Years before he joined ServiceNow, Eugene Chuvyrov was on the front lines of cloud adoption as a senior cloud architect, where he got a front-row seat to companies’ numerous mistakes.

“[Businesses] were ad hoc’ing this stuff,” he says. He saw firsthand how departments across organizations acted unilaterally to move their data to the cloud, built their own cloud control panels, and developed new processes on the fly. “There was never any enterprise wide strategy.”

The results, he says, were years of technical debt, sunk costs, and massive cleanup as companies realized they had overpaid for certain systems and lacked ones they desperately needed.

There are overlaps between the earlier cloud era and the current GenAI one, says Chuvyrov, who now spearheads product development in AI and other technologies at ServiceNow. Deploying GenAI quickly, without burning through resources, requires the entire organization to work together to build a companywide GenAI vision, he says. “The vision is a framework for people to control AI adoption rather than letting it control them.”

As part of that process, Van Over highlights the need for domain expertise within the organization. Companies should hire their own AI experts to evaluate and vet potential GenAI solutions, and there should be an open line of communication between the experts and leadership.

When a business considers a specific GenAI investment, leadership should identify the outcomes they’re looking for, what it’s going to take to achieve those outcomes, and whether the company can afford the financial and ethical risks, says Hughes.

Then leaders should design evaluation metrics by which they pick their AI vendors and communicate their strategy to the company, says Chuvyrov. Afterward, they should revise their strategy as needed for any future adoptions.

Chuvyrov suggests a safe way to get started is to identify low-risk use cases where GenAI can't cause major problems if things go awry. While experimenting, a company can draft a broad GenAI strategy and vision, including a data governance policy, to guide more ambitious implementations. Taking the time to create such policies will also allow for companywide conversations about how to use AI responsibly, put guardrails in place, and mitigate risks to customers, employees, and society at large.

Undertaking such efforts to ensure success doesn’t mean that companies must slow down adoption or otherwise be afraid to take the GenAI leap, says EY’s Kapoor. “I do understand the need to go fast,” she says. “That doesn’t need to stop. [Speed and strategy] can run on parallel tracks.”