What does contextual AI look like at scale?
At Streaming Media Connect 2026, Tavant brought together a group of media and advertising leaders for an honest, wide-ranging conversation about what AI is actually doing for the industry today. What emerged was a clear pattern: the companies gaining the most from AI are not treating it as a feature layered onto existing systems—they are building core advertising operations around it.
The panel featured Charlie Goodman from Roku, Joe Caporoso from DAZN, Aulden Kaye Yi from Philo, and Chris Grosso from Intersection, with Filiz Bahmanpour from Tavant. Over the course of an hour, the conversation moved through contextual advertising, real-time operations, measurement & attribution, and the road ahead. The discussion that followed was not a polished marketing narrative but a grounded, practical account of an industry that is genuinely figuring things out in motion.
The first thing that reoriented the room was the sheer operational scale at which these platforms run every single day. When Charlie Goodman described what Roku’s creative review pipeline looks like in practice, it reframed every subsequent conversation about AI, because at this volume, AI is not an enhancement. It is the only way the business functions at all.
“We see about six million creatives come in every month. The scale is astronomical.” — Charlie Goodman, Roku Roku’s approach: an “AI by jury” architecture — multiple LLMs evaluating each creative in parallel for brand safety, object detection, and advertiser intent, with humans making the final call on the edge cases that resist easy categorisation. |
As Goodman described it, the goal is not to replace human judgment but to augment it. “AI isn’t giving you the answer,” he noted. “It’s giving you a confidence level.”
What struck the audience equally was how the most effective deployments are not just technical achievements but organisational ones. At DAZN, managing live global sports events across multiple regions simultaneously, AI is woven into production, programming, advertising, and social workflows as a shared intelligence layer rather than a standalone engineering function. Joe Caporoso put the philosophy plainly, and it cut to the heart of why so many AI initiatives underdeliver in practice.
“The tech part of the business isn’t siloed — it’s done hand in hand with programming, advertising, and social teams.” — Joe Caporoso, DAZN |
Context itself also looks very different across platforms — from physical location signals in digital out-of-home networks to moment-level tone analysis in streaming content and real-time audience engagement during live sports.
Across the panel, there was a consistent and important distinction being drawn between data that arrives programmatically and intelligence that is genuinely contextual. Philo receives thousands of ad creatives each month and has built AI review into its workflow — but the standard that Aulden Kaye Yi’s team holds itself goes beyond brand safety. The goal is tonal alignment: ensuring that what runs connects with the emotional register of the moment it appears in, not just the content category surrounding it.
“The future isn’t about placing ads in a specific show. It’s about aligning with tone and sentiment across moments within content.” — Aulden Kaye Yi, Philo |
Chris Grosso brought the out-of-home perspective and with it a pointed observation about where AI’s structural impact on the ad tech ecosystem is most likely to be felt. Across digital out-of-home, where AI is already enabling screen-level creative optimisation and programmatic buying at a granularity that was previously impossible, Chris sees the same dynamic playing out industry-wide: AI simplifying and automating the transactional layers so that the humans in the room can focus on what actually creates value.
“AI is going to automate much of the transactional ad business in the next few years, allowing sales teams to focus on creativity and relationships.” — Chris Grosso, Intersection |
On measurement, attribution, and what comes next, the panel converged on a shared conviction: contextual and identity-based targeting must work together, not in competition, and the feedback loops that connect campaigns to outcomes are what make genuine optimisation possible. As Goodman noted, proving effectiveness is the real challenge — “It’s easy to guess what works. It’s much harder — and much more valuable — to prove it.” Looking ahead, the contextual opportunity extends well beyond the screen itself. As Filiz Bahmanpour noted, the next frontier involves signals that most of the industry has not yet seriously integrated into how campaigns are designed and executed.
“Contextual signals extend beyond content to include environmental cues like light and mood, enabling more sophisticated targeting.” — Filiz Bahmanpour, Tavant |
What the hour at Streaming Media Connect 2026 made clear is that the industry’s most important AI work is not happening in conference keynotes or strategy decks. It is happening in production pipelines, live event infrastructure, creative review workflows, and ad operations systems, every day, at a scale most observers do not fully appreciate.
Contextual advertising, in other words, is no longer just a targeting tactic — it is becoming an operational capability.
The leaders on this panel are not waiting to see what AI becomes. They are already building with it, and the gap between them and everyone else is growing.
[Editor's note: This is a contributed article from Tavant. Streaming Media accepts vendor bylines based solely on their value to our readers.]