Estimation in software has always been broken.
Not because people are bad at it, but because we keep asking it to do something it can’t: predict outcomes in complex, uncertain systems.
AI hasn’t changed that.
What it has changed is speed — and that makes estimation more dangerous, not less.
Speed increases confidence faster than understanding
AI makes it easy to move quickly:
- code appears fast
- demos look good
- progress feels visible
That often increases confidence faster than it reduces uncertainty.
You get to the good place quicker — or the bad place quicker.
AI compresses the time between “this seems straightforward” and “this is more complex than we thought”. That’s not a tooling issue. It’s a decision-making one.
Faster delivery makes scope creep quieter
When work is slow, scope creep is obvious.
When work is fast, scope creep becomes subtle.
It’s easier to say:
When work is slow, scope creep is obvious.
When work is fast, scope creep becomes subtle.
It’s easier to say:
- “let’s add one more thing”
- “while we’re here…”
- “we’ve got time”
Not because anyone is careless — but because speed makes additions feel cheap.
That’s exactly when constraints matter most.
The business question hasn’t changed
Despite AI changing how fast we can build, the core business question is the same:
What is this worth?
Despite AI changing how fast we can build, the core business question is the same:
What is this worth?
Not:
- how fast can we ship
- how much can we add
- how productive can we be
But:
- how much investment makes sense
- what problem are we actually solving
- when is “enough” enough
AI affects execution.
It doesn’t change value.
Why appetites still beat estimates
This is why I prefer appetites over estimates, especially now.
Instead of asking:
“How long will this take?”
We ask:
“This problem is worth ~X weeks of effort. What’s the best version we can deliver inside that?”
That shift:
- fixes cost up front
- forces early prioritisation
- encourages cutting scope
- keeps uncertainty visible
The hard conversation happens at the start — not at the deadline.
AI works inside the constraint
AI is incredibly useful within a clear appetite.
It helps:
- explore options faster
- find simpler approaches
- validate assumptions earlier
But the appetite itself shouldn’t move just because things feel faster.
If anything, increased speed makes the constraint more important, not less.
AI increases leverage.
The appetite provides discipline.
The appetite provides discipline.
The real shift
AI hasn’t fixed estimation.
It’s made it clearer that estimates were never the right tool for the job.
What we actually need is clarity about:
- what we’re willing to invest
- what trade-offs we accept
- when to stop
That was always true.
AI just raises the stakes.
AI just raises the stakes.