The attention economy is not a metaphor. It is a market. Publishers compete for the minutes people spend engaged with content, and advertisers pay for access to those minutes. For twenty-plus years, that market ran primarily on article pages, video streams, and social feeds. The inventory was measured in page views, watch time, and scroll depth. The attention was diffuse, often passive, and monetized at scale.
AI chat is a different kind of attention. Not better in every dimension, not a replacement for everything that came before, but genuinely structurally different in ways that are starting to shift where publishing and advertising are heading.
The Mode Difference
Scroll-based consumption is passive by design. Social feeds and content platforms are built to sustain low-commitment browsing, which serves both the medium-term goal of time on site and the short-term goal of finding the next thing to look at. The attention is real but it is shallow by design, and the ads placed in that stream are banking on frequency and impressions to create recall.
AI chat asks for active participation. You have to formulate a question. The cognitive work of articulating what you want to know is not trivial, and it changes the quality of attention you bring to the response. You are not scanning for something interesting; you are waiting for a specific answer to a specific question you just articulated. Reading mode is different. Retention tends to be higher. The mental context you are in when the response arrives is the context of a person mid-task or mid-decision, not mid-scroll.
This is why we built Velocity around the conversation turn as the ad unit rather than the session or the page. The turn is where the intent is sharpest and the attention is most directed. A session that contains a turn about comparing solar panel installation quotes has one clearly monetizable moment; the adjacent turns about weather patterns and roof slope calculations are context, not equivalent intent.
Where Publishing Volume Is Moving
There is a practical side to this shift that matters for publishers. Traffic from organic search to content pages has been declining for the categories most affected by AI overview features in search results. Publishers in niches like finance, health, and how-to content, which historically relied heavily on informational search traffic, have felt this acutely. The user intent that used to flow to a content page is increasingly satisfied inside the search interface itself or inside a standalone AI assistant.
This is not solely a threat to publishers; it is also an inventory shift. The publishers who are capturing that intent-based attention are not the ones producing listicles for search rankings. They are the ones building AI assistant products that their audience actively seeks out for specific, expert guidance. A financial planning assistant built by a personal finance media company. A cooking assistant from a culinary publisher. A legal guidance tool from a law-focused media brand.
These are not hypothetical. They exist and more are coming. The question for those publishers is whether their AI assistant products can be monetized, and how. Display advertising does not translate cleanly into a chat interface. Subscription models work for some audiences and not others. Contextual conversation advertising is one of the few monetization approaches that is native to the format.
What Advertisers Are Learning
For advertisers, AI chat inventory is still early-stage. Budgets allocated to conversational placements are a fraction of what flows into search or social. Part of that is measurement unfamiliarity, part of it is supply-side immaturity, and part of it is the legitimate uncertainty that comes with any format where established benchmarks do not yet exist.
What advertisers who are experimenting with conversational placements are learning is that the intent signal is highly specific. A user asking an AI assistant "what is the best HELOC rate right now" is expressing a financial product intent that is more explicit than almost anything inferrable from behavioral signals. The challenge is not intent quality; it is matching that intent to the right advertiser message at the right moment, which is exactly what Velocity's contextual matching is built to do.
The advertisers most positioned to benefit from this format are those in categories where active decision-making conversations are common: personal finance, home improvement, software purchasing, healthcare products, travel planning. These are the verticals where users turn to AI assistants mid-decision, where a contextually matched ad has the highest chance of being genuinely useful rather than intrusive.
The Trust Dimension
There is a counterargument worth taking seriously. AI chat interactions feel more personal than scrolling through a feed. Users who ask an AI assistant for advice about their financial situation or their health are in a different relationship with the product than users reading a generic news article. Placing advertising in that context carries a higher risk of feeling like a betrayal if done poorly.
We do not dismiss this concern. It is the reason Velocity enforces clear sponsorship labeling on every ad card and gives publishers full control over which categories can appear in their assistant. The sponsored card format was designed specifically around this constraint. We are not in the business of sneaking ads into conversations that feel like personal advice sessions; we are in the business of matching commercially relevant queries to contextually appropriate offers, and making both the match and the sponsorship completely transparent to the user.
The trust concern does mean that the bar for contextual match quality is higher in AI chat than in display advertising. A mildly relevant banner ad on a sidebar is forgettable. A mildly relevant ad card appearing in the middle of a conversation about something the user cares about is jarring. The standard for what counts as a good match has to be higher, which is exactly why we built a scoring system that uses relevance thresholds rather than serving everything that clears a basic category match.
The Longer Arc
The volume question is honest: AI chat is not yet a replacement for search or social in terms of advertising scale. The inventory is real but constrained by the current reach of AI assistants that have been integrated with monetization. That will grow. The question is whether the monetization infrastructure grows alongside the product adoption, or whether publishers who build AI assistants end up in the same position that app publishers were in circa 2010, with a mass of user attention and no clean way to extract value from it.
We built Velocity because we watched the mobile app era play out and saw what happens when a new format matures without a native monetization layer built alongside it. The ad tech that gets retrofitted onto formats after the fact is almost always worse than what gets designed in from the beginning. The window to build this correctly is now, while the format is still growing and the norms are still being set.
Publishers who are building AI assistant products today are in a position to shape what monetization looks like in this format. That influence does not last forever. Get the format right while there is still space to define it.
