Newsletter
|
Feb 28, 2025
Personal agents will filter ad content and recommendations for users
Copy Link
Copy Link
Digital advertising has become a significant force in the global economy, with revenue reaching $484 billion in 2022 and is projected to grow to $663 billion by 2027. It also represents ~65% of total advertising spending and accounts for 0.6% to 1.1% of US GDP. The average consumer is estimated to see 4,000-10,000 ads per day.
Major tech companies like Alphabet (Google), Meta (Facebook), and Amazon dominate the landscape, contributing significantly to US GDP (0.85%, 0.47%, and 0.17% respectively in 2023). This market share underscores the importance of digital advertising in driving economic growth and shaping consumer behavior.
Cookies are small data files that websites use to track activity and improve browsing experiences. They help remember user preferences, analyze behavior, and enable personalized recommendations and targeted advertising. First-party cookies are accessed only by the website that created them, while third-party cookies enable cross-site tracking.
Cookies and user data are fundamental to modern digital advertising, enabling personalized experiences and targeted campaigns. These cookies allow precise measurement of campaign performance. Companies create detailed profiles, optimize strategies, and deliver relevant ads by tracking user behavior across websites. The importance of user data extends beyond personalization and can also facilitate retargeting and help identify revenue-generating opportunities like discounts on previously viewed products. However, privacy concerns are prompting a shift towards first-party data and ethical practices, driving the need for innovative solutions that balance personalization with privacy.
Platforms like Google use cookies extensively for advertising and analytics, tracking user behavior, and delivering personalized ads across the web. Due to privacy concerns, cookie use is becoming increasingly regulated, requiring companies to balance personalization with user trust and compliance (Secure Privacy). Modern browsers also implement measures to limit tracking, such as blocking third-party cookies by default.
Consumers will soon have AI agents that filter and personalize advertising, providing a customization layer between consumers and advertisers. Consumers have historically cycled between prioritizing privacy and accepting personalized advertising, ultimately preferring relevance in the ads they see. While still low (46%), the proportion of users opting into app tracking has grown every year.
While consumers remain mindful of their data, advertising is here to stay; however, we believe it is on the verge of a transformation. Companies that depend on advertising and consumers seeking relevant recommendations both could benefit from personal AI agents that securely manage user data and engage with advertisers on their behalf.
As intermediaries, AI agents will filter and personalize content based on user preferences, enabling ad networks to bid for their attention. Users will not see or interact with ads in the traditional sense we do today.
This could start as a browser extension similar to an AdBlocker that filters out noise (irrelevant content) based on how the agent understands the user. Ad networks now need to bid to buy a user's attention from the agent and could even negotiate with an agent to offer the best deal to showcase and the agent filters. Similar to an ad network, this would happen in real-time as the user browses the internet or directly asks the agent for information (e.g., I want to book a beach vacation for my family). This could be a transformative method of ad distribution with a more personalized and positive ad experience for the consumer.
We believe that, much like cookies, AI agents will operate independently of hardware providers, social media platforms, or any first-party solutions. The development and commercialization of these agents will likely follow a similar trajectory to Consent Management Platforms. That said, there is a clear opportunity for companies like OpenAI, through its Operator offering, or Google, as an expansion its Vertex AI Agent Builder, to play a significant role in this space.
For “walled gardens” like iOS and Android, it makes sense for Apple and Google to develop their own native agents, given the vast amount of user data they are already entrusted with. A native solution within these ecosystems would ensure seamless integration and, through SSO (Single Sign-On) solutions, could extend across other platforms, including PC browsers and game consoles.
From a business model perspective, agent creators would generate revenue by acting as intermediaries between users and service providers. This model could resemble an Ad Network, where the agent company takes requests from DSPs (Demand-Side Platforms) and earns fees both for exposure and successful conversions. This could also decrease unprofitable spending as agents facilitating these interactions could deny requests based on the information they have and even get recommendations back to better align with their users.
Additionally, users could pay agents for negotiating and securing the best deals. For instance, if an agent is tasked with finding the best price for a product, the user might agree to a success-based fee upon purchase. This could also follow an affiliate-type model where agents are paid a percentage of revenue by the brand for conversions.
As gaming investors, we understand the importance of advertising as a business model and a user acquisition channel. With games moving to free-to-play, advertising has become the core mechanism for monetizing and acquiring mobile users. Agentic advertising allows you to reduce friction and improve cost conversion to install.
Agents will benefit both consumers and developers.
For consumers: an agent would understand your interests, play styles, preferences, and even spending habits. With an agent integrated into your smartphone and PC, and siloing data locally (or secured in the cloud), the agent could recommend games for you. Your agent will always be looking for games and apps in the background and can store that information for a user to recommend when needed.
Imagine an AI agent that has enough information about you to tell that you are about to churn due to limited recent engagement. That agent can either negotiate directly with the publisher for benefits or deals to offer you and observe if these perks drive retention. If not, the agent can recommend similar (or different) products based on how your preferences have changed. This agent would be constantly aggregating reviews, market data, and could even playtest games to determine if ads and promotions are legitimate (a serious issue in mobile game advertisements).
For developers and publishers: the main value comes from the long tail of personal agents and the data that they can aggregate across users. This data, aggregated across tens or hundreds of millions of players, can help developers build or test games. Imagine clear data around what games retain users best in real-time, or trends around new genres or playstyles that will attract users. This would allow developers to make data-driven decisions on how they build games.
For playtesting, there are two main values: (1) general playtests for bug fixes and user experiences, and (2) quality validation to verify if agents are willing to recommend their game to users. The incentive for these agents to test could be wrapped into deals they can offer their users (e.g., 50% off for all shop purchases for the first month).
Takeaway: Consumers will soon have AI agents between them and advertisers. We call this Agentic Advertising. This new entrant has the potential to fundamentally reshape digital advertising. Instead of brands competing for ad space on platforms that push content onto users, advertisers will engage directly with AI agents that act in users' best interests. This shift moves advertising from intrusive bidding wars to a model driven by genuine product relevance and user validation. In this new paradigm, success will not be defined by ad spend but by the quality of the product and its ability to earn an agent's recommendation.