It’s official: AI is no longer on the periphery. For many organizations, it's now a central part of how work gets done.
In our recent study From Pilot to Production: The State of AI Adoption in 2025, we worked with Zogby Analytics to survey 1,000 business leaders across the United States and Canada. The findings are clear: AI has moved out of the lab. It's being used to solve real problems, in real time, at meaningful scale.
According to the data, 68% of respondents say their organizations are already using AI in live production environments, and most of those systems have been running for at least six months, with 43% reporting plans to expand their use of AI over the next year.
So yes, AI adoption is picking up. But implementation isn't without its challenges.
As companies shift from planning to production, new pressures are surfacing. Nearly 70% report that at least one AI project is behind schedule. The reasons are familiar: incomplete or inconsistent data, difficulty connecting with older systems, and the complexity that comes with training and tuning customized models.
Even among the 86% of organizations that are actively adjusting their AI models, more than half say the process has been more involved than they expected.
In response, many are revisiting how and where they deploy their systems.
Two-thirds of respondents say they're planning to move some AI workloads off the public cloud in the year ahead. Whether that means shifting to hybrid environments or bringing workloads full on-premises, the motivation is similar: more control, more consistent performance, and greater assurance around data security.
Despite the technical and logistical hurdles, confidence in AI remains strong. Most organizations say they have the right governance in place, their data is protected, and their teams are managing AI with intent. The majority have established formal leadership roles - 86% now have a Chief AI Officer - and 91% say they can track how data moves through their systems from beginning to end.
But confidence isn't just about performance. It's about being able to explain how things work, and prove that they are working as expected.
That's where we focus our efforts.
Prove AI was built to give teams a clear line of sight into how their systems behave over time. It helps identify drift early, enforce internal standards, and build trust - both with stakeholders and within the business itself.
AI is now part of the operational core. It's being used to assist in software development, detect fraud, automate decisions, and improve processes across departments. But as deployment scales, so do the risks of poor oversight. Without the right guardrails, even well-designed systems can veer off course.
The takeaway from this report is straightforward: organizations are moving quickly, and expectations are high. But real, sustained success will depend on the quality of oversight that supports that progress.