Agentic AI and CTV advertising
How will the advent of agentic AI impact CTV advertising and streaming adtech in ways that previously predominant AI technologies (AI/ML, generative AI) haven’t left their mark on OTT monetisation technologies, workflows, management, and strategy? IAB Tech Lab COO and EVP of Product Shailley Singh offers a concise explanation of the current state of play, and what IAB Tech Lab is developing with its agent-to-agent communication framework, while FreeWheel VP of Product Architecture Jeff Ellin offers a candid take on where buyers, sellers, and publishers are with agentic and LLM implementation today in this discussion with Streaming Media Contributing Editor Nadine Krefetz at Streaming Media Connect 2026.
Bypassing labor-intensive human work
Krefetz asks Singh to provide a quick summary of where he would use an AI agent.
Singh says, “Agentic AI, or the AI technology that we are seeing today with LLMs and generative AI, is different than the previous software technologies.” Previously, AI was supposed to store and move information at the instruction of a human. Now, AI can “analyze and reason over complex systems and multiple data points at the same time; it can almost act like a human and make decisions and actually do the work.”
He provides the example of a seller agent bypassing the human requirement of setting up line items and dropdowns. “Once [a media plan is] decided, [AI]’s already integrated. It can actually go and set up and create all those line items for you. So those kind of things where it’s the labor-intensive human work, which is well-defined and structured, is I think the first low-hanging fruit where AI is being applied and can take over.”
Creative management and deals management
Singh references another panelist’s mention of creative management. “We all know creative management is a very laborious process: going back and forth between the client, the agency, and the seller and setting it up,” he notes. “In my first job in advertising, I was looking at this problem of creative management and unfortunately could never solve it. I think there were an average of 20-plus emails exchanged to finalise one creative that was ready to be used. And maybe [agentic] AI is something that can actually solve that.”
Singh says he’s also hearing feedback that agentic AI can help with deals management: setup, tracking, etc.
Standard protocols for agentic AI
“Now you guys are working though on agent-to-agent communication,” Krefetz says. “Do you want to give us a little bit of background on that?”
Singh replies, “So ultimately the aspirational vision and the goal is that there’ll be autonomous agents that would send the media brief, and on the other side, the seller agent would analyze that and provide the media plan or the media kits or the options that they have available. And then the agent would select and be able to [make a] decision on what media to buy, where to buy, and how to buy, and at what price.”
He discusses the variety of standard protocols needed to have agents speaking to each other, including MCP from IETF, “initially an Anthropic-promoted protocol that allowed you to expose any system that you have for an agentic communication.” He adds, “Google had the A2A where two agents can talk to each other and understand each other’s capabilities. So in our framework, we are kind of using both.”
Decision-making, education, and restrictions impeding progress
Krefetz turns to Ellin: “Jeff, you said when we talked earlier that some companies are at a different point in terms of their preparation. They’re willing to start this, but what are some of the things you hear about from people, either working with agent to agent or other areas, in terms of the publisher comments on the workflow?”
Ellin notes, “I think some of the ad-op stuff, publishers are pretty quick to utilize. When it comes to more of the agent to agent, how am I selling my inventory?” Internal decisions about what inventory to expose are involved. “Should I expose the rate? Who should I expose the rate to? It’s a big change in how they do business,” he says.
“But also we work with both buyers and sellers, and one of the things that we found is that buyers are much more progressive, meaning they’re willing to try stuff out a lot faster,” Ellin explains. “They might not be beholden to this large corporation and a lot of things having to happen, or a lot of sign-offs needing to happen, before they’re willing to really lean in. So some of the things that we’ve seen that might be a little slower on the publisher side also is just like, what LLM are you using?” Some publishers might not have “a corporate policy that has approved a very specific LLM to put their data in for every use case. So some of the things they want to go and do or that are available to them, they’re restricted right now,” he continues. It causes “more of a hesitation.” So it’s a matter of both education and being able to use it, Ellin says.
Standards creating interoperability
Singh agrees, “I think that last point is really important about the LLMs. People should understand that each LLM can have its own decisioning, has a different training, and within LLMs, each model has a different training.” That makes proper LLM selection important. “And that’s where standards come in, that they create the interoperability between agents,” Singh says. “So if you have a standard schema and resource schemas, which every agent requires, then you reduce that hallucination, you enhance the understanding between the two agents of the context in which each agent is talking or sending messages, you make that much more robust.”
Join us 11–13 August 2026 for more thought leadership, actionable insights, and lively debate at Streaming Media Connect 2026! Registration is open!
Related Articles
Streaming Media Connect May 2026 was all about delivering and monetizing sports streaming featured exclusive keynote fireside chats with Matt Graham of the Tennis Channel and Maria Malinowitsch of Verimatrix and a slate of expert panels packed with speakers from DAZN, Globo, Fubo, OneFootball, FloSports, NESN, IAB Tech Lab, Telestream, Zixi, BuyDRM, ViewLift, CacheFly, Servers.com, Tiledmedia, Norsk, and more. Check out a playlist with Streaming Media Connect sessions on Streaming Media's YouTube channel to catch the sessions you missed and revel in the ones you want to relive through the magic of VOD.
19 May 2026