Not a typo. And it’s not a complaint about bad software.

BCG's 2026 AI at Work study surveyed nearly 12,000 workers across 14 markets.
47% now spend more time managing the AI than doing the job the AI was supposed to help with. (BCG, June 2026.)

It gets stranger. Among frontline workers who use it regularly, 42% say that AI saves a full workday or more every week. A whole day, back in their hands. And two-thirds of them report getting little or no guidance on what to do with it. More than half pour the saved time back into more of the same work, only faster.

The time showed up. The purpose of the time did not.

Here is the line in the study that deserves your attention. A clear strategy lifts the business impact of AI by 25%.

Better tools? About 5%. The tool is not the lever. The tool was never the lever.

So why does almost everyone keep pulling the tool lever?

It’s because reshaping a job is easy. Reshaping work is hard. Reshaping a job means handing someone a faster engine for the work they already do. Reshaping work means asking what the work should even be now that the engine exists. Most of the time the honest answer is that it should be something different. The first is a purchase order. The second is a redesign. Companies are racing through the first and quietly skipping the second, which is exactly why the research is headlined: AI is reshaping jobs faster than companies are reshaping work.

Give a stagecoach driver a faster horse and you get a slightly faster stagecoach.

Not a car.

Inventing the car required someone to stop optimizing the horse and instead ask the purpose of the journey. That question, asked on purpose, inside a company, about its real work, is the discipline I call the Exponential Work Framework: redesigning the work around the new leverage you now command, instead of bolting leverage onto work you already had shuttling context and data between a dozen apps.
Copy. Paste. Repeat.

Last week's issue highlighted Organizational Intelligence Design as the practice of making the invisible map of how your company thinks legible enough to hand to a machine. Exponential Work is what you do once that machine is genuinely capable: you do not give it the old job to do faster, you redesign the job around what only humans should decide.

BCG's own people are circling the same idea from the outside. As one of their managing directors put it, the first wave of AI was about individual productivity, and the coming wave "is really about rethinking the human value-add inside."

That phrase, the human value-add inside, is the whole game. 61% of workers already believe that AI agents will be able to do half of their job in three years (same 2026 BCG study).

Why make a person faster at the half that is leaving?

The Exponential Work Framework evolves their role around the half that remains.

Reshape the work, not just the job. The companies that do it on purpose compound. The ones that buy faster horses get a very expensive stagecoach — and a workforce that spends its hard-won day managing the reins.

→ Half-Baked.
[ THE DIP ] The most valuable AI work right now looks like the least urgent. Redesigning a workflow is slow, unglamorous, and shows up on no dashboard this quarter, while buying a new tool ships a press release by Friday. The redesign is the thing that compounds, and it is precisely the thing that gets deprioritized because it is boring now. The dip is real. Sit in it.

→ Hot Take.
"Time saved" is a vanity metric. 42% saving a workday a week sounds like a win until you learn most of it gets reinvested in the same work, only faster. Saved time you pour back into the same work is not productivity. It’s a faster treadmill. And the people on it can feel it. That’s the tell: in the same study, satisfaction and cognitive load rose together. The treadmil feels productive while it quietly costs you.

→ Confession.
I keep wanting to measure my own AI leverage in hours saved, because hours are easy to count. They are also the wrong unit. The honest measure is whether the work I do now is work I could not have done at all before, not whether I did the old work in less time. I make this mistake more weeks than not.

> If AI gave your best person a full day back next week, do you know what you would want them to do with it? Or would it quietly refill with the work they already had?

If you do not have an answer to the first question, the second one already happened.

Reply if you see this differently.

Pick one role on your team that AI has clearly made faster.

Do not ask how much time it saved.
Ask the harder question: what should this role do now that it could not do before?

Not the same output quicker. A different output entirely. The one that was impossible when this person spent their week on the part the machine now handles.

Write down one thing the role can now produce that was previously out of reach. That sentence is the first line of a redesigned job. It is also the difference between a team that got a faster horse and a team that is quietly building the car.

Monday. One role. One impossible-before output.

This week's durable asset is the framework itself, gathered in one place you can keep coming back to: Exponential Work is less a read than a tool.

The whole 2×2 is there — Reinvent the work on one axis, Scale past one on the other. You can drop any role onto the grid and see whether you are building a faster horse or finally building the car.

The question that starts the redesign is there too, in a form you can carry straight into your next team meeting: what could this role now produce that was out of reach before? It is worth sitting with before your next planning cycle — especially if AI just handed your team a day back and no one has decided what the work should become.

There is a harder discipline underneath the redesign, and it is where this series is heading. Once you decide to hand part of the work to the machine, you hit the question that decides whether you stay sharp or quietly hollow out: what do you offload, and what stays yours?

Hand over the toil and your judgment gets room to compound. Hand over the judgment and you slide into cognitive debt without noticing. Where you draw that line — your waterline — is a coming issue.

https://chrishuberreitz.com/frameworks/exponential-work

This newsletter is called Signals from the Curve because there are two kinds of forecasting: the curve and the cliff. The cliff says everything changes at once. The curve says it is already changing, you just have to know where to look.

I write here every Wednesday morning about AI strategy, leadership, and the parts of work that compound. If something here changed how you are thinking, hit reply. I read every response.

— Chris

Chief of AI & Strategy at Essential Innovations · Founder, Attainable AI · Adjunct Faculty, Columbia University

chrishuberreitz.com · LinkedIn