James Cadwallader

The through-line

Cadwallader's consistent argument is that the internet's front door looks the same while everything walking through it has changed: consumers who once clicked through to websites now get their answers from models, and that makes this, in his framing, the biggest platform shift in marketing history [4]. He treats large language models as more than a new interface layer sitting on top of the old web [2], and he dates the practical opportunity to the moment models became web-connected, which is what made answer engine optimization a real discipline rather than a theory [3]. Everything else he says follows from that premise: traditional SEO is under existential pressure [5][6], brands now build content for machines acting on behalf of people [7], and AI is becoming a brand's most influential referrer [9].

The second half of his thinking is about what the marketing function has to become in response. He is not describing a defensive adaptation. He expects marketing to be rebuilt around agents and around a new hybrid role, with output scaling by orders of magnitude rather than percentages, and he says the marketing of today will look "like a cottage industry in comparison to where it ends up going" [1].

On the platform shift

The comparison he accepts is to the mid-2000s: a change in how people spend their time, shop, evaluate and purchase, which opens a window for whoever matches their approach to the new behaviour [1]. Where consumers once clicked, they now ask, and the consequence is that AI search is becoming the default interface for information retrieval [3][4][8]. He argues this puts every brand on the planet in scope, not just the ones with sophisticated search teams [2], and he has made the case with practitioners on the retail side that the SEO craft the industry spent two decades mastering may not survive in its current form [5][6]. His own company's positioning came from seeing the shift early: "we we we just about saw this market shift start to happen and then we built a business very quickly that captured um that captured it" [1].

On what visibility now means

The practical question he keeps returning to is how a brand stays visible when a model, not a search results page, decides what gets surfaced [3][8][9]. His answer runs through answer engine optimization, which he ties directly to models gaining web access [3], and through the shift from producing content for human readers to producing content for machines operating on people's behalf [7]. He has described this work concretely through brands using Profound to gain visibility and leverage inside models, including Eight Sleep and MongoDB [9], and he frames the endpoint as agent-led growth rather than search-led growth [4].

On the marketing engineer

He is emphatic that this is a real prediction and not positioning: "it's not a uh it's it's not a marketing tactic. We we we genuinely believe that there will be a new role in the marketing org called the marketing engineer" [1]. The role is "part technical, part very human," converging creativity with technical ability, and its job is to "build and deploy and customize agents that can do work on their behalf and let them do a better form of marketing, a a new kind of marketing, marketing that wasn't possible before LLMs" [1]. Crucially, these people do not need to be engineers in the classic sense: "certainly not coding wizards," because agents can already be built in natural language, with voice as the next step, "the highest bandwidth upload which is voice" [1].

Structurally, he places the marketing engineer horizontally rather than inside one silo. It "touches all parts of the marketing org," speaking to product marketing, growth and performance, brand, and PR and comms, equipping each of those existing functions with agents [1]. As for whether teams get smaller or bigger, he declines to prescribe: "you can either do more with less or you could do more even more with more," and calls it a philosophical choice for marketing leaders, while noting the pull of Bezos's two-pizza team and that "big teams become a problem" [1].

On agents being insatiable

His case for scale rests on demand from the machines. "I think we might end up doing a 100X or a 1,000X more marketing in the future with the same size team," he says, "cuz agents are insatiable. Agents want more and more information and you it falls on the marketing team to create and distribute that information" [1]. He is careful to say this is not hyperbole [1]. The corollary is that human judgment gets more valuable, not less: marketers have "this rare blend of creativity and operational excellence," which is "not going anywhere at all," and in a world where anything is possible because of AI, "your ability to create a strong brand and to communicate your story well comes arguably more important" [1]. What changes is the toolkit, in an industry he describes as having been stubbornly manual, where "most marketing took place has taken place in Google Drive, in Google Docs, in spreadsheets" [1].

On personal productivity and being AI-native

Asked how much more productive he is than in his previous startup, he answers "a hundred times more productive, not due to any genius on my part, but due to AI" [1]. He attributes it to having started the company after LLMs existed, so AI is native to daily work rather than retrofitted: calls transcribed through Grainola, follow-ups pre-drafted in his inbox before he is back at his desk, with his own role reduced to editing. He is clear it does not make the output impersonal, and estimates one such follow-up would have taken 45 minutes before [1].

On how the company is run

The org is flat and deliberately so, with roughly 12 or 13 direct reports, which he concedes is a lot for the company's size [1]. His model is "flat and high agency. We allow people to cook," and the operating principle is that "if you bring in really talented people, and you give them ton of tons of context, tons of agency, you know, they're able to do really incredible things," with AI helping because "it allows you to transfer ideas faster" [1]. He stays in the interview loop while hiring at scale across functions by compressing interviews to ten minutes, which he defends on evidence: after ten minutes of conversation, both parties have enough context [1].

The cost of speed shows up as a second-order problem he names directly. Product ships so fast that internal teams and customers struggle to keep pace: "we have customers who log into the product and they're like, 'Oh my god, it will change,'" which he calls a champagne problem while acknowledging it drives customers crazy [1].

On luck, timing and the co-founder choice

He resists a purely meritocratic account of the company's trajectory. Asked to split success between luck and skill, he leads with "lots and lots of luck," starting with meeting his co-founder Dylan, and observes that "when you build a company, you can choose many things, but it's what you only really get to choose your co-founder once," crediting "the craft and the obsession that Dylan brings to the table" [1]. He extends the luck to market timing and to having built post-LLMs [1]. The honesty cuts both ways: "in the same intellectual honesty we've worked our asses off" [1]. Twenty months in, he says the scale of attention was "beyond our wildest imagination" and explicitly not the product of strategy: "it wasn't part of some Machiavellian strategy that we had" [1].

Takeaways

• The premise behind everything he argues: the front door of the internet is unchanged but the visitor has changed, making this the biggest platform shift marketing has seen [4], with AI search becoming the default interface for information retrieval [3][8].
• Answer engine optimization became viable specifically once LLMs gained web access, which is what turned AI visibility into an addressable discipline for brands [3].
• He predicts a new role, the marketing engineer, "part technical, part very human," sitting horizontally across product marketing, growth, brand and comms, and building agents in natural language rather than code [1].
• Output, not headcount, is where he expects the change: "we might end up doing a 100X or a 1,000X more marketing in the future with the same size team," because "agents are insatiable" [1].
• Human creativity gets more valuable as production costs collapse: the marketer's blend of creativity and operational excellence is "not going anywhere at all" [1].
• His operating model is a flat org with 12 or 13 direct reports, high agency, and ten-minute interviews so he can stay in the hiring loop at scale [1].
• He claims to be "a hundred times more productive, not due to any genius on my part, but due to AI," on the strength of building the company post-LLMs and editing rather than drafting [1].
• He credits much of the outcome to luck, particularly the co-founder choice: "you only really get to choose your co-founder once" [1].