AI Agent Trust: Why Marketing Claims Don't Work
Why AI agents can't be sold on a brand, only shown proof. Part three of a three-part series on trust.
Two weeks ago, the data made the case: Canadians trust a billboard more than they trust their feed. Last week, a framework showed how that trust actually breaks down inside an organization, one comment, one unanswered question, one point of friction at a time. This week's question is the one sitting underneath both of those: when the thing deciding whether to trust your brand stops being a person, what happens to trust at all?
Rob Schlyecher, Founding Partner and Creative Director at Spring, has been writing about that shift for months. His starting point is a distinction he calls functional trust versus marketing trust, and it's less abstract than it sounds.
Functional trust vs. marketing trust
“Trust is built on how the user experiences the product," he says. “If it's an attribute, feature or level of performance that a customer can depend on while using the product it is creating functional trust. If it's a claim, slogan, or otherwise that isn't directly related to what the product actually delivers then it's the far more debatable status of marketing trust."
He argues the distinction matters more than it used to, because the two are getting harder to keep separate. “If marketing trust contradicts functional trust, AI will accelerate the process of your brand getting found out."
The audience that can't be bought
That acceleration has a name in Schlyecher's world: Agentic Endorsement, what happens when an AI agent recommends a brand based on verifiable performance rather than paid placement or SEO. It's a shift he compares to a familiar fable.
“To borrow an oldie, this is Emperor's New Clothes material," he says. “Marketers have grown used to purchasing approval. That has existed in many ways, not just from influencers or paid social, but from paid SEO. But that's not where AI agents search, at least not all the time."
The uncomfortable part for marketers used to buying their way into visibility is that this particular audience can't be bought. “You simply can't pay for agentic endorsement. You must find and proliferate verifiable performance attributes and communicate them to the right markets," he says. And the target keeps moving. “AEO performance changes as fast as AI grows."
Picture the version of this that's already happening. Someone asks an AI assistant to recommend a reliable cordless drill, and the agent isn't cross-referencing sponsored listings, it's pulling from reviews, forum threads and product data to find what actually holds up. No media buy changes that answer. The only lever left is whether the product earns it.
That framing lines up with what Scott Mitchell told MNC in part one of this series: 43% of Canadians say they're less likely to trust an ad once they know it was AI-generated. It also echoes what Christina Garnett described in part two, brands losing ground to competitors who can now replicate a feature or a UX improvement almost overnight. Three different vantage points, one shared conclusion: when the gap between competitors closes that fast, and the audience evaluating you increasingly isn't human, trust stops being a nice-to-have and becomes the only thing left to differentiate on.
Start with an audit
So what does a marketer actually do with that this quarter? Schlyecher's answer starts with a word most brands associate with compliance, not strategy.
“Audit," he says. “You need to know how your site is performing in AEO and figure out where you are failing in the areas that you can't market your way out of." From there, he says, the work widens out. “You need to do a full audit of markets and your product's competitive advantages and disadvantages."
That kind of audit tends to surface a specific, uncomfortable answer: a feature everyone assumed was a differentiator, but that competitors quietly matched months ago, or a claim the marketing team still leads with that customer support tickets have been quietly contradicting for just as long. Those are the gaps an AI agent will find eventually. Better to find them first.
It's not a glamorous starting point, but it's a concrete one. The next audience your brand needs to earn trust from might not read your ad copy, browse your site or scroll past your post at all. It'll audit you, and no amount of marketing trust will talk it out of what it finds.
