"The managers weren't really talking about skill gaps anymore… they were uncovering things around friction." -- Michelle Gonzales, West Bend Insurance

People leaders across industries are asking the same question right now: how do we prepare people for AI-enabled work? In this featured in-person keynote address at ELE's Spring Conference, Michelle Gonzales (Head of Talent at West Bend Insurance) and Deepa Kartha (Founder and CEO of CultureRox & Journyz) brought a critical truth to the surface: the real challenge sits one layer deeper. What their cross-organizational innovation sprint revealed is that organizations aren't just facing a skills problem. They're trying to make sense of work itself as it changes faster than anyone can formally redesign it. Workflows have drifted, decision rights are blurring, and the cognitive cost of navigating that gap is accumulating quietly—mostly in silence..

People shaping the work

The live keynote exchange drew on findings from a five-week innovation sprint conducted earlier in 2026. Participants represented a cross-organizational cohort of practitioners working across talent management, learning and development, and organizational development — each bringing direct experience of AI's pressure on their organizations. The sprint was designed and facilitated through a collaboration between West Bend Insurance and CultureRox/Journyz.

The problem underneath the problem

The sprint started with an assumption the field knows well: AI is creating skill gaps, and L&D needs to close them.

Then the questions changed.

Instead of asking managers "what skills do people need?", the sprint asked: "what has become harder to navigate?" and "where are people hesitating?"

The answers were different. Managers stopped describing skill deficits. They started describing something harder to name — friction, ambiguity, inconsistent judgment, and a growing sense that the rules of the work had shifted without anyone officially saying so.

"We started with questions and used the questions to understand the scenario." -- Deepa Kartha, Journyz

That shift in question design changed everything that followed.

The 4 signals that surfaced

What emerged across organizations wasn't a unified skill gap. It was four distinct patterns — each one a signal that work had changed faster than organizations could interpret it.

Risk inflation. The perceived cost of a wrong decision increased — even when actual risk hadn't changed. More escalation. More "just checking" behavior. Slower approvals. Growing dependency on leadership sign-off for calls that used to be made at the front line.

Workaround behavior. When formal workflows stopped fitting reality, people invented unofficial ones. Shadow approvals, undocumented shortcuts, side-channel decisions. A key signal shared during the keynote: people were no longer doing what their job descriptions said they were doing. The work had changed. The official process hadn't caught up.

Uneven adaptation. Some teams moved quickly. Others slowed dramatically — not because of capability differences, but because of structural ones. How a manager interpreted the changes. How clearly expectations had been reset. How much risk tolerance existed on that team. Treating uneven results as a talent problem misses the actual driver entirely.

Decision confidence erosion. Managers began second-guessing decisions they previously made without hesitation. Execution slowed. Escalation inflated. Review cycles multiplied. And beneath all of it, a reframe emerged on stage that stopped people:

"It is not change management. It is slow digestion of change." -- Deepa Kartha, Journyz

The moment that named what nobody had said out loud

One of the sharpest observations from the live interactive presentation wasn't a data point — it was an image.

Managers aren't failing to adapt. They're straddling.

Running old workflows and new ones at the same time. Without clarity on which rules apply where. Without organizational acknowledgment that the cognitive load of that dual reality is real, accumulating, and costly.

"We're kind of here in the old world and we're here in the new world." -- Michelle Gonzales, West Bend Insurance

The output looks the same. The execution reality does not. And most organizations are detecting that gap late — after performance problems have already appeared, after escalation has already inflated, after the workarounds have already become the unofficial standard.

A collective perspective emerged around what this actually requires: not faster training, not another framework. Something closer to organizational permission — to name what's hard, slow down before designing a response, and treat friction as a signal worth exploring rather than a problem to eliminate.

A new possibility for L&D

Here's what the sprint surfaced that most L&D strategies aren't accounting for yet.

L&D practitioners — because of their proximity to people, behavior, and organizational patterns — are already positioned to detect these signals before they become performance problems.

Not by abandoning what they do well. By expanding what they're listening for.

The sprint pointed toward an expanded role: organizational sensing, signal interpretation, behavioral pattern recognition, operational listening. These aren't foreign capabilities for L&D. They're an extended aperture on skills the function already holds — applied earlier, upstream of the intervention.

"We're uniquely positioned in L&D to notice where subtle shifts are in the organization." -- Michelle Gonzales, West Bend Insurance

The question the keynote left open: is L&D being asked to do this work? And if not — should it start without being asked?

---

What to try next

1. Change the first question. Before any upskilling or reskilling initiative, replace "what skills do people need?" with "what has become harder to navigate?" Run that question with five to ten managers. If the answers surface hesitation, inconsistent judgment, or workarounds rather than skill gaps — you're looking at the real problem. That's the early evidence signal.
2. Map the workarounds before designing the intervention. When people are inventing unofficial processes, the official workflow no longer fits reality. Spend two weeks documenting where shadow approvals, duplicated checks, or undocumented shortcuts are happening. That map is more useful than any needs assessment built on stated skill gaps.
3. Build a short discovery sprint before committing to a solution. Five weeks of structured inquiry — open-ended questions, behavioral observation, rapid pattern synthesis — can reframe the entire problem definition. The investment is small. The cost of skipping it is designing the wrong solution at full speed.