They're not coming alone anymore
I've written before about the boss who banned the word "consumers" from our office. It made people sound like they'd eat whatever we put in front of them. I've thought about that boss a lot this week.
Because the people we serve aren't swallowing what we hand them anymore. They're showing up to the conversation with backup. An AI agent that has read your return policy, your FAQ, three Reddit threads and a review site, and has its own opinion about what you owe them.
And sometimes that agent says something your company never said.
The numbers just came in
Capacity, a customer-support automation company, surveyed 1,000 U.S. consumers for its new Agentic Consumer Report. A few findings stopped me:
- 35% have already used an AI agent to handle a customer service request. Among Gen Z and Millennials, it's 49%.
- 44% say AI contradicted a company representative or the company's own policies.
- When the AI and your rep disagree, consumers are equally likely to trust either one.
- 36% would use AI to research return, refund or cancellation policies before they ever contact you.
- 23% would ask your rep to explain or cite the policy, because of what AI told them.
Read that third one again. Your trained, approved, on-brand representative now gets a coin flip against a chatbot nobody at your company briefed.
The surface story, and the higher one
The surface story here is a customer service story. Faster tickets, smarter routing, better bots. That's how most of the industry will cover it, and it's the story the CX software vendors want told.
The higher story is a communications story. Your brand now has spokespeople you never hired, never trained and can't fire. They speak on your behalf every day, to your customers, at the exact moment a customer is deciding whether to trust you. And they make commitments your team never approved.
I've spent much of my career advising and message-training the people who speak for brands, and sometimes serving as one myself, including years on air at WNBC. In every one of those roles, the first rule was the same: know exactly what you're going to say, and make sure everyone saying it says the same thing.
That rule didn't go away. The number of voices just exploded, and most of them aren't human.
I'd give this a name: agentic reputation management. It isn't a productivity problem. It's an unmanaged spokesperson problem. And right now, almost nobody owns it. Comms thinks it's a CX issue. CX thinks it's an IT issue. Legal hasn't heard about it yet.
You can't brief the agent. You can brief its sources.
Here's the part that gives me hope. These agents don't invent your policies out of thin air, at least not usually. They assemble an answer from what they can find: your site, your help center, and the third-party pages they trust. When those sources are clear and agree with each other, the agent has less room to improvise. When they're vague, outdated or contradictory, it fills the gap. Confidently.
So the old spokesperson rule still applies. You just apply it to everything an AI reads.
- Find out what AI is already saying. Ask ChatGPT, Gemini, Google's AI answers and others the questions your customers ask: Can I return this? Is it covered? How do I cancel? Write down where the answers drift from your actual policy.
- Make your own site the clearest voice in the room. One current, plainly written page per policy. No conflicting versions buried in old FAQs or PDFs. Structured so a machine can parse it, not just a person. This is where SEO and GEO, generative engine optimization, stop being acronyms and start being reputation management.
- Fix the echo, not just the source. AI leans heavily on third-party citations: review sites, publications, comparison pages, forums. Strong backlinks tell AI which of those sources to trust. But what matters is what they say. If a widely cited article still has last year's refund policy, that's your new spokesperson.
- Decide who owns it. Put comms, CX and legal in the same room. I've done this kind of cross-team build before, at Pumpkin, with teams that already had full plates. It works when everyone sees the same picture.
None of this guarantees an agent will never get it wrong. But it's the difference between your brand having one story out there or five.
That's the thinking behind The Higher Story AI Visibility Audit, my proprietary platform. It shows what AI tools, including ChatGPT, Gemini, Google and others, say about your brand, which sources they're pulling from, and a ranked list of what to fix first. Accuracy is part of visibility. Being named by AI doesn't help much if it's saying the wrong thing.
So, what's the higher story?
For years, we worried about what our spokespeople said in the interview. Now the interview is happening thousands of times a day, without us in the room. The brands that come out ahead won't be the ones with the slickest bots. They'll be the ones whose truth is so clear and consistent that every voice repeating it, human or AI, gets it right.
So here's my question for you: has an AI ever told you something about a company that turned out to be wrong? Or, if you're on the brand side, have you checked lately what it's telling your customers?
The Higher Story is a series about looking past the surface to the truth underneath: how that shapes the way we see the world, meet our challenges, and, in marketing and communications, do work that lands with more honesty and impact.
Source: Capacity, The Agentic Consumer Report, via PR Newswire, Oct. 27, 2026.