BRANDS VS. AGENTS
It's war, or the death of brand loyalty as we know it, or our response to Citrini
At this point you’ve very likely come across this week’s severely viral, severely consequential Citrini Research piece. A lot of ink and slop has been spilled this week both in favor of and against its core thesis, which was presented in the form of speculative financial fiction.
(Feel free to skip the next paragraph if you already know what we’re talking about. If not, keep on reading.)
For context: Citrini Research is a financial analysis firm focused on thematic investing research with a highly influential Substack. The piece in question, published this past Sunday, is a fictional macro memo written from June 2028, narrating the fallout of what they call the “human intelligence displacement spiral” – a negative feedback loop in which AI capabilities improve rapidly, leading to mass white collar layoffs, spending decreases, margins tightening, companies buying more AI, accelerating the loop. It’s all a “scenario, not a prediction,” and not really all that new: the idea that a (unit of AI capex spend) < (unit of disposable income) circulating in the real economy has been doing the rounds for a while now. Yet Citrini caused a very real market selloff earlier this week. Apparently this scenario wasn’t priced in.
What interests us here is not the macro thesis (though it’s worth your time as well as the numerous rebuttals) or the fact that, if you take its AI capability predictions at face value, we’re very close to the singularity and by extension, likely human extinction.
No, we’re interested in what this means for brands. In particular the mechanism that Citrini describes as the death of “habitual intermediation”.
The concept is simple. Over the past fifty years, advanced service-based economies built a rent-extraction layer on top of human limitations. Most people accept a bad price to avoid more clicks or cognitive labor – brand familiarity substituting for diligence. In Citrini’s scenario, AI agents dismantle this layer entirely. They price-match across platforms, re-shop insurance renewals, assemble travel itineraries, route around interchange fees etc. Agents don’t mind doing the things we find tedious.
Basically what Citrini is describing as the destruction of habitual intermediation is, from the POV of Nemesis HQ, a description of one of the primary economic functions of brands dissolving.
We like to think brands emerge from the sum total of impressions, interactions, expectations, and beliefs an audience has of a thing. Compress this, and you’re left with a cognitive shortcut. Brands reduce the friction of decision-making, but at a cost to consumers.
Think about what a brand does in the consumer economy. You don’t evaluate every CPG on its unique merits; you buy the one you’ve always bought, or the one that signals something about who you are, or the ones that your friends buy. Brands are, in this framing, a technology for managing – and preying on – human limitations of time, attention and cognition. There’s a human tradeoff to ruthless economic optimization. Creating loyalty for your brand is a way to arbitrage it.
So what happens when you hand over these decisions to an agent?
In Citrini’s 2028 world of agentic commerce, brand loyalty is nothing but a tax. When given license to do so, the agent doesn’t care about your favorite app or branded commodity, feel the pull of a well-designed checkout UX flow or default to the app your thumb navigates to the easiest on your home screen. It evaluates every option on price, fit, speed, or whatever other parameters you’ve set. The entire edifice of brand preference built over decades of advertising, design and behavioral psychology becomes, from the agent’s perspective, noise.
Though its not a wholesale “end of brands” it is a catastrophe for every business whose moat is made of friction. And a remarkable number of moats are made pretty much of only friction (think: insurance, financial services, telecom, energy, SaaS, most platforms, Big Grocery, airlines, cars, etc.). Generally speaking, this is a good thing (creative destruction!), and frees the labor of large swaths of branding and marketing professionals to societally more useful ends. It also rewires brands as social coordination mechanisms. In our brand analysis of Chinese handheld gaming device maker Anbernic we wrote:
“In commodified verticals [...] we quickly get to the heart of the matter. What is the true function of a brand? The simple answer is the brand is everything (as it’s the only distinguishing factor and potential source of larger margins) *or* the brand is almost nothing, i.e. a minimum viable coordination point that aggregates (ideally positive) impressions of a product and makes it discoverable.”
After the death of habitual intermediation we’re firmly in brand-even-closer-to-nothing territory as its coordination function needs to be only legible to machines. Again, ignoring questions of social unrest caused by mass unemployment or looming extinction, in Citrini’s scenario we have a consumer economy where a very large portion of products and services become essentially white label commodities, locked into a ruthless price and discoverability competition while agents wage war amongst each other for the best deals.
Which leads us think that as brand value drains from the products and services being transacted, it doesn’t fully disappear but reappears upstream at the agent layer. In a world where your agent handles commerce on your behalf, the relationship that matters is the one between you and your intermediary: your Claude, Clippy, Cluely or Samantha. You need to trust its judgment, taste and alignment with your interests.
And here, ironically, “habitual app loyalty” might return as an extremely powerful force – maybe as the world’s deepest moat. Once you’ve trained an agent on your preferences and developed a sense that it ‘gets’ you, switching to a competing agent carries a significant cost. This is a far deeper ‘brand’ relationship than the one you had with your food delivery app of choice or insurance provider. Yet it’s one where the social function of a brand dissolves as any personal agent (or whatever layer you use to interface with it/them) is only known to a single person.
Quoting ourselves again, this time from H is for Harvey:
“...it might be that brands simply aren’t a relevant framework here anymore. If users increasingly interact with models and agents fine-tuned and contextualized to their habits, vibe, preferences and goals, they drift further away from any coherent notion of brand as a coordination point for consumer behavior.”
If a tree falls… / if no one else knows your agent – does it really have a brand? We don’t know! But in the 2028 Global Intelligence Crisis we’ll have other more pressing things to worry about.