We are living through a defining moment in the history of work. A new type of organization has emerged and is being shaped by intelligent agents, on-demand expertise, and new models of human and machine collaboration. In the years ahead, students will study the decisions leaders make right now. The question is whether those future case studies will highlight innovation and foresight, or preventable mistakes and lessons learned too late.

This is not the first time organizations have stood at such a crossroads. During the digitization wave, companies rushed to automate without understanding the long-term implications. I remember working in banking as digital transformation accelerated. Leaders were eager to eliminate Customer Service Representative (CSR) roles, confident that digital tools would close the gap. But customers still needed someone to talk to when the system did not fit their situation, when something felt off, or when reassurance mattered more than a streamlined transaction.

We reduced too many CSR roles too fast and quickly saw the consequences. Wait times increased, customer frustration grew, and the experience inside branches deteriorated, and we had to rehire many of the very roles we had eliminated because we failed to look at the problem holistically and underestimated the complexity that comes from the interplay between technology, customers, and employees. In other words, we focused on the technology, and not enough on the humans.

Today we are witnessing the same pattern in AI adoption, except the stakes are significantly higher. And we already have examples. Klarna, the Swedish fintech company, laid off roughly 700 customer service and operational roles as part of an aggressive AI automation push, only to reverse course as customer complaints surged and service quality declined. The company ultimately acknowledged that the automation had gone too far and began rehiring human workers to restore the service levels customers expected.

Other examples of AI missteps have had reputational consequences that were just as costly. CNET had to pause its AI-written content program after widespread factual errors and plagiarism surfaced, prompting a rebuild of reader trust. Duolingo saw user criticism accelerate after shifting too much of its lesson creation to AI and reducing contractor involvement, a move that affected both course quality and the morale of the people who created the platform’s strongest content.

These examples echo the very same pattern we saw during digitization: a rush toward efficiency followed by a recognition that human expertise still matters deeply.

Every new technology arrives with both opportunity and risk, yet organizations often focus only on the upside. When leaders rush toward efficiency without understanding the tradeoffs, they often overlook the consequences that can ultimately impact customers, employees and business operations.

Rick Nason illustrated this perfectly when he joined me on the Reroute Reflections podcast. He offered a powerful definition of risk: “Risk is the likelihood of both good and bad things happening.” Reminding us that risk, like everything that involves people, lies on a spectrum. AI widens both ends of that spectrum. The potential upside is significant. AI can accelerate analysis, strengthen decision making, solve operational challenges in real time, and relieve employees of repetitive work. In other words, it thrives in partnering with humans on the simple and complicated problems we face.

In my recent conversation with Chinkal Patel, she shared how custom large language models are already transforming R&D functions by summarizing years of product development history in minutes. Organizations like Dow, Wells Fargo, and Estée Lauder are using agents to streamline logistics, customer support, and insights generation, creating meaningful operational gains. These examples show what becomes possible when intelligent systems support human work.

But this kind of progress does not happen by accident. It happens because the organization creates the conditions for AI to work. A key condition is ensuring that AI is introduced into an environment with the culture needed to support it.

This is where many leaders get it wrong. They treat AI as a technology change when it is fundamentally a cultural change. Chinkal captured this sentiment perfectly during our discussion. Traditional technologies are static and predictable. AI is dynamic. It changes daily. It learns and adapts. This level of fluidity requires a culture that is flexible, curious, and committed to continuous learning.

There is a clear gap between aspiration and readiness, yet the tendency is to push forward based on aspiration alone and ignore readiness, and we all know how that story ends.