The Infrastructure Stack Powering Monetization and Distribution in AI

AI is changing the economics of software.

As AI makes it easier to build products, the number of apps competing for attention continues to grow. This creates two fundamental challenges for AI-native businesses: monetization and distribution.

Monetization is becoming more difficult as software becomes easier to build and competition continues to increase. Distribution is becoming increasingly competitive as more products compete for the same users. At the same time, every interaction carries real inference costs, increasing pressure to build sustainable business models.

Advertising has the potential to address both challenges. It can generate revenue while connecting advertisers with users at the moment intent becomes actionable. But AI-native apps cannot simply adopt the advertising infrastructure built for web and mobile.

The systems powering today's advertising ecosystem were designed around pages, placements, audiences, and predictable user journeys. AI-native apps operate differently.

To unlock monetization and distribution in AI-native environments, a new infrastructure stack is required — one built in three layers.

Layer 1: Understanding intent

The foundation of this stack is intent.

For years, digital advertising relied on approximation. Systems inferred what users might want based on demographics, browsing behavior, content consumption, and historical patterns.

Search improved this model by allowing users to express intent directly through a query.

AI-native apps take it further. Rather than a single moment of expression, intent unfolds across the conversation. Consider a user shopping for a new computer. A search query might be "laptop under $1000." A conversation reveals far more: they're starting a graphic design degree in the fall, they need something that handles professional design software, they prefer a larger screen, and they're weighing whether to buy now or wait for back-to-school deals. Each message adds context. Each response uncovers more of what they're actually trying to accomplish.

This creates one of the most valuable signals on the internet.

But raw interactions cannot simply be piped into advertising systems. They contain sensitive information, require contextual understanding, and must be transformed into structured, privacy-safe signals before they can be used effectively.

This creates the need for an intent layer: a system capable of turning natural language interactions into actionable, privacy-safe intent signals in real-time.

Layer 2: Matching intent with demand

Understanding intent is the foundation. The next step is connecting that intent with the demand most relevant to it.

Historically, advertising was placed around content: banners beside articles, videos before content, interstitials between screens. AI-native apps have no equivalent placement or surface. The interaction itself is the environment.

This requires advertising experiences designed specifically for AI-native interfaces. Rather than relying on predefined placements, opportunities emerge naturally as interactions evolve. Matching also changes. Instead of targeting audiences based primarily on historical behavior, advertisers can be matched to real-time intent and context.

A user researching running shoes mid-conversation is a different signal from a retargeted cookie. The result is not simply a new advertising format. It is a new way of connecting users with relevant products, services, apps, and experiences at the moment they are needed. Crucially, the same intent signals that power monetization also power distribution, making this stack the shared foundation for both.

Layer 3: Creating competition

The final step is ensuring that relevant demand sources compete for the opportunity, so every interaction is priced at its true value.

This principle has shaped every major advertising ecosystem. When inventory is exposed to multiple demand sources, competition increases, pricing becomes more efficient, and monetization improves. The same dynamic applies to AI-native apps.

In conversational environments, the value of a single interaction can vary dramatically depending on the user's intent, location, category, and stage in the decision-making process. A user asking "which credit card should I get?" at the point of decision is worth far more than the same user casually browsing. Maximizing value requires systems that can evaluate competing demand sources and determine which opportunity creates the best outcome for each interaction.

Without competition, intent remains under-monetized. With competition, every interaction becomes a marketplace.

The infrastructure layer for AI

Search created infrastructure for intent. Mobile created infrastructure for apps.

AI requires infrastructure for interactions.

The opportunity ahead isn't just about building better models or more engaging interfaces. It's about building the systems that connect intent with demand, in real time, at scale, organically into the conversation. That infrastructure is what will determine which AI products can grow sustainably, and which remain caught between the cost of serving users and the inability to monetize them.

Velocity is building this infrastructure layer for AI-native apps. See how at velocity.io