What’s the Difference Between an Agency That Uses AI Tools and One That Builds AI Visibility?

Almost every marketing agency uses AI now. We certainly do. We use it for research, ideas, data analysis, writing, design, development and plenty of repetitive work nobody is particularly heartbroken to see automated. Asking an agency whether it uses AI in 2026 is a little like asking whether it uses email.

But “we use AI” has become one of those wonderfully flexible marketing claims that can mean almost anything. I’ve been in digital marketing long enough to have watched this movie before. A new technology arrives, some agencies genuinely develop expertise around it, and others update the website first and figure out the expertise later.

An agency can be terrific at using AI to make its own work faster and still know very little about helping a client become visible when people use ChatGPT, Gemini, Perplexity, Claude, Copilot or Google AI Overviews to research a company, product or service. Those are different capabilities. One is about how the agency works. The other is about whether AI systems can find enough credible information to understand your business, cite it and potentially recommend it.

Using AI Doesn’t Make You Good at AI Search

There are plenty of good reasons for agencies to use AI. Writers can organize research faster, designers can explore concepts, analysts can work through data and developers can troubleshoot code. I have absolutely no desire to return to the days when every tedious task required a human being staring at a screen for three hours.

But producing 20 articles in the time it once took to produce five doesn’t give those articles authority, and automating social media doesn’t suddenly make a company more credible. AI can make the factory faster, but that doesn’t mean anybody wants what the factory is making.

An agency working on AI visibility should be asking something different. When somebody asks Gemini or Perplexity about your industry, what does it know about you? When Google generates an AI Overview, are you anywhere in the sources? Which companies appear when someone asks an AI assistant for recommendations, and where is it getting the information it uses to make those choices?

That’s a considerably more interesting problem than figuring out how to produce another 50 blog posts.

There Isn’t One AI Internet

One reason I’m skeptical of anyone selling a universal AI optimization checklist is that the platforms themselves don’t appear to agree on what sources matter most.

Meltwater’s August 2026 AI Search Visibility Report analyzed roughly 7.3 million citations across eight AI platforms and found very different source preferences. YouTube led among Google’s AI experiences, LinkedIn dominated Perplexity’s citations, NIH led among ChatGPT’s top sources and Wikipedia led Grok.

I wouldn’t look at that research and immediately tell every client to start making YouTube videos or posting constantly on LinkedIn. Marketing has enough people turning correlations into strategies already. What it tells me is that AI visibility is an ecosystem problem. Your website matters, but so can industry publications, video, professional networks, reviews, directories and other places where credible information about your company exists.

I’ve been around digital marketing long enough to be suspicious whenever someone claims they’ve found the one lever that controls an entire new channel. AI search hasn’t changed my mind about that.

Your Website Still Matters. A Lot.

I’ve also become suspicious of the increasingly popular claim that your website doesn’t matter much anymore because AI systems get their answers from third-party sources.

WARC reported on a 2026 Profound analysis of 11.8 billion data points that found brand-owned websites accounted for 57% of AI citations globally, although that varied considerably by platform and industry.

I like research that makes the answer less convenient because that’s usually closer to reality. Your website matters enormously, and so does the rest of the internet. The goal shouldn’t be choosing between owned content and outside authority. They should reinforce each other so an AI system isn’t relying entirely on what you say about yourself or trying to piece together your business from whatever it happens to find elsewhere.

For a Kansas City business, that also means local authority isn’t created by squeezing “Kansas City” into every available paragraph. There should be actual evidence connecting the company to the market through customers, projects, organizations, coverage, listings and other credible references.

Apparently the Boring Stuff Matters Again

One of my favorite little ironies of AI search is that some of the least glamorous parts of digital marketing suddenly look pretty important.

Yext analyzed 155.5 million location-grounded AI citations across 1,623 brands and found 80% pointed to sources brands could influence, particularly their websites and business listings. When brand-owned domains were removed, directories and review platforms became surprisingly prominent.

After years of marketers chasing every new shiny object, we’re talking about accurate business information and directories again. Apparently MapQuest lived long enough to become relevant to AI. I respect the commitment.

More importantly, this is why I don’t think AI Authority can simply mean publishing more content. Sometimes the work is considerably less exciting: making sure Gemini, Perplexity, Copilot and other discovery systems can establish basic facts about the business and find enough consistent evidence elsewhere to have some confidence in them.

Show Me What You’re Measuring

If I were evaluating an agency claiming AI expertise, I’d spend less time asking what tools it uses and more time looking at what it measures.

Telling me AI helped produce 12 articles, 30 social posts and four landing pages is an efficiency report. An AI visibility report should tell me where the company appears, which competitors appear instead, what sources are influencing those answers, whether visibility is changing and, eventually, whether any of this produces business.

Adobe’s August 2026 AI Traffic Trends Report found AI-referred traffic outperforming other channels on revenue and conversion within Adobe’s data. That’s where this gets interesting to me. A citation in an AI Overview or a recommendation from an AI assistant is nice, but eventually somebody should probably ask whether all this new visibility helped sell anything. Otherwise, we’ve just invented another dashboard.

I’m Still Suspicious of the Secret Formula

AEO, GEO and AI Authority overlap with SEO, PR, content, reputation management, local search and brand building. That’s not a weakness. Marketing loves announcing that the old thing is dead whenever a new thing arrives, but most of the time the old disciplines simply get another job.

What I don’t believe is that somebody has discovered a secret door into AI recommendations. The research is almost reassuringly messy. Gemini doesn’t behave exactly like Perplexity. Google’s AI experiences don’t necessarily use sources the same way as standalone AI assistants. Industries behave differently, local searches introduce another collection of signals, and all of this continues to change.

I’d trust an agency saying, “We don’t know why your competitor is appearing there yet, but here’s how we’re going to figure it out,” considerably more than one promising to make you the preferred recommendation across every AI platform. One sounds like somebody doing the work. The other sounds like somebody found a control panel the rest of us haven’t been invited to use.

Using AI and understanding AI visibility aren’t the same thing. One helps an agency work faster. The other requires figuring out how different AI systems understand your business, where they’re getting that understanding and what you can realistically do to improve it.

That’s the expertise I’d be paying for.