The Water Cooler is Closed
- Jun 30
- 5 min read
The New Isolation of Organisations Successfully Working with AI
It’s Monday. You are sitting down in front of your laptop, head buzzing with ideas. You finally feel like you have gotten the hang of tools you have been experimenting with for the last few weeks. You know what to expect from Lovable, you prototyped a few nice UIs using Figma AIs, Rovo has made it easier to pick up new tasks from the Jira board. Slowly, the CLAUDE.md and copilot-instruction.md are getting populated with genuinely useful guidelines and expectations. Spinning up agents, using skills, checking-in on progress and monitoring the app, as it slowly grows in sophistication and usefulness, has become a familiar pattern. The week is going well, you feel productive, you solve real problems and make progress like never before.
You should feel excited, energised, motivated and yet… Friday comes and you feel like you could probably just keep going. It turns out Claude doesn’t have any weekend plans. Copilot isn’t going shopping to try out a new recipe it got from its friends and Codex hasn’t been to the exhibition that opened downtown last week. It finally hits you and the realization is stark: all your “colleagues”, Lovable, Rovo, Claude, Copilot, and Codex, are infinitely capable collaborators, but they aren't your peers. There are no water cooler conversations about the exhibition downtown or sharing a recipe that didn't quite work. Your progress is purely technical, a constant, efficient loop of creating and testing, uninterrupted by coffee breaks or team lunches. You are achieving more than you ever thought possible as an individual, pushing out features and iterating on designs at a blistering pace. But where did the “we” go?
The way most organizations deploy AI today is optimizing for individual output. This is true whether you are building software, closing deals, running campaigns or analysing markets. The tools and the mission differ but the pattern doesn't. One person’s superpower quietly becomes everyone’s habit. What used to require collaboration can quietly dissolve into individual tasks. Stand-ups and weekly meetings disappeared not because the team got better at communicating, but because there was nothing left to communicate. Anthropic’s own internal study points in the same direction, as one employee admitted: "I like working with people and it's sad that I 'need' them less now… More junior people don't come to me with questions as often." The texture of the working day changes. The gain for an individual is real, visible and desirable. Supported by leadership mandates and hyped up in the corporate narrative. The collective cost is hidden.
Organisations that chase innovation and growth can inadvertently dismantle the conditions that make both possible. The ideas that travel furthest inside a company rarely come from close teammates, they come from the accidental conversation with someone from a different team, a different floor, a different discipline. That's what the watercooler was actually for. How might we get everyone talking if most of the talking now happens with the large language models?
One vision of AI promises that we no longer need people thinking together to get smarter — the technology fills the gap. But when we think together we don’t just think better, we think differently. A landmark field experiment at Procter & Gamble revealed the most consequential dynamic: AI-augmented individuals performed at a level comparable to two-person human teams. This creates enormous organizational pressure to reduce team sizes and bypass collaboration. However, and this is critical, full human teams augmented with AI produced the best breakthrough (top 10%) ideas.
What this study suggests is that instead of framing the new AI tools as “individual production technology” we should deploy it as “coordination technology” which can enhance collective intelligence and amplify innovation. Currently, most AI deployments overwhelmingly favor the former paradigm and optimizes the individual at the expense of the collective and leaving the larger gain in innovation and productivity on the table.
The ideas that travel furthest inside a company rarely come from close teammates, they come from the accidental conversation with someone from a different team, a different floor, a different discipline.
But that lost innovation is only the visible cost. When people stop thinking together, they lose something more fundamental still - their emotional wellbeing. A 2023 study found evidence linking more frequent interaction with AI at work with “higher levels of loneliness, insomnia, and even after-work drinking.” Deploying AI as an “individual production technology” doesn't necessarily mean freeing up time for social connection — its internal logic of “productivity enhancement” naturally inclines one to intensify work.
We watch the news fill up with stories of large organisations “downsizing” due to expected AI efficiency improvements. This eats away at our confidence, leaving us simultaneously more anxious and more determined to prove our worth. Employees work at a faster pace, take on broader scope, and extend work into more hours, often without being asked. Left unchecked, this curdles into workload creep, cognitive fatigue, burnout, and weakened decision-making. It leaves even less bandwidth for the human interactions that sustain organizational health. Run too long, this dynamic hurts people and makes organisations brittle. Lonely, anxious employees don't take risks, don't challenge decisions, don’t come up with novel ideas and don't stick around. It doesn’t have to be that way.
The AI tools are here to stay and they have a lot to offer. As we gain something new, we must pay attention not to lose what was there before. It is our choice. We can look at the new capabilities as superhuman tools of individual production. We could also consider them a real augmentation of how organisations actually work — woven with people, their interactions, the weak ties that fuel shared cognition, unexpected discoveries and true innovation. It requires a broader view, a deliberate attention to the design of work, to the strengthening of social glue. We must be intentional about creating a rhythm that keeps people in contact with each other. A rhythm that combines deep focus with genuine human contact. For leaders in the modern organisation, this is today your most important job. And if your organisation hasn't felt this yet, plan for it now and you will avoid the trap. Otherwise our opening story has a rather dull ending: The tools let you scale up your output, but they also scaled down your circle of human interaction until only you were left. You're a one-person army, but the battlefield is quiet, and the victory feels curiously empty.
The views expressed in this article are those of the author and do not necessarily reflect the position of Stellar Capacity. It is published as part of our faculty contributor series to share independent perspectives and spark dialogue.
Work should be meaningful and bring out our best. Marcin Floryan is a member of Stellar Capacity’s faculty and a software engineer at heart. He is currently VP of Software at Marshall Group, and he previously held several engineering and product leadership roles at Spotify. He cultivates organisations where psychological safety, diverse perspectives and continuous learning spark innovation; and speaks and writes about what it means to lead well in a rapidly changing world.