AI Agents Automate Premium Ad Deals, Enabling Hyper-Contextual Targeting

The future of ad buying is here, and it’s not what you might expect. In this conversation with Ryan McConville of NBCUniversal, we uncover the surprising implications of deploying AI agents not just for planning or reporting, but for the actual transaction of selling premium ad inventory against a live NFL game. This isn't about replacing humans, but about unlocking "premium automation" that can handle complex, bespoke deals traditional programmatic can't touch. The hidden consequence? A potential acceleration of personalized advertising to an unprecedented degree, where ads don't just match a person, but a precise moment within content. This analysis is crucial for anyone in media buying or selling who wants to understand how to leverage these new tools for efficiency and a competitive edge, rather than be disrupted by them.

The "Premium Automation" Frontier: Beyond Programmatic's Reach

The industry has long spoken of automation in advertising, often equating it with programmatic buying. However, Ryan McConville, Chief Product Officer and EVP of Ad Products and Solutions at NBCUniversal, highlights a critical gap: much of television's most valuable inventory, particularly linear, remains undigitized and outside the open real-time bidding (RTB) ecosystem. This is where agentic AI presents a profound opportunity, not to replace programmatic, but to offer an alternative and complementary path. McConville describes this as "premium automation," capable of handling direct IO and programmatic guaranteed deals by automating the negotiation and setup of deal IDs, even orchestrating the entire process without necessarily leveraging a DSP or SSP.

This approach is particularly potent for linear TV, where workflows have historically been manual and time-consuming. By granting seller agents access to linear APIs--data on available units, forecasts, product catalogs, and rate cards--NBCUniversal demonstrated the potential to automate processes that have been unchanged for decades. Furthermore, by integrating with their own forecasting service and identity graph, agents can immediately provide household reach forecasts based on varying media plan splits between linear and digital. This capability, McConville notes, transforms complex planning into an immediate, interactive dialogue.

"The industry's talked a lot about automation for a long time and that has roughly equivalent has been roughly the equivalent of programmatic when people say like automated buying. But when you're a television company not all of your supply is available programmatically because a lot of it's not digitized still."

-- Ryan McConville

The underlying mechanism is elegantly framed by the concept that "AI is the new UI." Instead of building entirely new systems, agentic AI leverages existing APIs of established services--identity graphs, forecasting models, product catalogs--to provide answers through a natural language interface. This means an AI agent can query these services just as a human would interact with a traditional web dashboard, but with significantly greater speed and efficiency. The analogy of an MCP (Model Context Protocol) server acting as the AI's version of a website is particularly illuminating; it repackages data for AI comprehension, not human visual consumption.

This shift underscores a fundamental truth: the efficacy of AI agents hinges on the quality of the underlying data. McConville emphasizes that "garbage in, garbage out" is a critical concern. The real work, he argues, lies in ensuring data cleanliness and adherence to standards, enabling agents to fetch accurate information across diverse publishers and platforms without "hallucinating." This focus on foundational data integrity is the necessary precursor to realizing the benefits of agentic workflows.

The Human Element in an Automated Future

Despite the power of AI agents, the conversation consistently circles back to the indispensable role of humans. McConville stresses that these are "productivity tools" designed to accelerate access to information and facilitate faster querying, not to eliminate human involvement. Buyers and sellers remain in the loop, understanding and approving transactions. The introduction of buyer agents, such as the one RPA and Newton Research employed, mirrors the seller-side approach. A human at RPA can simply ask their buyer agent for available inventory, bypassing traditional emails or phone calls.

"I mean, listen, I think a the the humans are very involved and and I think will be very involved for a long time. These are not like these are productivity tools that speed up access to information and allow you to ask questions of information and get things in a much faster way but like you are still very much going to have humans in the loop like understanding what they're buying and approving what they're buying and like you know along along every kind of step of this process."

-- Ryan McConville

This accelerated access to information frees up human capital for strategic thinking. By automating operational logistics, both buyers and sellers gain more time to focus on campaign strategy and higher-level objectives. The analogy to travel booking is apt: just as Kayak automated the search process, AI agents automate the transactional elements of media buying, allowing humans to focus on the strategic decisions. Crucially, just as a traveler might still verify a booking directly with an airline for a significant trip, media buyers will continue to oversee and approve deals, even when facilitated by agents. Accountability for campaign performance, McConville asserts, remains with the human buyers and sellers, mirroring the established dynamic in programmatic advertising.

The Targeting Trilogy: Hyper-Contextualization and the End User

Beyond workflow automation, agentic AI is poised to revolutionize the end-user experience through hyper-contextual advertising. McConville introduces the "targeting trilogy": the right ad, to the right person, at the right time. While advancements have been made in ad formats and reaching the right person through identity graphs, agentic AI unlocks the "exact right time" by enabling unprecedented contextual relevance.

NBCUniversal's upcoming products for VOD and live content exemplify this. Advertisers can specify highly granular contexts, such as "family scenes around campfires with positive interactions," and AI can scan vast libraries of assets to identify the most relevant moments and suggest optimal ad pod placements. This means users will see ads that are hyper-contextualized to the content they are consuming, potentially even referencing it. For live content, AI can scan feeds in real-time, identifying moments like a fumble, and triggering custom creatives like "Don't fumble this opportunity."

"The more precise that you can get it and because the agentic ai workflow can handle things so much quicker than humans but even then like existing ad technologies."

-- Ryan McConville

Initial testing shows that these hyper-contextualized ads lead to increased ad recall and website visits, validating the effectiveness of this more precise approach. This evolution from programmatic's relevance to agentic AI's hyper-contextualization promises a future where ads are not just targeted, but intimately aligned with the viewer's immediate experience, creating a more impactful and potentially less intrusive advertising ecosystem.

Key Action Items

• Immediate Action: Invest in data hygiene and standardization across your product catalogs and inventory data. This is foundational for any AI agent's effectiveness.
• Immediate Action: Explore and pilot agentic AI tools for internal operational efficiencies, such as automating routine inquiries or data retrieval for planning.
• Immediate Action: Begin conversations with partners (buy-side and sell-side) about how AI agents can facilitate transactions, moving beyond just planning and reporting.
• Next 3-6 Months: Develop internal expertise on AI agent capabilities and limitations, focusing on how they can augment, not replace, human strategic roles.
• Next 6-12 Months: Pilot agentic AI for specific, high-value inventory types (e.g., direct IO deals, sponsorships) where traditional programmatic falls short.
• 12-18 Months: Evaluate the potential for AI agents to enable hyper-contextual ad placements, moving beyond broad audience targeting to moment-specific relevance.
• Long-Term Investment: Foster industry-wide standards and trusted registries for AI agents to build confidence and ensure secure, verifiable interactions.