Nick Gaudio has spent fifteen years making B2B companies interesting — head writer at theCHIVE, then brand at Rev, Rattle, ManyChat, and Sweep, which ServiceNow acquired. He also has an MFA and seven Associated Press awards from a stint as a reporter, which turns out to matter for what he argues here.
His position is blunt: the days of evergreen content are over. Assistants already contain the general knowledge your explainer posts were competing on, so nobody is coming to your website to learn what a thing is. They ask their assistant. What survives is a blog that behaves like a newsroom.
Why the evergreen post lost its job
The logic is hard to argue with. These systems hold the entirety of general human knowledge. A post explaining a standard concept in your category has no advantage over the assistant’s own answer, and no reason to be visited.
There’s one exception he’d keep: content whose purpose is to make the models understand what kind of company you are. That’s worth hammering during what he calls a liminal period. But it’s positioning work, not traffic work.
What takes the place of the explainer is the news cycle. At Sweep — which sold metadata security for Salesforce — a breach in the Salesforce ecosystem over the summer became a piece about securing your Salesforce metadata. The event supplied the attention; the brand supplied the angle.
“Build your content engine like a newsroom.”
Constantly looking outward for story-worthy things, then putting them through the brand wrapper. His bet is that both humans and assistants reward it — humans because they want a human take on something that just happened, and models because it’s new information.
The daily story pipeline, concretely
This is the most directly usable thing in the episode, and he does it at every company he joins:
- A free news API, with Boolean queries covering everything related to the company.
- A script to query it — he doesn’t code, so he describes what he needs and has a model write it.
- Google Sheets with Apps Script, so the sheet refreshes every day with story candidates.
- One tab per brand pillar. At Sweep, AI governance was its own tab. When a product needed covering, there was already a fortnight of relevant stories to contextualise it against.
He’s built the same thing for sales enablement — handing sellers current, specific reasons to reach out.
Earn the big change with a small win first
Asked how he’d build a content program from scratch, he doesn’t start with content. He starts with one specific problem the company already has — awareness, retention, feature adoption — and solves that first.
The reason is political rather than strategic. The work he actually wants to do is large: rewrite the messaging, redo the homepage, sometimes the logo. Product teams frequently don’t believe marketing does anything. An early, legible win buys the trust you need before anyone lets you touch the homepage.
Then comes the messaging house. He works mostly on category creation, which means finding the smallest unit of change a person has to accept for the new idea to land — and building everything around conveying that.
He treats content as a function of brand, never the reverse: a planet of brand with pillars about what was wrong with the world that needed changing, and content as moons orbiting it. The founding narrative has to carry genuine obsession. His analogy: you’d rather eat at the restaurant run by a chef obsessed with a cuisine than one opened to make as much money as possible.
And it has to be compressible. People can’t hold nuance — every layer you add is another day of their memory you’re asking for. Fifteen words or fewer: here’s the problem, here’s the solution, we’re obsessed. Everything else you publish exists to prove the obsession is real.
Interesting, believable, memorable
His filter for any piece of work, which he shortens to IBM. Interesting breaks the pattern; memorable makes it stick; believable is the one people skip.
He makes the point with a thought experiment: he could send his list an email with a deliberately offensive subject line and watch the open rate spike. That’s the end of the game, not the start of one. Attention has to be part of a mechanism that runs all the way through to the outcome, or you’ve spent trust to buy a number.
This is also why he’s changed his mind about repurposing. He used to make one big asset and break it into constituent parts. Now, with so much generated sludge in every feed, he’d rather be razor-sharp about the culture of each platform — LinkedIn and X aren’t the same audience in a different font; what a person is willing to be seen liking differs between them.
The corollary is not to betray the reader. Disguised ads that mimic the format of the platform until you notice they’re selling insurance read as a trick, and he thinks people are getting faster at spotting manipulation, not slower.
Deterministic versus probabilistic
The line that I think most SEO people should sit with:
“SEO is deterministic. This is probabilistic.”
His read is that the current moment — where assistants lean heavily on search infrastructure — is interstitial and won’t hold. What he’s betting on is that AI companies will end up rewarding the same thing Google eventually rewarded with site speed: whatever makes their own product better. In this case, sources that are consistently true, evidence-backed, and grounded in first-hand expertise.
That lines up with what Anastasiia Krasiukova found testing human against generated content — the differentiator is access to someone who actually knows something.
How Sweep got cited four times more often
Working as a team of one, he lifted Sweep’s visibility fourfold across a hundred discovery prompts. Two plays did it, and neither was a standard SEO play.
Novelty. An episode of South Park about AI sycophancy became an article about Salesforce’s AI tooling. The juxtaposition was strange enough that it became their most-read piece with humans, and then started appearing as a source in AI answers. He’s honest that he can’t fully separate cause from effect — people were sharing it, which may be what the models picked up, or the novelty may have been enough on its own.
Proprietary data. Their AI assistant handled roughly 15,000 interactions a month, and he mined them:
- Bucketed the questions by stage — discovery, design, implementation, management, governance.
- Found that 17% were asked after midnight local time, with design questions spiking after hours because that’s when nobody was interrupting them to ask what was wrong with the CRM.
- Paired it with a survey on how long each problem took to solve without the tool, which produced the comparison that mattered: minutes instead of hours.
It became a report of around sixty-five pages, with several thousand downloads. Two details worth stealing: product was nervous that publishing the questions implied the product didn’t answer them (it didn’t read that way), and because security people were downloading it alongside the RevOps audience, they tested a humorous version against a straight one — on the theory that a CISO gives you about six seconds.
Underneath both plays is one idea. You’re giving the model things it cannot know, because it has no eyes or ears inside your product. That’s the edge available right now.
Brand sits above all of it
Asked where brand fits relative to content, PR and reviews, his answer inverts the usual hierarchy: all of those are subsets of brand. Brand is where the company meets reality.
“A company’s existence is an argument.”
His favourite example of brand that isn’t content: Anthropic hiring a philosopher to work on AI safety. The claim is “we’re the safe AI company,” and the hire is the claim backed with real resources where the world can see it.
He contrasts two versions of his own work at Rev. Early on, people searched for transcription, clicked, paid, and forgot the company existed — a transactional relationship with no brand attached. Then he started writing a newsletter with actual jokes in it, and people wrote back to say they loved it. He kept a file of those replies; it ran to hundreds over two years. Asking that audience for a review was trivial, where a gift-card bribe would have been necessary before.
And the forward-looking version of the same instinct: with assistants in the middle, the valuable thing is knowing what your buyer doesn’t know to ask. If they can’t phrase the question, they can’t get the answer — and you can be the thing that sits between them and the assistant.
Where this ends up
He takes the pessimistic case seriously, by way of Maggie Appleton’s dark forest theory of the internet: as more of the web turns out to be bots with something to sell, trust frays, and people retreat to places where everyone is verifiably human. Walled gardens, Slack and Discord communities, and above all in-person events. He thinks even video stops being proof fairly soon.
Which produces the best career advice in the episode. If someone told him they were going into email marketing, he’d tell them to go and learn how to throw a party instead.
FAQ
Is the company blog dead? The evergreen explainer is. A blog that covers what’s happening in your category, from your point of view, isn’t.
What replaces evergreen content? Newsjacking plus proprietary data — the two things a model can’t generate for itself.
How do you find stories every day? A news API with Boolean queries per brand pillar, piped into a Google Sheet that refreshes daily.
Why does proprietary data help with AI visibility? It’s information the model cannot have. At Sweep it contributed to a fourfold lift across a hundred discovery prompts.
The takeaway
If your content plan for next quarter is a list of explainer topics, that’s the part to question. The two things with leverage are a story only you can tell, because the data comes from inside your product, and a reason for anyone to care this week.
Neither requires more volume. Both require someone with judgement and access.
That’s the kind of thing I’d look at first in an organic program. If you want a straight answer about yours, book a call — and for the full conversation with Nick, watch the episode or connect with him on LinkedIn.