There's a category of work that doesn't appear in any job description, doesn't show up in project trackers, and rarely gets discussed in team planning sessions. But if you watch where your best people actually spend their time, you'll find it everywhere.
It's the work of moving information between systems. Interpreting what the CRM says to make a decision in the project management tool. Pulling together context from three different places before you can have a meaningful conversation. Translating what sales told you into something delivery can act on. Monitoring a dozen data sources and deciding what's actually important.
Call it the coordination tax. It's the cost of operating across multiple systems, functions, and contexts — and it's paid primarily by the people who understand your business well enough to do the translation work that the systems can't do on their own.
Why it falls on your best people
The coordination tax is regressive in a specific way: it falls hardest on the people who can least afford to pay it.
Coordination work requires context. You have to understand what the data in the CRM actually means, not just how to read the interface. You have to know which project issues are genuinely urgent and which can wait. You have to understand the customer's situation well enough to write a response that's actually useful rather than technically accurate.
That kind of contextual understanding takes time to develop. In most businesses, the people who have it are the senior members of the team — the ones you'd most want focused on work that requires strategic thinking.
Instead, a significant portion of their day goes to coordination: extracting information, reformatting it, synthesising it, routing it to the right place. Work that needs to be done, but that shouldn't require the people doing it.
Where AI is particularly useful
AI is well-suited to coordination work for a specific reason: it can hold a lot of context simultaneously and reason across it.
The task of "here is what the CRM says about this account, here is what happened in the last three support interactions, here is the open proposal — produce a briefing note for the sales conversation tomorrow" is a coordination task. A human can do it. AI can do it faster, with more consistent structure, and without the cognitive cost of context-switching.
At Pattern, the most significant productivity gains we've found haven't come from automating discrete tasks. They've come from claiming the coordination work that sat between tasks — the research, synthesis, and translation that was consuming time without appearing anywhere as "work."
Some specific examples. Before a client workshop, pulling together the account history from our project tool, the relevant communications from email, and the current status of outstanding issues — and producing a structured briefing — was an hour of someone's time. It's now a few minutes. Before a sales conversation, enriching a contact record with company information, recent news, and relevant context from our CRM used to be manual. It's now automatic.
Neither of these is a dramatic transformation individually. Added up across a week, they account for a meaningful number of hours — hours that the people involved can redirect to the preparation and thinking that actually requires them.
What gets freed up
When the coordination tax is reduced, the visible change isn't just that people have more time. It's that the texture of their work changes.
Work that felt fragmented — constant context-switching, always partially caught up, always a few steps behind — becomes more continuous. People can focus longer on single problems. They're less often the bottleneck between systems. And the quality of their judgement tends to improve, because they're not making decisions under the cognitive load of having just synthesised three data sources.
This is harder to measure than "hours saved" — but it's the change that tends to matter more in practice.
Finding it in your business
Identifying where the coordination tax lives requires looking specifically for it, because it's designed to be invisible. It happens in email, in Slack, in "let me just pull that together quickly" conversations that don't get counted as anyone's time.
A useful approach: track where your most senior people spend their time for a week, with the specific intent of finding work that requires context but not judgement. Work that anyone who understood the systems well enough could do — but that defaults to the most experienced people because they're the only ones who know what they're looking at.
That's the work AI can claim. It's also the work that tends to compound — because when the most experienced people get it back, they usually put it toward something that produces more value than the coordination work it replaced.
The coordination tax isn't something most businesses have calculated, because it's never been a line item. But it's real, it's significant, and for the first time there's a reasonably tractable solution — once you've identified it clearly enough to act on it.
