What Survives the Prompt
Something has shifted in the rooms I sit in. I am still working out what to do about it.
TL;DR: AI has flattened the knowledge layer in the screen industry. The briefing documents are better. The judgement in the room has not improved. I have been thinking about what that gap actually means, for how expertise works, for what operators are actually selling, and for what is worth building now.
I have been in a handful of pitch and strategy sessions over the past year where something felt slightly off and I could not name it until recently.
The preparation was good. Better than it used to be, honestly. Comps were clean. Market context was accurate. Someone had clearly done their homework.
But when the conversation moved somewhere the preparation had not anticipated, things got thin quickly. The kind of thin that tells you the research was assembled rather than understood.
I do not say that to be uncharitable. I have assembled research myself. I know what it looks like from the inside.
What I kept noticing was the gap between the quality of the document and the quality of the thinking in the room.
And it made me start asking a question I am not sure I have fully answered yet: what is expertise actually made of, and how much of what I thought it was has quietly changed?
The part AI actually flattened
The honest answer is: more than I initially wanted to admit.
A lot of what passed for expertise in development, distribution, and brand content strategy was information access. Knowing which deals had closed. Knowing which platforms were actively buying in a specific format. Knowing the right precedent to cite in a negotiation.
That knowledge had real value because it was genuinely hard to come by. It took time. It took relationships. It took being in enough rooms that the patterns started to accumulate.
Now a decent AI tool produces a reasonable version of that briefing in twenty minutes.
I do not think that is a disaster. But I do think it changes what the actual differentiator is, and I am not sure the industry has caught up with that shift yet. Including, sometimes, me.
What I notice does not transfer
There is something that happens in certain conversations that I have not found a way to prompt for.
It is the moment where someone reads a commissioning signal not from the data but from what a particular executive did not say in a particular interview.
Or where they look at a co-production structure and know, from having watched similar structures fall apart, exactly which clause is going to cause the problem.
Or where they sit with a brand partner and understand within the first twenty minutes whether the content ambition is real or whether it is a budget reallocation dressed as strategy.
I do not know a clean name for that. Pattern recognition is close but not quite right. It is more like pattern recognition under conditions where the pattern is incomplete and the cost of misreading it is real.
That accumulates differently to knowledge. It accumulates through repetition and through being wrong in ways that mattered. Through deals that closed and deals that fell apart and the specific texture of understanding why.
I think that is what is actually getting repriced right now. Not intelligence. Not even experience in the general sense. Something more specific than that.
What I am still working out
The APAC context makes this sharper for me personally, because a lot of the markets I work across, Southeast Asia, India, Australia’s co-production infrastructure, are markets where the documented record is thin. The AI tools are trained on what has been written down. Much of what matters in these markets has not been written down.
So the gap between the briefing document and the actual deal is wider here. Which means the operator who has worked inside these markets carries something the tool genuinely cannot replicate yet.
I say “yet” because I do not think that advantage holds forever. The documentation will improve. The tools will catch up in some of these areas.
But it does point to something I keep coming back to: the expertise that ages best is not the kind built through aggregation. It is the kind built through doing the thing, inside the actual conditions, with real consequences attached.
I am not entirely sure where that leaves me, or anyone else trying to figure out what they are actually selling in a world where the knowledge layer keeps flattening.
But I think the question is worth sitting with. Because the people who answer it honestly, before the market answers it for them, probably end up in a better position than the ones who assume the old signals still mean what they used to.
That is roughly where I have landed. Well for now.
Though I suspect the more honest version is that I have not fully landed at all.
I keep returning to it in different forms, in different rooms, with different people. Which probably means it is the kind of question that does not resolve cleanly.
Maybe that is the point. The questions worth sitting with rarely do.
I am curious what you are noticing in your own rooms. Reply and tell me.