The Coordination Tax: Why AI Made Your Boss Faster and Your Meetings Longer

Your manager thinks coordination eats about 21% of their week. Their own itemized time logs say the real number is 41%, and nearly half of that is spent sitting in status meetings. That gap, the one between what your boss believes and what’s actually on the calendar, is the whole story of AI at work right now.

The people who use AI every day aren’t drowning less. They’re drowning more. Daily AI users among managers report the heaviest coordination load at 20.3 hours a week, versus just 9.1 hours for managers who don’t touch the tools, according to the In Parallel study reported by IT Brief. We think this is the most honest signal in the entire AI hype cycle, and it explains why your workday feels busier even as the demos get shinier.

The tax nobody put on the invoice

A Helsinki software company called In Parallel coined the phrase “coordination tax” straight from its own survey of 247 managers across five countries. They define it as the overhead effort required to keep people, decisions, and work aligned before the actual work can even start.

Their estimate of the damage is blunt: roughly $64,600 per manager per year, driven by an average of 16.5 hours a week lost to status meetings, re-explaining what’s already been decided, and searching for information that exists somewhere but nobody can find.

• Status meetings swallow close to half of all coordination time, per the same data.
• Re-explaining work happens because context lives in one person’s head, not in a shared place.
• Searching for information is the quiet killer, small on any given day and enormous over a year.

Interview Guys Take: Fair warning: In Parallel sells coordination software, so they have a commercial reason to make this problem loud. But you don’t need their sales pitch to believe the shape of it. Anyone who’s watched a 45-minute meeting exist purely so five people can confirm they’re not blocked already knows the tax is real. The vendor bias affects the pricing, not the pattern.

Why the AI users are the ones bleeding hours

Here’s the part that should stop you cold. The managers using AI daily aren’t the least burdened, they’re the most. Twenty hours a week on coordination versus nine for the abstainers, even though 86% of managers use AI tools at all.

In Parallel’s diagnosis is that AI tools lack access to organizational context: prior decisions, past discussions, standing commitments. So a human has to bridge that gap by hand, which means more status updates, more repeated explanation, and more meetings, not fewer.

• AI produces more output which means more artifacts for other people to review, question, and align on.
• AI doesn’t know your backstory so someone has to feed it context and then translate its answer back into your team’s reality.
• Faster individual work creates more handoffs, and every handoff is a coordination event.

The 39-point blind spot

The cleanest evidence for this whole mess comes from developers, the group everyone assumes AI helps most. A METR randomized controlled trial put 16 experienced open-source developers through 246 real tasks. The ones using AI took 19% longer.

Then those same developers were asked how it went. They believed AI had made them about 20% faster. That’s a 39-percentage-point gap between what happened and what they felt happened, and it’s exactly why the coordination tax stays invisible. You don’t feel the overhead of verifying, prompting, and correcting. You feel the speed of the first draft.

• The feeling of speed comes from the fast part (generating a draft), not the slow part (checking it).
• The overhead hides in the review, the second-guessing, and the meeting to align on what the AI produced.
• METR walked some of it back in a February 2026 update after finding 30-50% of invited developers refused to work without AI, a selection effect worth knowing about.

Interview Guys Take: The perception gap is the actual product being sold. If your boss feels faster and the org measures nothing, guess which one wins the budget meeting. This is why the same executives who can’t find the productivity gains keep expanding AI programs, they’re managing a feeling, and feelings are notoriously hard to put on a spreadsheet.

The firm-level numbers are brutal

Zoom out from one team to thousands of companies and the story gets worse for the hype. A 2026 NBER working paper surveyed nearly 6,000 CFOs, CEOs, and senior executives across the US, UK, Germany, and Australia.

The result: 69% of firms actively use AI, but 89% of executives reported no measurable impact on labor productivity over the past three years, and more than 90% reported no impact on employment. The PwC 2026 Global CEO Survey lands in the same place, with only 12% of CEOs saying AI delivered both cost and revenue benefits and 56% seeing no significant financial benefit at all.

• 69% adoption with 89% reporting no productivity gain is not a rollout problem, it’s a wiring problem.
• 90%+ no employment impact means the “AI is taking the jobs” panic and the “AI made us efficient” boast can’t both be true.
• The money isn’t showing up where finance can see it, which is the only place that ultimately counts.

Work about work was already the disease

AI didn’t invent this problem, it inherited it. Asana’s Anatomy of Work Index, built from more than 10,000 knowledge workers globally, found that 60% of the average workday goes to “work about work”: communicating about tasks, hunting down documents, and managing shifting priorities.

That leaves only about 27% for the skilled work you were actually hired to do and 13% for strategic planning. Per year, the average knowledge worker loses 103 hours to unnecessary meetings, 209 hours to duplicated work, and 352 hours simply talking about work.

• 60% coordination overhead existed before ChatGPT, so dropping AI into it just multiplies the busywork.
• 352 hours talking about work is nearly nine full work weeks a year spent describing the work instead of doing it.
• AI accelerates whatever you point it at and if you point it at a broken process, you get a faster broken process.

The IBM lesson: optimize a part, break the whole

IBM ran its own AI restructuring and learned the expensive way. When it optimized individual functions without coordinating across end-to-end workflows like quotation-to-cash or source-to-pay, it got only marginal improvements.

Its May 2025 CEO Study found just 25% of AI initiatives hit their expected ROI, with disconnected technology and processes across organizational boundaries named as the main reason. A Harvard Business Review analysis in April 2026 described the same split: executives experience AI as strategy while managers inside real workflows hit its flaws, and initiatives stall in the space between those two realities.

• Only 25% of AI initiatives reached expected ROI at IBM, and the gap traced to cross-boundary coordination failures.
• 66% of C-suite leaders told Deloitte’s 2026 trends that traditional functions must change, but only 7% had actually restructured them.
• The stall lives between layers where the strategy deck meets the actual handoff, and that’s where your extra meetings come from.

Where the gains are actually real

We’re not arguing AI does nothing. We’re arguing it works in narrow lanes and clogs the wide ones. A field study of 5,179 customer-support agents found AI access raised productivity 14% on average and 34% for novice workers, and Goldman Sachs documented roughly 30% gains in the narrow use cases of coding and customer service.

The pattern is consistent: AI delivers strong, measurable results in high-volume, repetitive, well-defined individual tasks. Those gains just don’t yet travel through coordination-heavy work that crosses teams. Meanwhile Stanford’s Erik Brynjolfsson notes US aggregate productivity grew 2.7% in 2025, nearly double the prior decade, so something may be brewing at the macro level even when it’s invisible at the firm level.

• Defined, solo, repetitive work is where AI shines and the numbers hold up.
• Cross-team, context-heavy work is where the coordination tax eats the gains alive.
• The macro picture is contested which means the honest answer today is “it depends where you point it.”

Interview Guys Take: If you want to future-proof your value, notice which side of that line your job sits on. The roles that survive this decade won’t be the ones doing the repetitive task AI does cheaply. They’ll be the ones absorbing the coordination that AI can’t touch, and coordination is a human skill that shows up in the fastest-growing corners of the job market more than the buzzy ones.

What this means for your next move

If coordination is the real work now, then the skills that get rewarded are the ones that reduce it: turning chaos into a decision, keeping a cross-functional project aligned, and making context legible to people who weren’t in the room. That’s leadership work, and it’s why the ability to walk through a messy project with a clear SOAR structure (Situation, Obstacle, Action, Result) matters more than knowing which AI tool is trending.

It also reshapes how you should read a job you’re considering. Ask what percentage of the role is defined solo work versus cross-team coordination, because one of those is getting automated and the other is getting more valuable. If you’re weighing a switch, run the math honestly first, since every career change carries its own hidden tax too.

• Coordination is now a competency so learn to name it, and practice it in panel interviews where alignment is literally the test.
• Managing sideways matters which is really the skill behind disagreeing with your boss without blowing things up.
• Value gets rewarded when named so if you’re carrying the coordination load, that’s your case in the promotion conversation.

The honest read on AI right now is that it made a lot of individual tasks faster and made the organization around those tasks heavier. Your boss feels the speed. You feel the meetings. Both of you are correct, and the gap between those two truths is the coordination tax with the meter running.

The number worth remembering is the 21 that’s really 41. Until leaders can see how much of the week actually goes to coordination, they’ll keep buying tools to speed up the work and wondering why the calendar keeps filling. If you can see the tax clearly and describe how you cut it, you’re holding the exact skill this moment is quietly desperate for.