The Tolerance for Friction Is Gone. Here's What That Means for Enterprise AI

Written by: Ben Durham-Kilcullen

Chief AI Officer, Ondaro

This article first appeared in Fortified Quarterly, Q2 2026.

Ben Durham-Kilcullen, Chief AI Officer of Ondaro, spends a lot of time in rooms where the question “will AI replace me?” is hanging in the air and no one's answered it directly.

In this conversation, he gets into what's actually changed, not just in technology, but in what employees expect from the tools their employers give them. He covers why most AI initiatives miss before they even start, what it takes for platforms like ServiceNow to move from tracking work to actually doing it, and why an AI strategy without a measurement framework is just an expensive experiment.

From your perspective, how is AI fundamentally changing what employees expect from workplace tools and digital experiences?

Employee expectations are eclipsing what the typical organization can facilitate in a scalable and secure fashion. Employees have had access to consumer AI: ChatGPT, Claude, Copilot, and now they're walking into work and asking why their enterprise tools don't work the same way. The expectation has moved from “help me find the right form” to “just handle this for me.” That's a fundamentally different product requirement, and it dramatically shifts their expectations and demands of the business.

What's really changed is the tolerance for friction. People used to accept that enterprise software was cumbersome because it was powerful. That trade-off is largely gone in users' minds. If the AI in their personal life can draft an email, summarize a document, and route a request without them clicking through five screens, they expect the same from the tools their employer provides. Organizations that don't close that gap are no longer managing a UX problem, they're dealing with an adoption and retention problem.

How do you see AI evolving the role of platforms like ServiceNow, especially when it comes to moving employees from simply finding information to actually getting work done?

ServiceNow has always been strong as a system of record, the place where work is tracked. What AI is doing is pushing it toward becoming a system of action. The difference is significant. Finding information means an employee still has to interpret it, decide what to do, and execute. Getting work done means the platform is completing steps on their behalf, within defined guardrails.

What makes ServiceNow particularly well-positioned for this is that it sits at the intersection of workflows across IT, HR, finance, and legal.

AI has the most leverage in platforms where the data, the process, and the decision authority already live in the same place. The challenge and the opportunity is governance. Giving AI the ability to act rather than just inform requires organizations to think carefully about what decisions agents can make autonomously, which ones need human review, and how they audit what happened. That's where the real implementation work is, and it's where a lot of organizations underestimate the effort.

Many organizations are investing in AI, but not all are seeing meaningful impact. What are the most common gaps you're seeing between AI ambition and real employee experience outcomes?

1. There's no measurement framework in place before anything gets built. Organizations launch AI initiatives without establishing what success actually looks like: no baseline, no target metric, no defined business outcome tied to the investment. Before you select a single use case, you should be able to answer: what KPI does this move, by how much, and over what timeframe? That could be time-to-resolve on service requests, employee self-service rate, hours saved per role per week, it doesn't matter which metric, but there has to be one. Without it, you're spending real money on something you will then struggle to prove worked.
2. Use cases get picked because they sound exciting, not because employees need them. Sometimes that is because they want AI to be more appealing, and sometimes that is because they lack the grounding in the user's day-to-day work. If you're automating a task someone performs twice a month, the ROI on adoption effort will likely be minimal. You have to start by understanding frequency: what are people doing constantly, and where is the effort going? Better yet, where could that effort be better applied?
3. The underlying systems and data aren't ready to support AI taking action. Organizations want AI that can take action, but the underlying systems aren't connected, the data isn't clean, and the permissions model wasn't built for automation. You end up with an AI that can answer a question but can't close a ticket, approve a request, or update a record. That can quickly devolve into a relatively expensive search or FAQ function.

For organizations looking to introduce AI into the employee experience, what are the most important foundations they need to have in place to ensure it drives adoption, trust, and measurable value?

As a leader approaching AI enablement, the first topic to tackle is employees' understandable concerns about AI. The question “will this replace me?” is in every room where you introduce AI to employees. If you don't address it directly with a clear, honest answer, you will never get real adoption. People may comply and perform for the demo and then go back to doing things the way they always have. Leadership has to own that conversation. It's critical at this juncture to over-communicate and to reiterate the organization's AI strategy and vision, which brings us to our next point.

Strategy has to come before tooling. Too many organizations acquire licenses and then figure out what to do with them. You need a clear point of view on what AI is going to do for your people, what it won't do, and how you'll know if it's working. That means role-specific use cases, not a general AI assistant, but an AI or AI-enabled tooling that knows what a service desk agent, a project manager, or an HR business partner actually needs.

Training has to match the specificity of the use cases. Generic AI literacy training has its place, fundamentals are valuable, but the fundamentals alone don't drive behavior change. People need to see their role, their tasks, and their workflows in the training content.

Last but not least, leaders have to go first. If direct managers aren't using AI visibly and talking about it openly, employees read that as a signal that it's optional, not serious, or in some cases, not condoned. Even worse, it can cause shadow AI (think shadow IT, but specifically AI tooling) to become pervasive. Leadership behavior sets the adoption ceiling.