picking one or two things—what we call domains or end-to-end workflows—that are going to drive one of three things: growth, productivity, or new businesses…it should ideally be linked to your business strategy.”

As with most things in business, you have to spend a little to (hopefully) make a lot. For companies like AI investment platform Brightwave, CEO Mike Conover told us, leverage is high for developing task applications—which, when combined with downside risk minimization, can result in payoff that sells itself.

You just have to make sure you’re playing it the right way and not using AI for its own sake. Education is the key. “You’re wasting other budgetary dollars because your team is underexposed to the strengths and limits of the technology,” Conover said.

Financial practice. A good way to approach AI integration is to measure how it can be deployed in a cost-efficient way. Controlling costs is key, Jonathan Kleiman, SVP of enterprise AI at StackAI, told IT Brew at the symposium: impacts of cost and revenue are an essential part of determining how much to budget for, and what can be offset by revenue.

“Identifying the areas that can drive the most revenue, or that can lower the significant cost reductions—once you identify those areas, [you’re going to understand] what are the processes, how much time each process takes,” Kleiman said, adding: “I would start with the use cases that are easy to build, that you can ship quickly, that you can deploy quickly, and that are gonna have significant business impact, and then move on later to the use cases that are harder.”

Ultimately pushing the responsibility for AI budgeting to IT teams and even senior tech executives is a faulty premise, Kayyali told IT Brew, because it assigns blame to the people who are trying to balance integration with demands from the business side.

“It’s not the CIOs’ problem, it’s the business’s problem,” Kayyali said.