"Agentic content" creates a sticky narrative
Everything used to be "content". Now it's increasingly "agentic" (or some combination of the two if you're lucky). But do these words actually mean anything?
I distinctly remember when everything became “content”, rather than just “posting on social”. It was 2013. I’d handed my notice in at Launch PR, ready to move to We Are Social. Daft Punk were riding high with “Get Lucky” on repeat everywhere. Arctic Monkeys just entered their quiff-topped imperial phase. Vine was the cool kid on the block (its iOS app launched in January of that year), and Instagram would soon add video as a direct response.
Over the space of a few months, the previous low-effort social approach of tweeting interesting links to articles would seem dated and fusty. Suddenly, if you weren’t making “content” (branded images, GIFs, videos), you were nowhere.
The 2013 narrative around “content” has proven remarkably sticky. You can find plenty of 2026 date-stamped advice that supports the notion that anyone can use “content” to build an online audience, an audience that will ultimately drive some kind of business outcome. It’s easy: all you need to do is optimise your posts, tap into what’s trending and utilise the right platforms and, hey presto, you’ve achieved marketing success.
And the narrative remains sticky because of its popularity, and also because it does actually work sometimes. You can have a viral hit; it’s just extremely unlikely. But it’s true that you could strike algorithmic gold with Reels or TikTok. The proof that it works sustains the mythical narrative.
That narrative shows no signs of slowing down, as evidenced by the new generation of genAI-powered start-ups springing up, particularly in San Francisco. Yes, the core idea of the product plays a part for many of these companies. But actually, their content marketing approach, and the associated potential virality, are arguably even more important.
Take the case of Cluely, the viral AI start-up that rose to prominence thanks to its claims that it can help you “cheat” your way to success. The story was the interesting part - the “personal mythology” around founder Roy Lee, as outlined in this memorable piece by Sam Kriss in Harper’s. Lee scored a viral YouTube hit with footage of himself cheating his way into an internship at Amazon (he declined the internship because that was the punk rock option).
When Kriss visits Lee at the Cluely offices, he finds that employees (including Lee) devote as much time to producing “viral content”(see example above) as they do to working on Cluely the product.
Indeed, when Kriss tries to use Cluely to “cheat” his way through the interview with Lee, the product doesn’t work. And when it gets working, it delivers entirely underwhelming results - and Lee seems less than bothered by this?
The Cluely example is one of many stories coming out of the world of viral start-ups, where the sell and the virality disguise the emptiness of the idea. Prime example being that viral pin you could wear which would record your life - which didn’t work. Big, well-funded ideas that, when you strip away the layers, are simply just a customGPT running on ChatGPT.
And it all comes back to the promise of the content marketing machine - make enough content, have enough killer ideas, and one will stick and generate riches. Doesn’t matter if the idea doesn’t work or if you don’t really believe in what you post. The virality is the goal - the virality generates the returns, either directly from platforms or indirectly via investment into your product. And investors don’t appear to mind - the new AI goldrush relies on just one hit to justify the outlay.
The challenge with the reliably sticky narrative is that it only works 0.08% of the time (that’s the percentage of YouTube videos that reach a million views, which maybe isn’t even that viral anymore?). The rest of the time, your efforts to go viral, to convert content marketing into leads and sales will fail miserably. The story of “content” through the 2010s often overlooks an equally important trend - the rise of democratised advertising tools.
The vast majority of content marketing works when there is paid spend to ensure what you produce actually reaches the audiences you care about. Content is important not because it sometimes goes viral; it’s important because it provides the assets you need for effective social media advertising.
But that’s boring, and requires budget allocation, and so the narrative of “content” and its organic power persists. But we’re perhaps beginning to see the rise of a new iteration of this narrative. Because even though the playbook for many of the new wave of genAI-powered start-ups is more of the same (big idea that kind of sort of works, plus content to promote it), we have a new buzzword on the block that’s gaining momentum while losing meaning in equal measure.
I’m talking about “agentic”. It seems that, despite agents not quite becoming the mass-market product we envisaged last year, the concept of “agentic systems”, “agentic processes”, and simply being “agentic” has caught on in a big way.
The Sam Kriss piece I mentioned before lays out an example of this application clearly:
“Highly agentic are people who just do things. They don’t timidly wait for permission or consensus; they drive like bulldozers through whatever’s in their way. Agency is now the most valuable commodity in Silicon Valley. In tech interviews, it’s common for candidates to be asked whether they’re “mimetic” or “agentic.” You do not want to say mimetic.”
You might reasonably ask what’s new about being “agentic” - the whole “just build it” ethos has been decidedly mainstream ever since Eric Ries published The Lean Start-up way back in 2011. Make an MVP (minimum viable product), get some users, keep going. Or don’t get any users, then “pivot” away to something else.
The premise in 2026 is the same - except, I guess, that you don’t even really need to be able to code to build something now. You can vibe code or use ChatGPT’s Codex to whip up your MVP in record time. But as any sensible person with any knowledge of technology will tell you, the vibe-coded MVP will only get you so far - at some point, you need to get serious and actually make something that works.
For the “highly agentic” crowd, that part feels secondary to the content marketing initiatives. And a bit too much like boring hard work. As Kriss puts it,
“As far as I could tell, being a highly agentic individual had less to do with actually doing things and more to do with constantly chasing attention online.”
The challenge with the narratives encapsulated by “make more content” and “be more agentic” is that the slightly ridiculous newsworthy headlines obscure the fact that both approaches do have value.
You can build awareness, generate leads and drive sales with a well-thought-through content marketing plan, but it takes time, effort and financial investment. You can build some interesting prototypes using Codex and Claude Code, which, with time and financial investment, could help automate processes or replace outdated legacy systems that were too expensive to mess with. But success in these fields doesn’t come from “bulldozing” through anything - in plenty of agencies, your IT restrictions might make some of this literally impossible.
“Being more agentic” is, to steal Katherine Dee’s phrase she used this week, yet another “costume” the ambitious wear to capture attention online. Katherine was talking about conspiracy culture in her piece (highly recommend), but you can apply it just as easily to the content grifters and the agentic AI uber-evangelists: “They are wearing the clothes, and…the clothes come off easily... A provocation on Monday that they’ll walk back by Wednesday.” For provocation, read business idea or content series.
The takeaway for comms people from the continued popularity of the content myth, and the rise of “being more agentic”, is to beware the buzzphrases. If it seems too good to be true - free marketing, simple automation of your processes - it probably is. You might, to go back to 2013’s biggest tune “Get Lucky” once or twice, but those are the exceptions that prove the content marketing rule: success is hard.
If you’re meeting a supplier that promises an “agentic” solution, spend time unpicking what that actually means - and potentially ask “is this just a custom GPT?”.
Building products and campaigns takes time. Success invariably comes from doing the boring things consistently, even when you don’t want to. I practice guitar when my wife and daughter are in bed - I often don’t feel like it, but I talk myself into “just five minutes”. Invariably, I end up playing for an hour and feeling satisfied. That kind of approach isn’t going to generate any viral hits (man attempts to play the same Smiths song for an hour, not really getting anywhere). But experience tells me that’s where success lies.