There is a quiet crisis playing out inside organizations that have invested heavily in AI. The tools are deployed. The licenses are paid for. The dashboards show activity. Yet when leaders look at actual productivity gains, the numbers tell a different story. Time savings are minimal. Quality improvements are hard to point to. Employees are using AI, but they are not necessarily getting better at their jobs because of it.
This gap between the metrics saying yes and outcomes saying not really is exactly where this conversation begins. The distinction matters because leadership is often reading the dashboards and concluding success, while employees are quietly navigating a technology that was handed to them without enough context, customization, or culture to make it genuinely useful.
To explore why this gap exists, and what organizations can do about it, Kristian is joined by Kelly Pallanti of Northstar PMO, Kimberly Durst of KLYNE, Sid Parak of Medable, Melonie Boone of Boone Management Group, and Tanesia Leflore of RiteFit Enterprise. Together, they bring perspectives from HR consulting, organizational change management, tech leadership, and workforce strategy, making this one of the more grounded and honest conversations on AI adoption currently happening.
What the Numbers Are Actually Telling You
The first theme to emerge from the discussion is what several of the guests call performative adoption. Organizations are deploying AI because they feel they have to, not because they have thought carefully about how it fits into specific workflows. Melanie Boon says the result is activity without impact:
“They’re like plopping these one-size-fits-all AI solutions and expecting employees to be able to use them, but then they’re not really giving direction on the how.”
Kimberly Durst sharpens the point by drawing a distinction between usage and readiness. A high login rate does not mean employees understand what AI is good for. It does not mean they are using it in ways that add value to the business. It means they are complying with an expectation. “Just because the usage numbers are really high and people are attending the trainings doesn’t mean that they actually adopt it,” she says.
“AI is moving really fast. But it’s actually moving faster than the whole clarity around it.”
Tanisha Lafleur puts a finer point on where leadership is getting this wrong. “Leadership prioritizes metrics, and some of these metrics are vanity metrics, over what the organizational culture is,” she says.
“If the organization does not feel that psychological safety, where employees feel like they cannot speak up, then you won’t hear that voice. That gap becomes inevitable.”
The silence that leaders sometimes interpret as smooth adoption is, in many cases, a sign that employees do not feel safe raising concerns.
The situation places middle managers in an especially difficult position. They hear from senior leadership that AI deployment is going well. They hear from their teams that it is adding friction, not removing it. “They’re hearing it from the senior leaders who are like, implement, this is great, we’re going to make trillions tomorrow,” says Melanie Boon. “And then their teams are saying, you’re asking me to implement this thing and I don’t even know what this AI can do.” That structural squeeze, without permission to communicate upward honestly, is where adoption stalls.
What Actually Works
The conversation shifts decisively when the guests turn to remedies, and the consensus is clear: the solution begins with culture, not tools. Kimberly Durst argues that organizations should stop chasing broad deployment and instead go deep on a single use case. “Pick one task that people already do every week and go deep on that,” she says. “Not a big abstract AI program, but one concrete use case.” That focused approach generates better signals about how employees are actually adapting and what genuine value looks like.
Tanisha Lafleur pushes the case for what she calls an evidence-based approach to readiness. Before deployment, organizations should run a baseline assessment, identify internal champions who can support skeptics, and establish clear guardrails. “Stop looking only at the dashboards and start asking your people if technology is actually making their work better,” she says. That shift, from compliance monitoring to genuine dialogue, is what separates organizations that succeed with AI from those that generate impressive activity reports and little else.
Sid Parak describes the model that has worked for his teams at Medible. Rather than surveillance and top-down mandates, he creates what he calls a fertile ground for experimentation. Employees are encouraged to try things, build things, and fail without consequence. Lunch-and-learn sessions let people show what they have built, no matter how small, which builds collective confidence rather than isolating individual high performers. Parak explains:
“I make it very clear that I’m optimizing for value, not the number of tokens you’re using every day.”
Kelly Palanta closes the loop on the HR angle, arguing that the function has an opportunity to take a genuine leadership role in AI adoption if it is willing to move beyond compliance and toward integration. One practical step she advocates for is embedding AI expectations into job descriptions, something almost no organization is currently doing. More broadly, she recommends building cross-functional AI councils that bring together finance, HR, IT, and senior leadership, because no single department has the full picture.
“It won’t be one particular department,” she says. “It really is those who are champions for it.”
Closing the Gap Requires a Shift in What You Measure
The through line connecting every insight in this conversation is that the AI adoption problem is fundamentally a culture and communication problem, not a technology problem. The tools exist. The capability is there. What is missing is the organizational infrastructure to deploy that capability honestly, safely, and in ways that employees can actually use.
A 2025 Gallup study on workplace AI adoption found that the single largest barrier to engagement is an unclear use case or value proposition. That finding maps directly onto what this panel describes. Organizations are handing people software without telling them why it matters for their specific role, what they should and should not put into it, and where their own judgment still leads. Until that context exists, usage metrics will continue to look good and outcomes will continue to disappoint.
The practical message for leaders is straightforward. Stop measuring activity and start measuring value. Create the psychological safety for employees to be honest about what is working and what is not. Empower middle managers to be conduits rather than compliance officers. And as Sid Parak puts it, if even one workflow can be modified to make someone’s life easier, focus on that. The noise outside is loud, but the signal inside your own organization is where the real answer lives.
