Over the past year, the concept of agentic media trading (where AI agents on both the buy-side and sell-side negotiate ad purchases) has rapidly evolved from theoretical discussions in blog posts to real-world use cases. That speedy evolution has continued into 2026, with NBCUniversal announcing it is running tests of cross-platform agentic media transactions in partnership with FreeWheel, Newton Research, and independent agency RPA. The US broadcaster claims these tests will involve several firsts for the industry, including the first use of AI agents to automate live sports inventory on linear TV.
In a video demonstration of the workflow released by Newton Research, a specially trained agent working on behalf of RPA (and its unnamed client) provides a list of key campaign details via Newton’s platform (in the example, asking for live sports across linear and streaming via direct IO in Q1 this year, with the objective of finding high-profile inventory to support a specific campaign). A seller agent deployed by NBCU then assesses this brief and responds with some inventory packages to fit the brief.
RPA then assesses these options, and asks via a prompt for more details on one of these packages, after which the NBCU agent provides a more comprehensive proposal. RPA then orders the campaign, still through Newton’s platform, providing its buying guidelines. With the order confirmed, NBCU then automatically creates the order within its own order management system and outlines next steps for activation.
There’s still plenty of human involvement in the process, both during the order as agency staff review and confirm suggestions provided by the agents, and once the order is booked when it comes to activation. And the companies involved say the tests are a big step towards a whole new form of media buying.
“This step forward will redefine how inventory is bought and sold, and what better place to start than within our live sports inventory,” said Mark Marshall, chairman of global advertising and partnerships at NBCUniversal. “Together, we’re on the leading edge of innovation and this is just the beginning of how technology and premium content will transform the advertising industry.”
From MCP to CTV
Sports is an interesting starting point for NBCU’s tests, as it’s an area where a lot of inventory is still bought and sold through direct deals, particularly on the linear side. It’s high-value, high-cost inventory where advertisers might be looking for quite specific inventory to meet their objectives, hence why they don’t necessarily want to buy through programmatic tools. But there’s still lots of time-consuming manual work involved which teams on both the buy-side and sell-side could do without.
Hence the use of AI agents, which can intelligently discover relevant inventory based on a quick brief and iron out campaign details much quicker than two humans would be able to. Human workers still ultimately choose which inventory to buy and give approval before any final decisions are made, but the agents massively speed up the process.
“This partnership illustrates the potential of agentic AI to hyper-streamline strategic media intelligence and transactions in service of business outcomes,” said Jim Helberg, CEO of RPA. “By reengineering manual processes with operational efficiencies we are enabling greater human focus and expertise on key strategic and marketplace nuances.”
There are still plenty of questions around what an AI-powered future for ad trading looks like, particularly around the role of existing programmatic processes and technologies, as well as the role of human traders. Will it always be the role of a human, for example, to pick which inventory packages to buy and give final approval, if a buyer’s own agent can make those same decisions just as effectively?
But the speed at which tests and real-world use cases are being announced suggests these questions will start to be answered sooner rather than later.
Follow VideoWeek on LinkedIn.
