In the span of just a few months, the experience of building software has fundamentally changed. Engineers are now working alongside agents that write, refactor, test, and even reason through complex codebases. Tasks that once required days of focused effort can now be completed in hours or even minutes.
When the core mechanics of how software is built shift this quickly, it matters beyond engineering. Software development has historically been one of the most complex and cognitively demanding forms of knowledge work. If AI can meaningfully compress and reshape that process, it’s reasonable to expect similar effects in other domains that are built on analysis, writing, coordination, and decision-making.
That expectation is exactly what I hear in conversations with business leaders. Having watched what’s happening in engineering, they recognize the acceleration and feel pressure to respond. They purchase tools, launch pilots, and assemble AI task forces in anticipation of similar gains across the organization. Yet despite the tools being available, the organization itself often doesn’t fundamentally change.
I believe that the problem is rarely the technology. It’s time.
Most employees are operating at capacity. When calendars are full and expectations are unchanged, experimenting with AI becomes something people are supposed to figure out on their own. Under those conditions, adoption will always be uneven and slow.
If organizations want real returns from AI, the first move is simpler than most strategy decks suggest. Leaders need to create deliberate space for experimentation, protected time where employees can explore how these tools reshape the work they already do.
AI Must Be Experienced, Not Announced
AI can’t be introduced the way organizations typically roll out new software. It’s not a system employees log into occasionally, nor a feature owned by a small innovation team experimenting on the side. If it remains isolated, it will never reshape how the company operates. For AI to matter, it has to become part of how everyday work gets done.
That integration doesn’t begin with a memo; it begins with discovery, and discovery requires protected time to explore.
In the organizations where I’ve seen this work, leaders make participation mandatory, not in terms of outcomes, but in terms of engagement. They block the time, get teams in a room, and treat it like a focused hackathon centered on real workflows. There is structure and direction, often tied to priority business problems, but within that frame, people are given freedom to experiment. It’s applied play, using AI to draft proposals, analyze reports, review contracts, write code, or prepare presentations, and see what actually changes.
There’s a parallel here to childhood development. Decades of research show that play is the primary mechanism through which children develop critical life skills. A 2018 clinical report from the American Academy of Pediatrics concluded that play develops executive function, self-regulation, language, math cognition, and social skills. Play isn’t a break from learning; it is learning.
The same dynamic applies inside organizations. Through this kind of structured experimentation, employees build their own mental models. They learn by trying, breaking, adjusting, and sharing. No outside consultant can map AI capabilities to someone’s job as effectively as the person doing the work.
Building a space where teams can openly share how they solved their problems can help others learn and encourage them to try these tools themselves. At Confluent, our engineering team has a Slack channel where engineers show how they use popular AI tools, including Claude Code and Cursor, to reduce operational tasks, minimize unnecessary time spent on investigations, and solve incidents faster. Since starting this channel, in four weeks, Claude Code usage in engineering increased from 10% to 81%, showing the advantages of the technology.
Company-led training paired with hackathons or workshops can also help increase AI usage and uncover new use cases for the technology. For example, Confluent hosted two AI trainings before a workshop for marketers to increase usage of Gemini, Glean, and Notebook LLM. The hackathon produced many ideas to solve challenges with lead scoring, customer stories, creating campaign briefs, and calculating the total cost of ownership. In two months, Glean usage increased by 32% with more than 89% of marketers using the technology, and Gemini adoption by 3%, with 83% of marketers actively using it.
Without dedicated time, only the most enthusiastic employees will engage deeply. The rest aren’t resisting; they’re simply overloaded. When you’re trying to survive Tuesday with a full calendar, sticking with familiar methods feels faster than learning a new one. But the “aha” moment, the realization that this fundamentally changes how you work, only comes through firsthand experience, and that experience has to be made possible.
Why This is a Leadership Issue
This is ultimately a leadership issue.
When organizations deliberately block even a few hours, bring teams together, and remove performance pressure so people can apply AI to real work, a predictable pattern emerges. Initial skepticism softens and even fear starts to turn into curiosity. Teams begin identifying use cases that leadership never would have surfaced in a strategy session. Informal knowledge networks form as people share what worked and what didn’t. Over time, those small shifts compound.
There is strong precedent for this in business.
3M’s famous 15% time policy, giving engineers protected time to explore projects of their own choosing, produced the Post-it Note, one of the company’s most successful products. Google’s 20% time yielded Gmail, Google News, and AdSense. What the play research tells us is that this isn’t just a nice management practice; it actually reflects something fundamental about how humans learn and innovate.
When applied to AI, this kind of experimentation has another important effect. It distributes capability across the organization; everyone becomes collectively more fluent. Speed increases not because AI is magical, but because employees learn how to shape it around their own context. That is where meaningful gains, sometimes multiples of previous output, begin to show up. It’s a cultural shift in the business.
None of this happens accidentally; it requires leaders to send a clear signal.
Bottom line, it’s a choice about priorities. When leaders protect time for experimentation, participate themselves, and normalize learning in public, they communicate that this is real work, not a side project. They make it safe to try and experiment.
Without that signal, AI remains an optional science experiment, and optional initiatives rarely transform how an organization operates.
The Competitive Implication
The companies that pull ahead won’t be the ones with the largest AI budgets or the most tool licenses. They’ll be the ones where AI is woven into daily work, where people share what they learn across teams, and where adaptation happens faster than formal strategy cycles.
In the early days of the internet, the winners weren’t simply the firms that bought websites. They were the ones who rethought distribution, marketing, and customer engagement around a connected world. Similarly, with cloud computing, the advantage didn’t come from signing a contract with a provider; it came from redesigning architecture and operating models to move faster.
AI will follow the same path, but its reach is broader. Unlike prior waves that reshaped distribution or infrastructure, AI touches the work itself. It changes how decisions are made, how content is created, how software is built, and how knowledge flows inside the organization. It is likely to be the most consequential digital shift in history.
And yet the research on human development suggests the response shouldn’t be more strategy, it should be more play. The Harvard Center on the Developing Child has shown that the executive function skills built through play in childhood, like working memory, cognitive flexibility, and inhibitory control, are stronger predictors of academic and life success than IQ. In the same way, the organizations that will lead in the AI era won’t be the ones with the highest-IQ strategy decks. They’ll be the ones who built the adaptive capacity to learn, pivot, and integrate new tools through deliberate, structured experimentation.
I don’t believe the competitive edge will come from access to technology. It will come from how quickly an organization learns to use it and how deliberately leaders create the conditions for that learning to happen.
