Buyers Now Ask AI for a Short List, and Your Site Copy Decides If You Make It

For most of the web's life, a page had one reader to satisfy: the person scanning it, weighing whether to call or fill out a form. That reader now shares the page with software. Assistants and search engines parse the same copy, decide what a business is, and hand buyers a short list of names, often before the buyer visits a single site. Copy that reads well to a person can still tell those systems almost nothing, and getting surfaced now means writing for both readers at once.

Nick Lookabaugh is the Founder and CEO of 945 Studios, an agentic digital marketing agency in Pittsburgh that works mostly with home-service and construction businesses across the Kiski Valley and greater Pittsburgh. For the past five months he has rebuilt client sites, and his own, around a single idea: make a business legible to the AI systems that increasingly decide who gets found. He calls it entity maxing, and it starts with the one thing every model reads first, the words on the page.

That legibility isn't optional in his read. "If you're not findable with the AI agents, you're going to become invisible," says Lookabaugh, who sees agent-driven search shifting from novelty to default across the industries he serves. The work, then, is making sure a model can read a page and come away with the right facts in the right order.

Writing for two readers

Lookabaugh's method centers on a semantic triple, a plain statement of who he is and what his company does, written once and repeated word for word everywhere he controls. The same sentence sits on his website, his Google Business Profile, his LinkedIn, and his social accounts, so a model checking any of them meets the identical claim and the same identity holds up as reputation across the web rather than fragmenting into slightly different versions. "It's me repeating the phrase, 'Nick Lookabaugh is the founder and CEO of 945 Studios,' and placing that in as many places as we can," says Lookabaugh.

He also orders a page around how a crawler moves through it, front-loading the facts he most wants retrieved. "Putting the important points at the top means whenever the crawlers or whatever AI model is ingesting the information, they'll have the important things at the beginning, so it's easier for them to remember," he explains. He'll accept a trade in polish to get there, letting copy run a little stiff to a human eye when it reads cleanly to a model, on the logic that anything buried low is easy for a model to skip.

When buyers stop calling

The behavior he's building for is new enough that he has no old playbook for it. "We now have to optimize for agent searching, and we didn't have to do that before," says Lookabaugh. He points to features rolling out in Google where a buyer can send a single quote request to several businesses at once, with no call and no conversation, a path that suits exactly the people who never wanted to pick up the phone and puts a premium on being one of the names an agent returns.

Getting the details right is part of the same job, because a wrong fact is expensive to walk back once a model has absorbed it. "As soon as the LLMs pick up something that's incorrect, it's hard to teach them that it's the wrong piece of information," he says. Consistency across every profile is the cheapest insurance he's found against getting recommended by AI with stale or garbled information attached.

Tools the agents can run

Findability is one half of Lookabaugh's shift; the other is rebuilding his own production around agents. Instead of opening Photoshop for an hour to make a flyer, his team built reusable micro-tools that agents operate as in-house software, spending the time once to save it repeatedly. "I told the agent exactly what I wanted and the template to use, and it made five images in 45 seconds," he says, describing a batch of social posts from the day before. The client sites his shop ships now are static rather than CMS-driven, a choice he credits for load times far faster than sites built by AI on heavier stacks.

The skill he'd tell any operator to develop is the one underneath all of it: learning to build the small pieces of automation yourself rather than waiting for a vendor to package them. "I don't like the term AI agent and gravitate toward how Claude labels it. They call it a skill, and it's just a simple piece of code that does something you want," says Lookabaugh, who treats handing work to agents as the highest-leverage move available to a small team right now. His own agency site, rebuilt in four days with those tools after sitting neglected for years, brought in the shop's first inbound lead from ChatGPT.