Why I’m tired of watching companies wait for a crisis to run themselves well
Every few years, the business world rediscovers something it could have been doing all along. In 2020, it was remote work. Offices emptied, Slack and Zoom replaced hallway conversations, and companies realized their people could, in fact, be trusted to do their jobs without being watched.
Now, in 2026, the impetus for change is AI. Teams are flattening. Approval chains are shrinking. Middle layers are being compressed into dashboards. Executives are describing these moves as “bold, necessary transformations”.
I have watched this pattern with a mix of amusement and frustration. At Kuma, we’ve been operating virtually for twelve years. We have deliberately kept our middle management layer thin. Our people own their careers.
We’ve delivered data security services to serious customers the entire time. Not because of a pandemic. Not because of a new class of models. Because this is simply a better way to run a company.
What bothers me is not that companies are changing. It’s that they seem to need an external reason.
Real efficiency doesn’t require a permission slip
COVID drove leaders to send people home. AI is giving them cover to rethink organizational design. Both shifts, in isolation, are positive. When done properly, flexible work environments expand the talent pool and allow employees to work wherever they are most productive . AI is genuinely useful for reducing coordination overhead, drafting, analysis, and routine decision-making. I am not anti-remote or anti-AI. I am pro-both.
But framing these as the reason for change reveals something striking: many organizations were running inefficiently on purpose, and waited for a socially acceptable excuse to stop.
The pandemic didn’t make remote work viable. Remote work was already viable. The pandemic simply made it awkward to deny. AI isn’t making flat organizations possible. Flat organizations were always possible. AI is just making it harder to justify five layers of approval for a decision that a single competent operator could make in ten minutes.
When a company announces, in 2026, that it is restructuring around AI to empower frontline teams and compress its management stack, I read between the lines. What they are really announcing is that they have tolerated the opposite arrangement for years and finally found a narrative that lets them fix it without admitting it was broken.
The things companies are discovering now, we built from day one
When we started Kuma, we weren’t striving to be progressive or novel. We were trying to build the best possible data security company with the smallest possible friction between our people, our clients, and the problems we were solving. Every structural choice in building our operating model flowed from that goal.
We decided not to open an office because offices are expensive, geographically exclusionary, and introduce a false proxy for productivity: presence.
We decided not to build a heavy middle-management layer because layers slow decisions and distance employees from the customers they serve.
We decided not to dictate time-off policies because adults who are trusted with sensitive customer data can also be trusted to manage their own rest.
We decided to give employees ownership of their career paths because a person closest to their own strengths and interests is better positioned to chart their growth than a manager two rungs above them filling out a form.
None of this was radical. None of it required an external movement to motivate. It required asking, honestly: What organizational design actually best serves the business and the people in it?
The real cost of middle management is distance
There is a fashionable critique of middle management that frames it as a cost problem. Too many salaries, too much overhead, too many meetings. That critique is true but ultimately shallow. The deeper cost isn’t financial. It’s distance.
Every layer between a client and the person solving their problem is a layer of translation, delay, and distortion. A client describes a pain point. The account manager relays it to a product manager. The product manager files a ticket. The engineering manager prioritizes it against other tickets. The engineer builds something. By the time the fix ships, the original context has been summarized four times, and the person who wrote the code likely has never spoken to the person who experienced the problem.
Lean operating models don’t just save money; they save understanding. When an engineer can talk directly to a client, when a salesperson can ship a small fix, when a support lead can change policy without convening a committee, solutions get better and they get there faster. AI tools can accelerate this, but they aren’t all that makes it possible. What makes it possible is a deliberate choice to keep the distance short.
Trust is the precondition, not the reward
A lot of the operating model conversation gets reduced to policy: unlimited PTO, no return-to-office mandate, asynchronous-first communication. Those are artifacts, not principles. The underlying principle is trust.
If you trust your people, you don’t need to police where they sit or how many vacation days they’ve taken. You don’t need three approvers to authorize a one-thousand-dollar decision. You don’t need a manager whose primary job is to watch other people work. If you don’t trust your people, no policy document will save you, and no AI tool will rescue the culture. You will simply be surveilling them more efficiently.
Companies that have historically withheld trust are now using AI as a pretext to finally extend it, sometimes to fewer people than before. That’s a reasonable strategic move, but it’s not a reinvention. The reinvention would have been to extend trust in the first place, and to design the organization around the assumption that it was warranted.
Stop waiting for a movement
If you’re a leader reading this and thinking about how to modernize your operating model, I want to invite you to consider that the frame itself is part of the problem. “Modernize” implies there is a timetable, a trend, a peer group to keep pace with. There isn’t. There is only the question of whether your organization is designed to let great work happen, or designed to protect existing structures from being questioned.
You don’t need another pandemic to allow your team the flexibility to work from where they do it best. You don’t need the next wave of AI to cut an approval chain that never served the customer. You need the willingness to look at your company honestly and remove the things that are not earning their place.
At Kuma, we didn’t get here because we were prescient. We got here because we were stubborn about one question: Does this genuinely make the work better, or does it just make us look more like existing organizations that are considered reputable? When the answer was the latter, we didn’t do it. Twelve years later, the industry is slowly arriving at conclusions we bet on in year one.
That is encouraging. It’s also a reminder: The companies that will lead the next decade are not the ones riding the current narrative, but the ones who would have made the same choices without it.