LAS VEGAS — Enterprise architects, engineers and CIOs flooded Las Vegas this week to discuss the latest and greatest in enterprise solutions at Gartner's Application Innovation and Business Solutions Summit. Attendees learned about the various new ways to make AI work for you, because it's here and it's not going anywhere — and yet, you've probably been going about it all wrong. Fortunately, if you just make a few small changes, the world could be your oyster.

And yet underneath the relentless pulsation of AI evangelism, anxiety persisted. 

When the opening keynote speakers, Gartner Distinguished Vice President Analyst Jason Wong and Senior Director Analyst Brent Stewart, evoked the dramatic term SaaSpocalypse, they also had to acknowledge the undercurrent of truth. An informal poll revealed that the room was full of people who believe that legacy SaaS applications are on their deathbed. Whether due to replacement by agentic AI platforms or by custom-built apps fueled by vibe coding, an entire industry was perceived to be under siege — even after we were reminded that only 5% of AI pilots reach production.

Related:What Apple's AI update reveals about the future of build vs. buy

The nervousness only grew as the conference continued. 

It slipped into conversations that were meant to be about triumph; during ServiceNow's case study highlight, Plat4mation Vice President of Americas Greg Clock acknowledged that his customers repeatedly tell him they are scared of AI but feel unable to avoid it. Concerns about job security emerged repeatedly throughout the conference, with Deepak Seth, senior director analyst at Gartner, describing AI as culturally anthropomorphized: "it can do what Sandra in HR can do." Even as he reassured the audience that AI agents are not a direct replacement for humans, he warned of technical skills growing obsolete overnight, of the broken talent pipeline, of the C-suite preemptively using AI to replace "fixed costs" — meaning human employees.

AI anxiety as a prompt for human connection

Surprisingly, the undercurrent of anxiety didn't seem unwelcome; in fact, attendees seemed almost relieved to address it. Perhaps the most interactive and engaging point in the entire event was when the audience was asked to directly confront the most extreme "what if?" AI scenarios we could call to mind and share in that panic together.

That talk was "Most Maverick Predictions: The Human Body as the Next Computing Platform," delivered by Frank Buytendijk, distinguished vice president analyst and chief of research at Gartner Futures Lab. Sporting patent duochrome dress shoes and a complementary purple pocket spare, Buytendijk posed to the room a series of possible futures: What if 30% of AI projects fail by 2028 due to "silent subversion" and sabotage? What if by 2030, more than 10,000 people will have been intentionally or unintentionally killed by AI?

Related:Why bank AI projects stall at approval

Clearly, these doomsday questions stirred curiosity among session participants. When the presentation was over, a crowd gathered at the foot of the stage. The audience was desperate to continue the conversation, with many jostling alongside fellow attendees for an opportunity to converse with Buytendijk. Then, instead of waiting their turn to approach the analyst, people began to share with each other. For a group of paranoiacs, there was a surprising amount of laughter. It felt like the most human moment of the three days of sessions. All it took was shared fear — and unrelenting curiosity — about what AI might be doing to our world. 

Maybe everyone just needed to feel solidarity; Buytendijk cited a finding from Atlantic Council that 62% of 357 global strategists and foresight practitioners surveyed expect the world a decade from now to be worse off than it is today.

A much needed AI reality check actually promotes progress

Perhaps it's a sign that we're entering the part of the AI hype cycle that Gartner has coined the "trough of disillusionment," but there was a clear sense of caution in the atmosphere at Caesars Forum. It permeated even the most optimistic of lectures, sometimes emerging in caveats and asides, sometimes blazing across the glowing screens in the form of chilling statistics. Yes, you must use AI, there's no going back, you can't be left behind — but don't go too fast, don't lose the "human-in-the-loop," don't jump into the void without your parachute.

Related:The agentic shift at the Snowflake Summit: Finding a platform's 'right to win'

This warning undertone, the emphasis on vigilance and precision, was not what I expected to absorb so deeply during the three days in Las Vegas. I don't think it's what most attendees expected either, so accustomed are we all to messages of unwavering conviction that AI has all the answers. 

But actually, the experience at Gartner's event was grounding. Once the full reality of AI in the enterprise was accepted — the dangers and the unknowns, as much as the opportunity and the thrill — that's when positive change became tangible. It allowed new technology use cases and strategies to sink in, rooted as they were in a measured approach. As Gartner Senior Analyst George Sellner put it, the best value from AI won't come from "moon shots," but from applying AI where it reduces friction in delivery. They weren't presenting us with the holy grail, but with a toolkit.

And there were plenty of practical, accessible AI applications shared during the summit. Senior Director Analyst Aaron Lord laid out several AI security frameworks for software engineering leads, with a clear Monday morning, 90-day and 12-month action plan. Sellner used F1 pit crew strategy and a Great Ormond Street hospital's case study to bring a detailed AI scaling guide to life. And Peter Vaccarella, global head of solutions consulting at Camunda, made replacing legacy technology feel not just necessary but thrilling as he outlined the company's new agentic platform, ProcessOS.

In a time where AI investment is high and yet return is murky, these Gartner sessions made the path to true ROI feel possible, even accessible. We may have entered the trough of disillusionment, but the signs are there that we'll make it into the next part of the cycle: the slope of enlightenment.

About the Author