IT teams are getting leaner, thanks to AI, and its role in the industry is only growing.

According to Gartner, 25% of IT work will be done by AI alone by 2030; the remaining 75% will be done by humans and augmented by AI.

AI will at least touch all IT work by the end of the decade, the report said.

Right now, it’s playing a significant role in speeding up resolution of user issues and handling larger request volumes without more hiring, according to Kurt Muehmel, head of AI strategy at Dataiku.

‘Human Readiness’

The transition will be rocky for companies that “[treat] AI like pilot projects rather than as production infrastructure,” Muehmel said. They have trouble naming what models their teams use or defining their value, and may even be losing money on their AI because their technical capabilities and business values aren’t aligned.

“What needs to change is building the governance layer first,” Muehmel said. “Companies need to track performance, ensure compliance, and prevent model drift before running hundreds of experiments.”

IT spending is poised to surpass $6 trillion in 2026, and to get value from that spend, not only does the tech need to be ready, but people do, too. It’s what Gartner calls “human readiness,” and transparency plays a significant role in that.

“When business leaders can’t audit recommendations, they won’t scale them. And when IT teams can’t explain how an AI system reached a decision, they won’t rely on it,” Muehmel said. In Dataiku’s most recent data report, 95% of data leaders said they could not fully trace AI decisions end-to-end if they were asked to provide the reasoning to regulators. With “complete lineage tracking, audit trails, and performance monitoring,” people feel more confident harnessing and trusting AI in IT.

CIOs need full visibility to get the maximum value out of their AI, and they need to take the helm of its deployment. Muehmel recommended:

• Finding high-value use cases that leverage the edge of current capabilities.
• Building in governance and monitoring from day one to make production-grade AI.
• Creating architectural systems that are able to manage hundreds of models efficiently.

“Success requires connecting three things that organizations usually keep separate: data that provides context and a competitive advantage, reasoning that combines human expertise with AI capabilities, and a trust layer that enables scaling,” Muehmel said. “Companies achieving results aren’t just buying better models. They’re building systems where technical and business teams work together, where every decision is traceable, and where they maintain control as they scale from tens to thousands of applications.”

The latest buzzword is “agentic.” Not only is ServiceNow investing heavily in the tech, but also about half of the companies surveyed by PwC have broadly or fully adopted agents.

And among the latest and most promising frontiers in agentic is Voice AI. In 2025, two things have become clear, said Alex Levin, CEO of contact-center software company Regal: “Voice AI technology is finally good enough, and companies understand that they need to invest in it.”

Over the past couple of years, the progress in what agents can actually achieve has been immense.

“LLMs are increasingly able to replicate the complex parts of human workflows,” Levin said. “For example, we have a customer that has automated negotiations via Voice AI. That’s something that wasn’t possible two years ago.”

But still, he cautioned against over-promising: “Most AI agent projects fall short in execution.” Plus, he added, “building directly on LLMs is like building on sand.”

What’s important is that companies properly incorporate and take advantage of the tech. That, he says, is how AI agents can succeed.

“Success isn’t binary, it’s iterative, and the teams that understand that are the ones succeeding,” Levin said. He’s seen insurance and healthcare lead the Voice AI charge, taking over support, lead nurturing, and operational calls. AI can reduce cost-to-serve while maintaining the high level of consideration people expect.

And that’s just in the past two years. In the next year, Levin foresees organizations adopting AI agents in droves, just like they did with mobile apps. The agents themselves will have improved latency, cadence, tone and emotion, making them feel more human.

“Over 90% of interactions will be voice as it’s what customers want (when they can’t self-serve) and the cost will be lower than any other channel,” Levin said. “Chatbots as we know them today won’t exist a few years from now. People don’t want to type; they want to pick up the phone and get something done.”

Everyone has their preferences. That doesn’t change when they’re using AI, which is why Amazon is considering ways to personalize the tech for its users.

The Seattle-based company has just filed a patent application for “dynamic language model updates.”

The system would enable a language model to quickly learn about topics that matter to a user at a given moment. First, users would give the system examples of their interests, citing names, terms or topics.

Then, the system would utilize those examples to create a “weight map” that tells the language model which topics and ideas are especially important to the user and should be emphasized.

Using the map, the model is updated and tuned to pay closer attention to the user’s interests and related trending topics when generating answers.

The patent application also mentions “boosting,” which supercharges emphasis on preferred topics. Basically, it helps the system become even better at relaying the most important, pertinent info.

Amazon has been focusing on ways to make AI more beneficial to users, including through a patent for personal data collection. Its Alexa assistant has been a particular priority, though the personalization efforts are being affected by organizational and tech issues.