Last night we ran the second edition of AI Marketers Anonymous in London. Same format as the first: senior marketers and CMOs in a room, no vendor pitches, no keynote polish, no pretending anyone has this figured out. Very much a working/therapy session for people who are quietly knackered by the pace of change and tired of LinkedIn telling them they're behind.
Here's what was actually on marketers’ minds this time.
The job question is louder now than it was a month ago
Last month, the conversation circled around "what is the agency for?". This month, it sharpened into something more uncomfortable: “What are the people for?” Layoffs at agencies are no longer hypothetical, and the room split on how to read them. One camp framed AI as replacing a bundle of tasks rather than whole roles, drawing the obvious parallels to ATMs and spreadsheets, both of which created more work than they destroyed over a long enough horizon. The other camp pointed out that they're personally watching tools do in minutes what used to take a team a week, and that "it'll all balance out in ten years" is cold comfort if you're running a P&L this quarter.
The more obvious and, in fact, useful framing landed on retraining. The companies treating AI as a cost-cutting exercise are getting short-term margin and long-term mediocrity. The ones treating it as a way to give existing people a superpower are quietly pulling ahead.
There are no experts, only people who've experimented more
What separates useful operators from confident ones is hours on the tools, plus the underlying domain knowledge to know when the output is good. Prompt engineering only works if you already understand the “laws of physics” of the thing you're prompting about. Brand building, media planning, data analysis - the true fundamentals of marketing still matter! AI amplifies the marketer underneath it.
Rent is coming due
CIOs and marketing leaders are drowning in AI startups. Every category has fifteen tools, half of which will not exist in eighteen months and we’re all trying to use the latest one and burning tokens which we haven’t yet truly understood the cost of yet.
The frontier labs are not currently profitable. Prices are going to go up, and the supply chain (energy, data centre capacity, chips) is a real physical constraint, not a software one. The current pricing is a land grab. That big rent bill is due.
Physical is the new premium
Tangible things, physical events, hard-copy reports, and hand-written notes, are doing disproportionate work right now precisely because AI can't fake the effort.
This is the same dynamic we touched on last month with budgets shifting to experiential and creator-led work. It's a rational response to find arbitrage in the things which AI can’t do.
Distinctiveness is the moat, and AI is pulling the other way
AI is trained on what already exists, so it produces what already exists, slightly remixed. Pointed at brand work, it tends to regress everyone toward the category average. The brands that win are the ones that are meaningfully different, and meaningful difference still seems to require an original human point of view and some emotional risk.
The compliance and ethics layer is starting to bite
First, AI-automated SEO at scale is starting to get punished by search engines, and the penalties can be effectively permanent. Cheap traffic now, no traffic ever. Second, the proliferation of AI meeting recorders and transcript tools is creating a new category of workplace ethics question that most companies haven't thought about.
Where the room left it
The same place it left it last month, honestly. Theory is settled. Implementation is where everyone is stuck. The specific stuck-ness this time was about pace: how do you invest in learning a tool when the next model launch invalidates the workflow you just built? The rough consensus was to stop optimising for tools and start investing in judgment, taste, and the human skills that compound regardless of which model is on top this quarter.
Whilst the rest of the marketing world has been shouting about AI for the past year, we at Rival have been building. Testing, learning, and figuring out what an actually AI-native marketing consultancy looks like in practice. The TL;DR is the same as last month: it's a marathon, not a sprint.
If you were in the room last night, thank you for making it the kind of conversation that doesn't happen on a stage. If you weren't, we run this regularly. The next one is worth your evening.
