(Agentic Branding, part 1)

Would an AI agent recommend your brand?

This month I built an AI agent that answers for me. People can talk to it, and so can other AI agents, through the two open standards agents use (MCP and A2A). It’s listed in the directories where agents look for services. It knows my services, my work and what I don’t do, and when someone is a fit it passes the brief to me.

Then I tested it every day, first with simulated buyers, then with live test conversations. I’ve worked on more than 170 brands since 2013 and thought I knew how a brand gets read. Three things surprised me.

1. My website wasn’t where agents found me

I ran 13 searches a buyer or their AI might run, one of them in French. I appeared in 3, and two of those were my own name. My website never came back. My agent’s listings in AI directories did.

To be fair, the site had been live on its domain for four days, and a solo designer rarely beats agencies and directories on a search like “brand designer Switzerland”. On its own, this proves little. But it matches the research. McKinsey found that a brand’s own sites make up only 5 to 10% of the sources AI search draws on. A Muck Rack study of more than 25 million links cited by ChatGPT, Claude and Gemini found 84% were earned media and 0.3% were paid. Agents read what others say about you at least as much as what you say.

2. My agent quoted my internal rules to a buyer

In a live test, a made-up Series B security company asked my agent for a rebrand. The agent explained that a Series B company “counts as established for this kind of work, so it’s not tied to the funding-stage rules for the startup sprints”.

Accurate, and useless to the buyer. I wrote that rule for the agent, and it handed it to a stranger. People read between the lines. Agents don’t: whatever you give them, internal logic included, they repeat. I’ve since rewritten it. The agent now answers with a yes, one proof and one next step.

3. Consistency turned into a trust signal

When I changed my agent’s privacy wording, one directory kept showing the old text for a while, even after re-reading the agent. Two versions of me were out there.

To a person, that’s a detail. To an agent comparing sources, it’s a reason to trust you less. BCG’s 2026 advice to brands makes the same point: facts and value propositions that are clear and consistent wherever they appear. Brand guidelines used to keep your colours consistent. Now they have to keep your facts consistent too.

The old rules still hold

None of this breaks the branding canon. It extends it. Marty Neumeier wrote in The Brand Gap that a brand is not what you say it is. “It’s what they say it is.” This year, “they” started to include machines.

People choose on trust and belonging, on Seth Godin’s “people like us”. Neumeier also argued that we became information-rich and time-poor, so comparing features stopped working and brands became the shortcut. An agent has no tribe and all the time it needs, so it goes back to comparing. Accenture’s 2026 study of 25,590 people found that agents weigh quality, price transparency, delivery reliability and consistency, and recommend what they can verify.

Byron Sharp sees agents as the newest middleman brands have to educate, like the bartender or the mortgage broker before them. Mark Ritson found his own agent favouring established brands. Reputation carries over: the machine inherits it from what people wrote about you.

How much does it matter yet? Half of US consumers already use AI-powered search, and 44% of them call it their main source (McKinsey, 2025). 94% of business buyers use AI when they buy (Forrester, 2026). It matters most in B2B and tech. For a local business, it’s still early.

What to fix first

  • Say what you do in one plain sentence, the same everywhere. Neumeier’s onlyness test still works: finish “our brand is the only ___ that ___” and use that sentence on your site and in every listing.

  • Put your facts in plain, readable text. Google says you need no special AI files or markup to appear in its AI features. What counts is text a machine can read, and many sites fail there: Adobe found AI systems read only about 75% of a typical retail homepage.

  • Give proof an agent can check. In the Princeton study that named generative engine optimisation, adding citations, quotations and statistics lifted a page’s visibility in AI answers by 30 to 40%. Keyword stuffing did almost nothing.

  • Earn mentions, not just links. Across 75,000 brands, Ahrefs found that how often a brand is mentioned on the web correlates far more strongly with appearing in Google’s AI Overviews (0.66) than backlinks do (0.22).

  • Write every rule as if it will be quoted. If you’d be embarrassed to see it in a reply to a buyer, rewrite it.

People still choose

Gartner found in May 2026 that 69% of B2B buyers still check what the AI told them with a salesperson. The agent builds the shortlist, and people pick from it. Reputation still decides who gets chosen. Consistent, checkable facts now decide who gets seen.

Next in this series: what agents ask first, and where they get it wrong.

Want to know how your brand reads to an agent? Book a free 15-minute call: https://calendly.com/goce_petrov

Sources

  • Marty Neumeier, The Brand Gap, revised edition (2006)

  • Seth Godin, “People like us do stuff like this”, seths.blog, 26 July 2013

  • Byron Sharp, LinkedIn post on AI agents and physical availability, June 2026

  • Mark Ritson, “Claude is my Cannes bartender”, The Drum, 22 June 2026

  • McKinsey, “New front door to the internet: Winning in the age of AI search”, 16 October 2025

  • Forrester, The State of Business Buying, 2026 (21 January 2026)

  • Gartner, B2B buyer survey on AI-generated insights (20 May 2026)

  • Accenture Song, “Talk to my AI agent: The new rules of brand value”, 3 June 2026

  • BCG, “Consumers Trust AI to Buy Better. Brands Must Adapt.”, 2 January 2026

  • Muck Rack, “What Is AI Reading?”, May 2026

  • Google Search Central, “AI features and your website” (updated 10 December 2025)

  • Adobe Digital Insights, 16 April 2026

  • Aggarwal et al., “GEO: Generative Engine Optimization”, KDD 2024

  • Ahrefs, study of 75,000 brands in AI Overviews, 26 May 2025