“I think the idea of search was a construct defined by the UI,” Simon Poulton told me.

That line has been living rent-free in my head ever since our interview — mostly because it’s such a brutally simple way of describing how marketers built their entire religion around a box.

For twenty years, the entire advertising industry has been yelling into that box like lunatics — “yoga pants!” “CRM software!” “best pizza near me!” — believing that the right combination of keywords would summon the gods of visibility. It worked, sort of. But it was a hack — a linguistic illusion that turned human curiosity into algorithmic obedience.

Now, that box is gone.

Search, as we knew it, isn’t dying. It’s mutating. It’s becoming conversational, contextual, agentic.

Here’s the uncomfortable truth: humans aren’t your audience anymore.

Brands are now talking to machines that talk to other machines. AI agents are the new intermediaries — digital emissaries that don’t just answer questions, but negotiate outcomes. They fetch, compare, transact, and decide before a human even blinks.

And yet, marketers are still out there buying CTR dashboards like it’s 2015, hoping their next “keyword optimization sprint” will save them. It won’t. Because no one is typing keywords anymore. They’re asking — and the thing answering isn’t a person. It’s a system.

Poulton said it perfectly: “You’ll have all kinds of agents operating with you, operating in various different ways, maybe even talking to each other.”

That’s not a metaphor. That’s literally what’s happening. ChatGPT talks to Google APIs. Gemini checks Wikipedia and shopping feeds. Perplexity cross-references PDFs. They’re forming an invisible supply chain of answers — and your brand is just one dataset in that chain.

If your information isn’t machine-legible — structured, signed, and current — you’re invisible to the only audience that matters: the algorithmic middlemen deciding what’s worth showing.

Marketers used to obsess over brand voice, color palettes, and witty copy. Cute. That still matters for people, but here’s the punchline: most of your “people” now meet your brand through machine intermediaries.

The “content” that wins isn’t the prettiest — it’s the most parsable. A brand that publishes a schema-rich product feed with accurate metadata will outrank your tear-jerking 60-second campaign every time.

In an agentic world, you don’t win with persuasion. You win with clarity.

Clean APIs. Fresh data. Consistent provenance signals.

Your “copy” isn’t for humans anymore — it’s for systems deciding whether you’re credible enough to surface in an answer.

The tragedy is that most marketers still think they’re selling to consumers. They’re not. They’re selling to the consumer’s agent.

Here’s where things get weirder.

Agents aren’t neutral. They have memories. They cross-check data, build trust scores, and develop reputations for which brands are reliable. That means your mistakes — broken feeds, outdated prices, contradictory policies — don’t just frustrate users; they degrade your machine reputation.

Every inconsistency is a credibility scar that your next AI buyer remembers.

Welcome to algorithmic reputation management, where the systems keep score — and they never forget.

If this sounds dystopian, it’s not. It’s just overdue accountability. For years, marketers played fast and loose with truth, hiding behind dashboards and PR spin. Now, the machines can tell when you’re lying.

When Poulton said the UI defined search, he wasn’t lamenting the death of a box. He was diagnosing an entire industry’s addiction to visible inputs. The interfaces — the clicks, the rankings, the keywords — were training wheels. They gave us the illusion of control.

Now, we’ve been shoved into a world that runs on invisible logic: model weights, schema accuracy, cross-agent trust. You can’t “optimize” your way into that. You have to engineer for it.

That’s the new work.
Not storytelling. Not copywriting. Not “content strategy.”
Data architecture as persuasion.

Marketers spent years building for people who typed. The next decade belongs to those who build for systems that interpret.

If your brand can’t be read, parsed, verified, and trusted by machines, you might as well not exist — because the new intermediaries won’t even know you’re there.

Search didn’t die. It evolved. And as Simon Poulton reminded us, it’s no longer about shouting into the void. It’s about teaching your data how to speak for itself.

Because in the age of agents, the real question isn’t who’s searching for you — it’s what’s searching for you.