The speed is the blessing
Something is happening inside the businesses we work with, and on the surface it looks like exactly what everyone hoped AI would deliver. People are building. Apps that would have taken a 3 months to scope are appearing in a fortnight. Internal tools, customer-facing services, whole websites, assembled by people who a year ago would have needed a development team and a budget line to attempt any of it. The capability has been handed out to everyone, and everyone is using it (or starting to).
The title of this piece isn’t a comment on the people doing the building. They’re often the sharpest people in the business, which is precisely why they got there first. The monkey in the metaphor isn’t about intelligence. It’s about what happens when the power to act arrives long before the context needed to aim it, and that gap is the thing worth talking about.
The doing is now the thinking
Here is the shift underneath all the building. It used to be that the slowness of making things was where the thinking happened. A new customer portal took months, and somewhere in those months someone from legal read the terms, someone from security asked where the data was going, someone senior asked whether the thing was worth doing at all. Nobody designed it that way. The friction just created the room, and the room is where the organisation’s knowledge got applied.
AI has removed the friction, and the room went with it. When the build takes an afternoon, the question of whether it should be built rarely gets an afternoon of its own. The doing has become the thinking, in the sense that the first serious consideration of a thing now happens by looking at the finished version of it. And a finished thing has momentum. It has a demo. It has an owner who is proud of it. The moment for asking whether it handles customer data properly, or makes claims the business can’t stand behind, or quietly breaches a licence agreement, arrives after the moment it would have been easy to act on the answer.
The knowledge you just walked past
What makes this genuinely risky rather than just untidy is what the builder doesn’t know they don’t know. When someone prompts an AI to build a customer sign-up flow, the model will cheerfully handle the consent language, the data storage, the terms of use. It will produce something plausible for all of them. But plausible is the trap. The person prompting usually has no way to judge whether the consent language actually meets Australian privacy requirements, whether the storage arrangement would survive a security review, whether the terms create obligations the business never agreed to carry. They don’t know the right thing to ask for, so they can’t tell whether they got it.
Meanwhile the people who could tell, the compliance manager, the security lead, the lawyer on retainer, the operations veteran who remembers why the last version of this idea was shut down, never saw it. The organisation’s accumulated knowledge wasn’t overruled. It was bypassed, at speed, by someone acting in complete good faith with a tool that never says “you should check this with someone.”
That knowledge is not paperwork. It’s the residue of every mistake the business has already paid for once. Bypassing it doesn’t make the risk disappear. It just moves the discovery to a worse moment.
Aim the speed at the room, not around it
None of this argues for taking the tools away, and it certainly doesn’t argue for a committee between every employee and every prompt. The speed is real value. The mistake is spending it all on production and none of it on the conversation.
The better move is almost the reverse of the current pattern. Use AI to make the thing quickly, then use the time you just saved to put it in front of the people whose knowledge used to be built into the slowness. A working prototype is the best briefing document legal has ever received. A live demo gets a sharper answer out of security in twenty minutes than a specification ever did. The speed of AI can pull the organisation’s knowledge into the work earlier and more concretely than the old process ever managed. It can gather the room, faster, around something real.
That’s the discipline worth building while the tools are still new: the habit that speed of production buys time for judgement, rather than replacing it. The businesses that get this right won’t be the ones that built the most things this year. They’ll be the ones where the building brought the team together around each thing worth doing, and where the knowledge the organisation spent years paying for still got its say. The gun isn’t the problem. Knowing what you’re aiming at was always the job.