THIS ISSUE

01 What we’re focused on: Doubling down on AI messaging isn’t positioning, it isn’t unique, and it can cost you trust. We’re looking at what four public AI-first bets already cost, and what's left to compete on once expertise is free.

02 An insider POV: I tapped Katie Kregel, SVP of Global Corporate Marketing at IDC (a global market intelligence and advisory firm) for some insight. She’s in a business so deeply embedded in trust, and I wanted her POV on the role of trust in the AI era.

03 Where to look first: The four things a model, and a skeptical human, are both grading before either one trusts you.

IN THE BUSINESS OF TRUST

IDC sells trust for a living. Every report, every ranking, and every "leader" quadrant is a research firm asking a market to believe its numbers over a rival's. IDC recently replaced its old portal-and-PDF delivery model with an AI platform called Quanta: subscribers can now ask a question over email, or inside Claude, and get an answer pulled straight from IDC’s existing research instead of digging through a report.

It’s a faster way for customers to reach the same analyst-produced information, but it raised a big question: does putting AI between a verified analyst finding and the reader change how much people trust the research itself?

I spoke to Katie Kregel, SVP of Global Corporate Marketing, for The Brand Equation podcast last month about the new delivery model. What stuck with me was how deliberately IDC is talking about AI, what’s changing, and what isn’t. Nothing about IDC's methodology changed. The analysts still do the research and the rigor hasn’t shifted.

IDC is betting on AI to save time for customers and free up analysts from routine questions, and that’s that. Customers aren’t left asking what “AI-powered” means, because the answer is clear. A lot of brands are about to have to start making similar distinctions, whether they're ready to or not.

SAME RUBRIC, DIFFERENT GRADER

The E-E-A-T (Experience, Expertise, Authority, Trust) framework hasn’t changed. But trust now takes more. An AI model decides whether it can find enough evidence about you that isn't self-reported. The human's pass, after that, is to decide whether to believe the model.

A model’s fluency makes it sound equally sure about a fact and a guess, which is training buyers to stop trusting confidence as a signal at all. That leaves an opening that a lot of brands haven’t caught yet: naming your limits precisely instead of projecting certainty everywhere.

A model still can’t do that for you. If your brand is willing to discuss its limits, that looks a lot more credible today.

AI-FIRST CAN COST THE MARKET’S TRUST

Within about a year of each other, Duolingo, Shopify, Klarna, and Fiverr each publicly told employees or customers that "AI comes first, headcount second".

Their core numbers didn’t move. Duolingo's daily activities kept climbing straight through its backlash. Shopify's memo didn't dent its merchant base.

The cost showed up in how employees and long-time users talked about the brand, and in a Fast Company survey finding that close to half of professionals have never even heard the term "AI-first," while among those who have, 27% now expect an AI-first company to feel less human and 25% expect a worse customer experience. The damage done was a slow tax on how much benefit of the doubt your brand gets the next time you need it.

After the announcements and the response, the companies didn’t walk back the technology. Duolingo's CEO called his own memo poorly worded and said the company was still hiring at the same pace. Klarna's CEO admitted the company had "focused too much on efficiency and cost" after rehiring the support staff it had publicly credited AI with replacing. What all four have in common: each making a public claim about their values before testing it with the people it would impact most.

Fragmentation is the biggest risk. AI multiplies every surface a brand appears on, so small inconsistencies that used to go unnoticed now show up everywhere at once. At that scale, it stops reading as refinement and starts reading as drift. You can’t fix it with a bigger AI announcement. It’s about staying recognizable everywhere the model might find you, the way a McDonald's fry tastes the same in Tokyo and Texas.

YOU CAN’T MESSAGE YOUR WAY TO TRUST

As AI-generated content floods every channel, buyers are pulling back toward the brands they already believe, rather than fanning out to whoever ranks highest today. When anything can be produced at scale and sound convincing, verifying each individual claim is too expensive to do every time, so trust becomes the shortcut people reach for instead.

That's why brands that already have a strong customer relationship benefit from AI-generated noise instead of losing out to it.

A brand can claim trustworthiness in its messaging, and a brand can actually have it, but those are not the same thing. Trust can't be written into existence. It's a balance that builds from every promise a brand has kept to its customers, one delivered thing at a time.

That raises the obvious objection: doesn't a track record take decades to build? Not necessarily. Age is just a proxy for one thing: enough time to prove you do what you say. A newer brand can earn that kind of faster, not by claiming maturity it doesn't have, but by being unusually specific and honest about what it does and doesn't do yet.

Edelman's 2026 Trust Barometer found "my employer" is the single most trusted institution people have, at 78%, 14 points ahead of business overall (64%) and 25 ahead of government (53%). 70% of people now say they're unwilling or hesitant to trust anyone whose values or information sources differ from their own.

People aren't losing trust across the board. They're narrowing who they extend it to, and getting inside that narrower circle is the competition to enter.

INTEGRATING THIS INTO THE BUSINESS

1. Run a citation audit before you run a campaign

Pick 10-15 real buyer questions in your category, ask them logged out of every AI tool you use daily, and note what gets cited and what doesn't. In half a day, you can figure out which pillar is truly weak versus which just feels weak.

2. Move your strongest proof off your own domain

A case study on your site is a claim. The same numbers in a client's LinkedIn post, a G2 review, or a third-party roundup is evidence a model can verify. Find a way to get your single best result living in at least two places you don't control before the quarter ends.

3. Decide who owns the AI story before your CEO tells it

When a CEO announces an AI shift in a shareholder letter or an all-hands before marketing has agreed on the message, the brand ends up reacting to its own leadership instead of working to shape the story. Get in the room early enough to have a position ready.

4. Investors and buyers want opposite things from your AI story

It’s a tightrope you have to walk. Investors want aggressive efficiency and margin gains. Buyers want to hear that nothing about how you serve them just got cheaper or worse. Decide which audience your external language should be written for, because one message can’t do both.

The full episode gets into what IDC briefed its analysts on before the change went live. It’s worth the listen if you're weighing a similar announcement of your own.

WATCH THE FULL EPISODE