The video could not be loaded. Press Play to try again, or read the transcript below.
- With the player focused:
- Space play or pause
- C captions
- JL back or forward 10 seconds
- F full screen
AI Is Making It Easier to Build the Wrong Thing
Plausible decisions compound into the wrong direction.
Transcript
Agentic delivery will make execution faster and more abundant. Unless the same intelligence goes into understanding intent, it will also help us travel much further in the wrong direction.
It begins with an idea that isn’t yet completely clear. AI produces a plausible suggestion. “Yes, that will do.” The next output builds on the first, then another builds on that.
Nothing looks obviously wrong in isolation. But several iterations later there’s a substantial gap between what was wanted and what has actually been created. Call it the AI rabbit hole.
The AI has not necessarily failed. It followed the direction it was given. A provisional suggestion became an accepted decision. AI makes that surprisingly easy.
Transformation has the same problem. Understanding what is really needed is often compressed or skipped. Used carelessly, AI accelerates every step. Assumptions become requirements, requirements become designs, and designs become working software before anyone has tested the thinking underneath them.
This isn’t an argument against agentic delivery. Use AI to shorten development cycles, generate code, create tests and produce documentation.
But applying AI only to execution leaves half the opportunity untouched. Use it to understand too. Compare interpretations, expose contradictions, and test assumptions.
The article behind this video was written that way. AI interviewed its author, and challenged where the position was unclear. It contributed knowledge without being allowed to manufacture the point of view.
One customer kept saying they wanted to go faster. The obvious response was to take the familiar sequence of workshops, build, testing and enablement, and compress it.
That wasn’t what they were asking for. They wanted to start with the out-of-the-box build, then use testing and enablement to find the few changes they needed.
Best practice is whatever’s best for this customer, with these people, in this situation. AI is exceptionally good at reproducing established patterns. Without critical review, you get the same best-practice answer every time, whether or not it fits.
The context that changes a recommendation is not always captured in a process document or a data model. It may be tension between two departments, the way a particular leader makes decisions, or the moment in a meeting when a sensible suggestion doesn’t land.
AI can analyse what was recorded. It cannot yet be relied on to navigate every human dynamic around it. So soft skills do not become less important as delivery is automated. They become part of how AI is steered.
The operating model has to preserve the customer’s intent and the evidence behind each decision across every project boundary, instead of resetting its understanding to zero.
When evidence changes the outcome, record why and who decided. Unrecorded drift is moving the goalposts. Evidence-led, authorised revision is learning.
Use AI on both sides of delivery, so faster execution stays connected to an outcome someone has consciously chosen and can still defend.
