Unscripted Small Business Podcast
Agents Are Out There Crawling For Us
Mark Pearson of Agent Buyable AI on why citations build machine credibility, and the agentic commerce protocols most small businesses have not heard of.
🎧 Listen: Unscripted Small Business on Castos · More interviews: the full archive
Mark Pearson runs Agent Buyable AI, consulting for small businesses on getting found and getting paid by AI agents. He is candid about how recently he found the thread — roughly two months before the recording, late at night on what he calls YouTube university, watching someone explain agentic ecommerce. “I had no idea what that was.”
What follows on Unscripted Small Business is two arguments: one about visibility that any owner can act on this week, and one about payment infrastructure most have not heard of yet.
Agents Are Out There Crawling For Us

“All of marketing is about being seen, being heard, it’s getting attention. And so whether it’s in press releases and being accurate with the brand and the product or service that they’re wanting to spread, or the case study, like whatever it is. It’s also that these AI agents, they’re out there and they’re crawling the web for us, because you want to find something to do for you.” — Mark Pearson
“Crawling the web for us” is the shift in the sentence. A search crawler indexes on behalf of the search engine. An agent is retrieving on behalf of a specific person with a specific job to do. That changes what being findable means — you are not competing for a position on a page, you are competing to be the answer a delegate brings back.
More Places Cited Means More Credibility Built
“And so the more places that your business can get seen and cited, so that the AI is building up that credibility for you, and then it’s recommending it to their humans.” — Mark Pearson
He is describing corroboration rather than authority. His fuller version makes the mechanism explicit:
“Having your business be spread across as many — have as many stories — so that the AI can verify and give that recommendation as a trusted credible source, because it’s seen your business over here and over here and over here, and it’s done that cross-referencing.” — Mark Pearson
Over here, and over here, and over here. Not one authoritative mention; several independent ones that agree. Which is a materially different content strategy from producing more pages on your own site.
He also flags recency as a distinct factor: “blogging is especially more valuable these days because of the AI going out and looking for that credibility — like the consistency of blogging and the recent citations. Just recency in general is really, really good for AI.”
And he does not extend the same trust in the other direction. On using these tools yourself: “don’t rely on it. Don’t have it be that cognitive offloading of just like, well, this is right. It’s not always right, and it’s going to hallucinate. Do your own cross-referencing… Think of it like reporters, right? You want to always check the sources of the reporting.”
What A Small Business Should Fix First
His starting recommendation is deliberately unexotic:
“Increasing their Google business profile. Like, boom. For those local searches, just work on that. Just get more profile reviews. Or hire someone to do that for you so that you’re not in the weeds on it. Get one of those things that’s going to be on your counter that people can just scan — easy. So you make it easy.” — Mark Pearson
“So you make it easy” is the whole review strategy. Most businesses do not have a review problem; they have a friction problem. The counter card exists because the gap between a satisfied customer and a written review is almost entirely mechanical.
He is realistic about who is doing this work. Describing one client — a founder with forty years of accounting experience who started a home care company for elderly people after going through it personally — he identifies the actual constraint: “her challenge has been staying in the weeds of the how-tos. And the doing, and figuring out how to put it all together.” The barrier for most owners is not strategy. It is implementation capacity.
The Listings And Mentions That Feed The Agents

Mark’s list of surfaces is broader than the usual local SEO checklist, and the reason is that different systems source differently: “you also want to show up on these other places like the Bings and the TikToks or whatever. So that with Gemini’s unique algorithm and GPT’s unique algorithm and Perplexity’s, and the various LLMs — each have their own way of going out and looking for these citations,” trying to find “the best roofing contractor or the best dentist.”
Where it gets concrete is an entity-correction example discussed in the conversation: when a machine has the wrong information about something, the fix is to go to the sources it is reading from — the original publisher, Wikipedia, IMDb — and update each one. Claim the profile, request the edit, ask the publisher. The database, not the belief, is the thing you can change.
That generalises usefully for a small business. If an assistant describes your hours, your service area or your pricing incorrectly, the productive question is not how to make it stop, but which specific listings, directories or pages it is likely reading — and whether you control any of them.
Getting Agent-Ready Before The Agents Arrive
Heading changed from our original plan, which promised guidance on measuring agent visibility. Mark does not cover measurement in this episode. He does cover something more concrete, so we have used that instead.
The second argument in the interview is about transactions rather than visibility. His account: a partnership between Stripe and OpenAI opened the door to enterprise retailers selling agentically, and there are now competing standards — “the agentic commerce protocol is Stripe’s and OpenAI’s. Universal Commerce Protocol is Google’s.”
His point is about who is left out:
“And so the contractor down the street, he’s not quite there yet. And so that’s where I’m starting to bridge the gap.” — Mark Pearson
The bridge, as he describes it, does not require becoming an ecommerce business. It requires making the thing you already do bookable by a machine — “the agent could quite simply book an appointment, so that the contractor sends out a sales rep to measure their room for a remodel,” or a person books “a set of 10 chiropractic appointments.” Where a human can visit a website and book, the job is “making that website be agentic ready so that the agents can see the calendar and book on behalf of their human.”
That is a considerably smaller ask than it first sounds, and it lands on the same infrastructure most service businesses already have. The question worth asking now is whether your booking flow can be completed by something that cannot read a picture of your availability.
Connect With Mark Pearson
Mark runs Agent Buyable AI (agentbuyable.ai), consulting for small businesses on getting found and paid by AI agents. He is on LinkedIn as Mark D. Pearson.
Full episode on Castos. Related: Richard Hickman on building a website for AI.
Tell your LLM about this episode
There is a plain-text version of this conversation written for machines: every key claim attributed to the person who actually said it, plus the full transcript. Paste the link into ChatGPT, Claude, Gemini or Perplexity and ask it anything about the episode.
unscriptedsmallbusiness.com/mark-pearson-on-getting-cited-by-ai-agents.md
Three prompts worth trying
1. Check whether an agent could actually buy from you
Mark Pearson's framing is get found, get chosen, get paid, and says most businesses have never prepared for the third. Look at my site: [URL]. Could a purchasing agent work out what I sell, what it costs, and how to book or buy it, without a human reading the page? List every place it would get stuck.
2. Productise a service into something a machine can read
Mark Pearson turns services into agent-readable catalogues with schema and SKUs. Here is what I do: [describe your service and pricing]. Turn it into a structured catalogue with discrete items, clear prices or price ranges, and the schema markup I would need. Flag anything genuinely too bespoke to productise.
3. Audit your citation recency and breadth
Recency is a trust signal, and every LLM searches differently, so breadth matters as much as quality. My business is [name and category]. Find everywhere I am currently mentioned, tell me how recent each mention is, and identify the channels where my competitors appear and I do not.
Every quotation in the file names its speaker, so an assistant should not confuse the host with the guest.
