The Death of the Pause

I'll start with a disclosure, because it doubles as evidence. AI helped write this, though not the way most people picture it. It began with me going back and forth with an AI on my own thoughts, testing ideas the way I'd test them on a colleague over coffee. This happened on my drive, over two hours of discussion, edits and revisions. Once the argument held up, the work moved to a drafting agent that turned those conversations into something readable, and from there to a publishing agent that gets it out the door. I used Grokbots for this. xAI only introduced Grok Bot on August 11, 2026, barely three months before I sat down to write this. I run eight Grok Bots in my company, and the number keeps growing.

The date matters to my argument. Agents were around before last summer, but in my experience building one often meant hiring a developer and waiting unless one was technically capable themselves. The difference is Grok Bot made it something an owner could do on a Sunday afternoon: create an agent, show it a workflow once and put it to work, with no technical background required. When it gets that easy, adoption stops being a pilot project and spreads through a company in weeks, and I'm as guilty of that as anyone. All of those new agents produce work that eventually lands on somebody's desk, and the faster they multiplied, the faster the people at the top filled up. Leaders were already carrying more than they could handle before August. For a lot of them, this was the straw that broke the camel's back.

Without really setting out to, I've built an organizational bot pyramid. Several agents work together, each handing its output up to the next, and at the top sits a "Chief" bot. But ultimately there is still one human who has to read it, check it and decide whether it goes out under his name. Every layer below me got faster. The decision needs at the top didn't move an inch, and that, in miniature, is what this piece is about.

Why the pause matters

For most of the history of business there was a gap between something happening and somebody acting on it. A letter took days to arrive, a report took a week to pull together, and a proposal sat in a folder until Tuesday's meeting. We spent decades trying to close that gap, because slow companies lose. But the gap was doing a job we never gave it credit for. It was where judgment happened. Somebody read the letter twice, noticed the numbers didn't add up, or slept on the proposal and came back with a better idea.

I've recently heard that gap referred to as the pause, and the name stuck with me because it describes something I'd felt for years without having a word for it. You couldn't buy it and you couldn't delegate it, because it came free with the friction of doing business. AI is the technology that finally removed it, and I say that as one of the people removing it fastest.

I used to think of the pause as a nice-to-have, the kind of breathing room you enjoy when a quarter goes well. I've come around to seeing it as one of the load-bearing walls of a company. The pause is where a leader asks what happens after the obvious thing happens. You cut a price to win a deal this month, and the pause is where you notice it also resets what every other customer expects to pay. Or you hire quickly to cover a busy season, and only on reflection do you see that the culture you spent ten years building is about to be outnumbered. Those second-order questions almost never show up in the report itself. They surface when someone has the time to sit with it.

The pause is also where strategy actually gets made. Nobody sets a five-year direction between back-to-back approvals. The decisions that define a company, like which market to leave, which client to let go or which bet to make while competitors are still debating, come out of the quiet hours when a leader can see the whole board instead of the next move. That's where you spot the cliff while there's still road left to brake on.

Take that away and the company keeps executing at full speed with nobody looking up at the road. That's a structural risk, and I think most owners are still treating it as a scheduling problem. The damage rarely arrives as one bad week. It builds over years, as a string of decisions that technically got approved but that nobody really made.

Ten times the decisions

Back in 2023, before agents were doing much of anything on their own, an Oracle survey found that 74% of respondents said the number of decisions they make every day had grown tenfold in three years. Oracle sells data tools, so I hold that figure loosely, but the line that stuck with me came next. Seventy-two percent said "the sheer volume of data and their lack of trust in data has stopped them from making any decision."

Those people had plenty of information. What they'd lost was the room to weigh it, and when there's no room to think, a lot of capable people simply stop deciding. Since then we've handed the keyboard to software that sends, books, posts and builds on its own, often before anyone has looked.

98% more work, 91% longer to approve it

Software shows the next step first, because it's measured so closely. When Faros AI studied engineering teams with heavy AI adoption, it found they merged 98% more pull requests while the time spent reviewing them went up 91%. The code got written faster than ever, and the people who had to sign off on it became the constraint. In Faros's own words, the bottleneck shifted to review.

Every office that produces documents is living some version of that now. Proposals, reports and status updates can be produced in seconds, and each one eventually lands in front of somebody who has to decide whether it's any good. A good report used to take effort, so there weren't many, and reading each one carefully was a sensible use of a senior person's afternoon. Volume destroyed that scarcity, and the only thing still genuinely scarce is the attention of whoever decides what's worth acting on.

Step back and you can see the shape of the whole problem. The traditional org chart is a pyramid, with a broad base doing the work, a middle layer checking and shaping it, and a small group at the top setting direction. AI has turned that pyramid into a funnel. The wide mouth sits at the bottom, where output now pours in faster than it ever has, and the narrow neck sits at the top, where one leader, or a handful of them, still has to decide what any of it means. In a big company the neck is an executive team. In most businesses, it's the owner.

It barely matters how the extra output gets made. Some companies are replacing people with AI, and others are handing their people AI and asking for more, and either way the mouth of the funnel gets wider. The coordinator who used to send a handful of proposals a week now sends several times that, and the agent that started as an experiment now runs around the clock. Each of those is a productivity win at the bottom, and each one eventually needs a yes, a no or a not-yet from somebody near the top. The neck doesn't widen to match. The owner's day still has the same number of hours, and there's still only one of them.

The layer that used to narrow the flow before it reached the top is thinning at the same time. Gartner predicted in 2024 that through 2026, one in five organizations would use AI to flatten their structure, "eliminating more than half of current middle management positions," and plenty of smaller firms are getting there simply by not refilling roles.

So the pressure pushes upward, and it lands on people who were already carrying the bank, the payroll, the regulators and the customer who will only talk to the boss. Something has to give at the bottleneck, and the first thing to go is the pause. The afternoon you set aside for thinking fills with approvals, and the drive home turns into one more call. A leader can absorb a heavy week. What a funnel produces is a heavy week that never ends, and the hours that used to go to judgment get spent keeping the line moving.

When everything funnels up to a few people at the top, something has to give, and the first thing they give up is time to think.

I see it from my own desk. Clients have been telling me lately that they're having a hard time getting through to senior management, and when they describe the people they're trying to reach, the same word keeps coming up. Everyone seems overwhelmed.

93% are using it. 2% are seeing the return.

This is how companies go over the cliff, and it rarely looks dramatic from the inside. A KPMG Canada survey of 753 business leaders found 93% were using AI in some form, while only 2% said they were seeing a return on their generative AI investments. I don't read that as proof the tools fail. To me it says a lot of organizations sped up the bottom of the pyramid without anyone at the top having the time to decide what all that speed was for.

I'm not slowing down on AI, and I doubt many people reading this will either. I have stopped pretending the speed is free, though. Every agent I add widens the mouth of my own funnel, and every one of them ends in the same place, with one person deciding whether the work is right. So I guard the pause now the way I guard cash flow, with time on the calendar that nothing new is allowed into and a short written list of the decisions that are actually mine.

Coming full circle...

At the start of this piece I described my own small pyramid of agents, where every layer below me got faster and the decision at the top didn't move an inch. That's the whole argument, scaled down to one desk. We removed the pause that used to sit between something happening and somebody acting on it. We turned the pyramid into a funnel and poured more through the bottom than any leader can catch at the top. With the neck that full, judgment and strategy are the first things to go, and a company can run at full speed for a long time with nobody really steering before anyone notices the road has run out.

That is what's crushing leadership right now. Most leaders I know are working harder than ever, and the work keeps coming. The part that's quietly disappearing is the one stretch of time when their work was actually leadership, the quiet hour where they could look up from the next decision and see where the company was headed. Getting that hour back is now part of the job for every owner who has adopted these tools as quickly as I have, and in my own business it has moved to the top of the list.