Maybe YOU’RE the robot?
Is a new age of advertising upon us?
This month, we are plugging KidSmart! To learn how to donate time, money, or resources, please visit https://kidsmartstl.org/.
SOURCES
https://digiday.com/media/time-has-started-serving-ads-to-ai-agents/
https://digiday.com/media/wtf-is-markdown-for-ai-agents/
About Us
https://vercel.com/blog/the-rise-of-the-ai-crawler
https://siteefy.com/how-many-websites-are-there/
https://www.semrush.com/website/time.com/overview/
https://time.com/article/2026/08/05/jeanine-john-taele-arrested-charged-gun-trump-golf-club-california/
The Marketing Gateway is a weekly podcast hosted by Sean in St. Louis (Sean J. Jordan, President of https://www.researchplan.com/) and featuring guests from the St. Louis area and beyond.
Every week, Sean shares insights about the world of marketing and speaks to people who are working in various marketing roles – creative agencies, brand managers, MarCom professionals, PR pros, business owners, academics, entrepreneurs, researchers and more!
The goal of The Marketing Gateway is simple – we want to build a connection between all of our marketing mentors in the Midwest and learn from one another! And the best way to learn is to listen.
And the next best way is to share!
For more episodes:
youtube.com/themarketinggateway
Copyright 2026, The Research & Planning Group, Inc.
Transcript:
Every now and then, I read a news article that absolutely makes my jaw hit the floor. And that happened earlier this week when I came across a piece on the website Digiday titled, “Time Has Started Serving Ads to AI Agents.”
Time, as in the website for the magazine, Time. If you go on their website, you’ll see that they have stripped the experience down to the bare minimum.
There’s a menu bar at the top, a popup for an AI chatbot on the side, and very simple layouts with a photo at the top of the page next to the byline box and then plain text articles underneath.
Sure, there are banner ads at the top and graphic ads along the side, and even occasionally in between paragraphs of the articles, and there’s also an endless scroll of recommended content underneath, but this version of the website feels so stripped down that it’s almost like reading the old “printer friendly” versions of articles you used to be able to make almost any website produce.
It feels like the old days of the World Wide Web, when bandwidth, storage and memory all prevented websites from being too big for browsers and web surfers to be able to handle. It’s refreshing at first to not have a web browser experience that’s so cluttered it’s hard to find the actual content, but then you begin to realize after awhile that it’s also so stark and plain that it’s also not engaging and interesting.
It’s just text, almost like what you’d see ChatGPT or Claude spit out when you prompt them with a question.
And in fact, it’s designed so that the text can be shown to AI agents in markdown format without all the HTML extras so that the bots have an easier time parsing the text. And now, it includes ads specifically targeted towards bots.
And this is all very much by design, because Time’s new philosophy is that the age of human readers is over. AI Agents are its new target audience.
And look, we’ve brought this on ourselves, because the Web has become borderline unusable for people who aren’t savvy enough to use ad blockers, aggregation tools or tightened-down browsers to block all of the junk that shows up on the unfiltered vanilla version of websites today.
As was writing the script for this episode, I visited a few websites I regularly read in my locked-down Firefox browser using plain old Microsoft Edge and it was illuminating just how much I’m not seeing when I browse the Web.
Prior to this year, the people running Time would have been annoyed that I, as a free rider, was consuming their content without being shown all the ads they want to show me, and they’re not wrong. I counted seven rotating ads in an article approximately 500 words long and many more in the endless scrolling situation beneath it.
Had I remained on the website for even just a few minutes, I would have been served dozens of ads during my time there.
And by the way, I do subscribe to a magazine service that includes Time, print ads and all, so I don’t feel too bad.
But Time’s solution is actually to stop worrying about me as a reader. I’m old news. Now they’re serving ads directly to AI agents, because those are the readers of the future.
So let’s talk about this cutting-edge trend and if it’s something we’re going to be seeing more of soon. Because I have a feeling that if this works, the Web’s about to split into two divergent experiences: what the bots see and what the human readers see.
And it may be that way sooner than you think, because it’s already happening!
I’m Sean in St. Louis, and this is the Marketing Gateway.
So let’s begin by talking about the broader topic of AEO, or AI Engine Optimization. This is still a new and very evolving field that’s branched off of SEO, or Search Engine Optimization, and there are other names for it, like GEO, LLM SEO, AIDO and AI SEO, all of which have their own wrinkles.
These more broadly fit into the philosophy known as the FOUND Framework, an idea created by an AI visibility strategist named Christopher Littlestone.
He suggests that for websites, AI systems should be targeted through a five-part strategy that spells the word FOUND –
Foundation, which establishes who you are and what you do.
Optimization, which makes your messaging easier for machines to parse.
Utility, which shows you can solve human problems by answering questions or providing solutions.
Niche Authority, which means making it clear to the AI system what your distinct specialty is.
And Data-Driven improvements, which involve measuring how AI systems are recognizing your content and using it.
Now, mea culpa here, I’m not sure how influential this framework is beyond Littlestone’s own community of followers, and I believe it’s proprietary, at least as much as any acronym framework can be.
But I have seen it referenced in multiple presentations at this point, so it’s making the rounds among the people who are talking about this topic.
I’ve linked an article to it in the show notes if you’re curious.
Where it’s useful for us is in understanding how AI chatbots and agents are viewing websites, at least at this point in 2026.
Let’s start with the web crawlers, which used to be primarily employed by search engines and data scrapers, but which have recently been in heavy use by AI tools attempting to build their own web-based training data sets. Web crawlers are basically programs that take trips around the internet by following links and then take snapshots of some or all of the content they find.
According to a study from Vercel.com, GoogleBot’s activity accounted for about 80% of these in 2024, with 4.5 billion fetches per month. GPTBot, Claude and AppleBot were each in the hundreds of millions, and PerplexityBot was in the 10s of millions.
We can presume all of these numbers have gone up as AI has become more capable of searching the Web in response to prompts and as these tools have become more ingrained in search engines.
So, how much activity is that? Well, according to Siteefy, there are around 1.5 billion websites in the world, with about 217 million, or 15%, actually active.
And, in the moment that it took me to read that statistic, about 400 more were created. The Web is a really big place, folks.
But anyhow, if we just limit the activity to the active sites, AI tools are probably pulling an average of 28 pages per active website per month. Though some, obviously, a lot more than others.
Time.com, for example, is probably getting visited a lot more than, say, Zombo.com, the website where anything is possible on its single page.
Adobe’s Semrush estimates about 12.32M visits a month for Time.com, ranking it around the 1,342nd most-visited website in the US and number 4,207 for the world.
The higher your relevance on the Web, the more AI requests your website is going to get, both from web crawlers and from inquires being made by AI search tools, which are becoming a major source of traffic as well while leading to the phenomenon we call “zero click search” where users are getting information directly from AI chatbots that search the web for them, not the website itself.
AI agents are basically doing the same thing as AI search, but on steroids, gathering information autonomously based on prompt instructions and then compiling it for users.
An AI news agent, for example, might consult several sources regularly and then build a brief, up-to-date summary of the key news of the day for a user based on their interests and affinities, sort of like a personal aggregation service.
And that’s where these markdown pages have become a stopgap solution to help reduce the cost of serving full HTML webpages with all the bells and whistles, because at scale, it costs a lot more to serve rich media like interactive displays and videos than it does to serve text.
AI bots and crawlers and agents also don’t see or pass on digital ads, which are necessary to fund those server costs.
And if the AI bots and crawlers and agents are only looking for text anyway, it doesn’t make sense to serve them more than they need, right?
That’s the whole philosophy of an AI visibility framework like FOUND – figure out what the AI bots are looking for, serve them that content and then benefit from higher visibility and traffic.
Well, in theory, anyway. The truth is that there’s not a lot of evidence that the bots are willing to accept markdown data over HTML.
AI being a black box with the platform owners siphoning off data from conversations behind the scenes means we really have no idea what the bots are doing beyond looking at cause and effect scenarios.
There’s also no evidence that reducing the amount of data you’re serving to AI offsets the problem of losing readers to AI, and that’s more or less what’s driven Time to try the new approach of serving ads directly to the bots.
So, here’s how it works. The markdown files contain sections labeled “sponsored content” that include information about brands in a Q&A or FAQ format.
Remember, from our FOUND framework, that being useful and carving put a specialty is something these bots look for.
The idea is not for the bot to pass that on to the user. The LLMs would simply instruct their bots to ignore sponsored content in markdown files to avoid that. The strategy is to instead get the bots reading the markdown files to incorporate that new information into their broader data sets.
And because it comes from a trusted source, like a top-ranked content publisher, and likely repeated multiple times, the AI system picks up the brand data as information and may be more likely to repeat it to other users over time.
Time’s ad tech platform, Mobian, wants to scale this up to the 15% of brands already trying to increase AI visibility with markdown pages, and we’ll see if it catches on.
For now, the two brands that have signed on to this, Ally Bank and Project Management Institute, are definitely on the frontier of digital advertising here.
I’m not sure how much this will help them in 2026, given that when I’ve run queries in various search-enabled chatbots on both brands, I’m finding references to their own websites and related review sites like Bankrate, not publishers like Time.
But let’s envision what the Web looks like in the year 2030.
If present trends continue, we may well have what is being called a “two-track internet” where content is being served to humans in one format and being served to AI bots in another.
This goes way beyond SEO, where hidden aspects of a webpage, backlinks, authority rankings and behind-the-scenes auctions for search terms influence search engines.
This is effectively two content streams that are meant to target two entirely different audiences.
Humans continue to browse websites that are designed to serve targeted digital ads and endless amounts of content.
Publisher sites create content for humans that’s picked up by social media and aggregators and continue to compete for eyeballs based on traditional web principles like stickiness, rich media, social hooks and so forth.
But AI bots see a different version, one that’s specifically made for them, and perhaps even rewritten by an AI tool to make it easier for the bots to parse the language used.
And embedded in this version of the web is a completely different form of advertising that’s designed to influence these bots so that when they’re consulted about a brand, they’re not quoting reviews or the website of the brand; they’re quoting publishers as if those publishers endorsed the brand or associated it with something positive.
I do see some danger here for brands. Today on Time, there’s a headline that says, “Feds Investigate Armed Man Arrested at Trump Golf Course Before President’s Visit”.
It’s an article about someone who appears to have been planning an assassination attempt. Dark stuff. And when I look at it as a human, there is indeed an ad for Ally Bank there in the rotation of ads I’m seeing.
Now I know, as a human reader, that Ally Bank isn’t in any way connected to assassination attempts on US Presidents.
I can correctly deduce that the ad is just there because branded ads show up in content regardless of its topic.
But can the LLMs distinguish that if, in their training data, there wind up being millions of references to Ally Bank along with this article every time a bot pulls it to summarize it?
I’m not sure. Nobody is. And it would be really awful for Ally Bank to have an incident occur where the AI chatbot starts telling users it’s an ally to would-be assassins with special financing plans for their evil plots.
And lest you think that’s something a chatbot couldn’t or wouldn’t do, remember that they have and continue to do things like this all the time. They’re not capable of thought or reasoning.
These are just really sophisticated word guessing generators.
They’re simply evaluating training data parameters probabilistically and generating output based on what their process determines is the most likely word to place next in sequence.
So the likelihood is that the LLM platform owners are already or soon will be instructing their bots, crawlers and agents to ignore sponsored content just so something like this doesn’t happen.
And given how quickly AI platforms change, what we may see is a digital arms race of evolving techniques eventually engineered by AI systems to place AI-facing ads intended to influence AI bots through zero-day vulnerabilities before those holes are covered or those problems are patched.
We may also see LLMs open up their own ad networks to allow influence to come from the inside rather than the outside, hopefully with some well-established guardrails for brands firmly in place.
Which means that the race to see who can successfully influence AI bots with ad tech is on, and as we hurtle towards a two-track internet, we can expect to see a lot more about this topic as a subset of the broader discussion around AI visibility.
As for me? I’ve been reading a lot more books and printed publications lately because, at least for now, those are being produced with humans in mind.
I’m Sean in St. Louis, and this has been The Marketing Gateway. See ya next time!
AUGUST PLUG
This month’s plug is for KidSmart, a charitable organization based here in St. Louis, MO with the mission of empowering children in need to succeed academically. And as we head into back to school season, KidSmart will be out there handing out $104 million school supplies to around 80,000 kids in the area.
They also operate a free education supply store and have a truck that brings school supplies directly into neighborhoods in need.
As you might imagine, August is their Super Bowl, and they need both financial and volunteer support this month. So go to kidsmartstl.org to learn more.
Leave a Reply