The illusion of less work
AI does not remove work; it multiplies it. When output becomes cheap, volume explodes. Without leadership deciding what “enough” looks like, optimisation just accelerates spend. That is why AI does not reduce the need …
Connections
Index
- The context
- The promise every technology makes
- Why work never disappears
- Frequency beats effort
- AI as a demand multiplier
- Where the cocktail fantasy breaks
- Why this is a leadership problem
- What top-down FinOps actually changes
- The real next action
1. The context
I was scrolling LinkedIn the other day and came across an advert I have now seen in many variations.
On one side of the screen, someone doing things “the old way”: manual work, visible effort, mild despair. On the other side, someone using AI: relaxed, smiling, sitting somewhere sunny, holding a phone in one hand and a cocktail in the other, while the system quietly does the work for them.
It is a great image.
And I do not believe it for a second.
AI is not going to remove work. It is going to change its shape, and dramatically increase its volume.
We have been here before.
2. The promise every technology makes
Almost every major technological invention starts with the same promise:
“You will work less.”
The washing machine.
The hoover.
The computer.
Email.
Automation.
Now AI.
Each time, the story is identical: less effort, more time, better life. And each time, the outcome is… different.
3. Why work never disappears
The washing machine did not eliminate laundry. It eliminated waiting. As a result, clean clothes every day became normal.
The hoover did not end cleaning. It made “reasonably clean” unacceptable.
Computers did not reduce paperwork. They multiplied it.
Technology does not remove work. It removes friction. And when friction disappears, expectations rise.
What used to be “nice to have” becomes “obvious”.
What used to be “occasional” becomes “constant”.
4. Frequency beats effort
Email is the clearest example.
When letters took days to arrive, you wrote a few and you waited. The delay imposed discipline. You thought before writing.
Email removed that delay.
So what happened?
We did not stop communicating. We communicated more. Five letters a week turned into tens—sometimes hundreds—of emails a day. Then messages. Then notifications. Then follow-ups asking if you saw the previous message.
The effort per action collapsed.
The frequency exploded.
That is the real effect of “time-saving” technology.
5. AI as a demand multiplier
AI fits perfectly into this pattern.
Generating a report used to take days. Now it takes minutes.
Writing a proposal used to take hours. Now it takes seconds.
Finding leads used to be expensive. Now it is trivial.
So we do not stop.
We generate more reports.
More proposals.
More leads.
Not 2× more.
10×.
100×.
Sometimes 1,000×.
AI is not a labour reducer.
It is a demand multiplier.
6. Where the cocktail fantasy breaks
The advert shows someone “letting AI do the work”.
What it does not show is what happens next.
Someone still has to:
- qualify the output
- filter the noise
- handle edge cases
- deal with errors
- explain results
- take responsibility
And crucially: deal with the volume.
Generating 1,000 leads is easy.
Handling 1,000 leads is not.
AI moves the effort downstream. It does not erase it. And downstream effort is often more expensive, not less.
The cocktail is marketing.
The throughput is the bill.
7. Why this is a leadership problem
The problem with AI is not that it produces the wrong answers.
The problem is that it produces too many acceptable ones.
Once output becomes cheap, teams will always ask for more of it. More leads, more reports, more forecasts, more experiments. From their point of view, this is rational behaviour: the tool exists, it is fast, and, for now, it looks almost free.
No individual team is responsible for the total volume.
No individual team sees the full bill.
No individual team decides when “enough” has been reached.
That decision only exists at the leadership level.
This is why AI breaks bottom-up optimisation. Improving efficiency at the execution layer does not reduce cost; it increases demand. Without an explicit decision about limits, optimisation simply accelerates spend.
Someone has to decide:
- which outputs actually matter
- how much volume is justified
- what trade-offs are acceptable
- where the stop line is
Those are not technical decisions. They are business decisions.
And business decisions cannot be delegated to a model, a tool, or a team trying to be helpful.
8. What top-down FinOps actually changes
This is where top-down FinOps becomes essential.
Bottom-up FinOps is good at asking:
“Can we do this cheaper?”
Top-down FinOps asks:
“Should we do this more?”
AI makes that distinction unavoidable.
Efficiency gains will always turn into volume gains unless leadership intervenes. Someone at the top has to decide:
- what demand is acceptable
- what success actually means
- what happens when limits are reached
Without that, AI becomes a perfectly optimised washing machine with no off switch.
9. The real next action
The next action is not another optimisation pass.
The next action is to force a decision upstream.
Before scaling AI, organisations need to answer:
- What outcome are we optimising for, not what activity?
- What is the acceptable cost per real outcome?
- Where do we stop when volume explodes?
- Who is accountable for saying “enough”?
AI will happily do more.
The cloud will happily bill for it.
Finance will happily report it.
Only leadership can decide when to stop.
That is why AI does not reduce the need for top-down FinOps.
It makes it unavoidable.
And no cocktail will save you from that.