Time has started selling ad space inside the pages AI agents read
3 August 2026 · 5 min read
In late July 2026, Time started selling advertising to machines. Working with the AI ad-tech platform Mobian, the publisher now places sponsored FAQ blocks inside the markdown versions of its pages: plain-text copies stripped of design and images, built so AI crawlers and retrieval agents can read them cleanly. Ally Bank and the Project Management Institute were among the first buyers. Time and Mobian say no publisher has sold advertising aimed at AI agents before, though AdExchanger notes that Blankspace, an AI company founded in late 2025, already serves ads to AI agents through CDN integrations.
The detail that matters for a B2B company sits one layer down. Digiday reports that Time's sales team is pitching these slots to brands that have already converted their own corporate websites into markdown, on the reasoning that they are already thinking about what an agent reads. A machine-readable version of your site has quietly become a qualification for a media buy most people have never heard of, and the list of companies that qualify is short.
What did Time actually launch?
Time has been converting its web pages from HTML into markdown: text-only versions that drop the layout, the images and the scripts, leaving the words an AI system needs. Inside those files it sells one sponsored slot per page, formatted as an FAQ carrying a brand's information and messaging, and labelled as sponsored content. Digiday reports that Time measures AI bot traffic to each markdown page and charges a premium for the placement.
The economics are blunt. Time's COO, Mark Howard, told Digiday that bot traffic on the site now exceeds human traffic on most days. Under an advertising model that bills on human impressions, every one of those visits costs the publisher bandwidth and returns nothing.
Why would a brand pay to appear in a text file?
Because the text file is what the answer gets built from. When a buyer asks ChatGPT or Perplexity to compare vendors, the model retrieves source pages and composes a reply. A brand buying an agent ad is paying for a place inside the material the model reads while it forms an opinion, on a domain it already treats as credible.
Whether it works is unsettled. There is no standard for how a model should treat a sponsored FAQ against an unpaid one, and that gap is where the doubts sit. Rob Derow, managing director at BCG X, told Digiday that if the large language models come to see the practice as cloaking, those pages could be devalued the way Google punished the equivalent SEO tactic.
What does this say about a B2B company's own website?
The machine-readable copy of a site has become a distribution surface with a price attached to it.
Agents are already reading your site. The question worth asking is what they find when they do. A page whose substance lives in JavaScript, inside an image, or in a PDF behind a form gives an agent almost nothing it can quote.
- Load your key pages as plain text, with styling and scripts removed, and read what is left
- Confirm your service description, pricing approach, location and contact details survive that stripping
- Move anything load-bearing out of graphics, diagrams and gated PDFs and into readable text
Should you write a separate version of your site for the bots?
Serving agents a materially different story from the one humans get is the tactic penalised in every previous round of search, and it is the specific risk flagged around Time's markdown ads. The safer reading is that a machine-readable layer should say what the site already says, only cleaner.
The buyer behind the agent has not stopped checking either. TrustRadius released its 2026 B2B Buying Disconnect report on 15 July 2026, drawn from 1,862 technology buyers and 444 vendors surveyed globally in January 2026. It found 63% of buyers used AI during their research, and of those, 94% fact-check what it tells them at least some of the time. An AI answer gets a company onto the shortlist. A person then goes looking for whether it was true.
What should a B2B team do about this in August?
Most of it needs no budget. The thing Time is monetising is the thing that was already worth doing: publishing facts a machine can retrieve without guessing at them, the same structural discipline AEO rewards. We ran this on SalesHive's own site first. It is the first account inside our own operating system, and that is where our AI-visibility baseline starts.
- Keep one description of what you do and use it identically on your site, LinkedIn and every directory
- Watch whether paid placement inside retrieval sources becomes normal, because AI visibility stops being free if it does
The takeaway
Advertising has arrived in the layer AI agents read, which confirms that layer now carries commercial value. Before anyone buys a slot in someone else's markdown, the cheaper work is making sure your own pages say something an agent can lift and a buyer can verify. If you want that checked on your site and the gaps written up, start the intake with SalesHive.
This is the work the Organic Engine does for you — built, run, and visible in a live dashboard.