If there is one thing we have learned at The Media League (TML) over the past year, it is that the industry is currently vibrating between two equally dangerous extremes: a blind, frantic rush toward any tool labeled "AI," and a stubborn, legacy-driven refusal to see the paradigm shift staring them in the face.
This week in London, we had the privilege of stepping onto the stage at the Connected TV World Summit (CTVWS). We want to start by extending a sincere thank you to the organizers for providing a platform that remains a cornerstone for high-level M&E discourse. However, we have to be honest: when we looked at the schedule and realized we had exactly ten minutes to distill the entirety of our AI² Research corpus into a coherent narrative, we briefly questioned our own sanity.
How do you take a year’s worth of deep-tissue analysis—covering 200 use cases , 90 AI techniques, and nearly 1,000 companies —and turn it into a 600-second pitch that adds value without inducing a collective migraine for the audience?
From the Engine Room: Synthesizing the Chaos
Fernando: From my perspective as the lead on the research side, the challenge wasn't just the time limit; it was the integrity of the data. My team and I have spent months building a data corpus that now serves as the foundation for our AI² Enquire Agent. We aren't interested in "AI hype." We are interested in AI ROI.
The synthesis process for this presentation was a brutal exercise in prioritization. We had to move past the "Awareness & Experimentation" phase that characterized 2024 and 2025. The reality we are reporting now is far more consequential: 2026 marks the beginning of First At-Scale Deployments.
We spent hours debating which slides would serve as the most effective "intelligence report" for an audience of executives who are tired of hearing about ChatGPT and want to know how to solve Inference-Led Margin Compression. We settled on a "Top Numbers" approach:
-
201 Use Cases across 7 categories.
-
91 Techniques across 6 categories.
-
948 Companies analyzed, with 313 being AI-native.
When I handed the final deck to Thierry, I knew the data was bulletproof. But I also knew the "10-minute freak-out" was real. We were asking him to explain the Cost Curve Paradox and the Sovereign IP "Closed Loop" while the timer was counting down in his peripheral vision.
From the Lectern: The High-Stakes Delivery
Thierry: Stepping onto that stage, my goal was simple: be the market analyst the industry needs, not the one it wants. It is a delicate balance. On one hand, you have the "AI-naïve"—those overindexing on generative tools while disregarding the brutal realities of business continuity and existing margins. On the other, you have the "AI-resistant," who are sleepwalking into a Commoditization of Craft.
In the world of truthful consulting, our job is to point out the negligence on both sides. If you are ignoring the 30–70% reduction in localization costs made possible by ML-driven ASR and NMT, you are failing your shareholders. Conversely, if you think an LLM is a replacement for a robust content strategy, you are failing your brand.
The ten minutes went by in a blur of data points and, surprisingly, a sea of smartphones. It is a strange feeling to see the most senior leaders in the room treat your slides like a tourist attraction, but it confirms the hunger for actionable intelligence.
I was particularly struck by the engagement during our breakdown of the Top 5 Use Cases Driving Cost Savings:
-
Subtitling & Translation: Reducing costs by 30–70%.
-
Content Tagging: Slashing manual logging labor by 40–80%.
-
Clip Extraction: Automating highlights to reduce editing labor by 50–90%.
-
Streaming Optimization: Shaving 30% off bandwidth costs via predictive optimization.
-
Ad Inventory Optimization: Improving fill rates and forecasting through hybrid ML/GenAI models.
The feedback immediately following the session was a testament to the fact that the industry is ready for this level of depth.
The Hallway Effect: A Thank You
The best conversations at any conference happen after the session ends and the timer stops mattering. We want to specifically thank the colleagues who found us in the hallways and gave us the kind of dialogue that does not fit into scheduled programming.
Martin Prins and Chris van der Linden brought sharp, grounded perspectives on where the tech stack is actually heading — not where vendors would like it to go. Ben Kenn and David Watkins challenged our market framing in exactly the ways we needed. Francois Godard offered the kind of rigorous analytical pushback that makes the research better. Alejandro Casal Gomez, Jerome Vial, and Paul Eriksson got into the practical realities of at-scale deployment with the directness of people who have actually tried to implement this. Tom Monrod and Maria Aguate kept the conversation anchored to the human dimension of the agentic shift — a perspective that is easy to lose when the data is this compelling. And Thomas Decieux, Juliet Greco, and many others shared their own front-line stories, which, frankly, are often more instructive than anything we can model.
These are the conversations that make CTVWS worth the journey.
The "Truthful Consulting" Paradox
As we reflect on the week, we find ourselves back at that central tension. We are naturally excited about the Transformational potential of AI. We see a future where Biz Agentic workflows solve the "Chasm" issues of tech complexity.
However, as consultants, we have to remain the "adults in the room." We have to warn against the Reliability Gap and the Rights & IP Complexity that can sink a project before it even launches. We have to be the ones to say that while AI can make TV more engaging and profitable , it can also lead to Inference-Led Margin Compression if you don't own your infrastructure.
To those who are over-indexed on the hype: you are risking your business continuity. To those who are under-indexed: you are risking your survival. In the 2026 landscape, the gap between being "AI-aware" and "AI-integrated" is no longer a strategic choice; it is a market mandate.
Final Thoughts: The Path to NAB '26
The AI² Research program was never intended to be a static report. It is a living, breathing intelligence service. The slides that were so heavily photographed in London are just the surface.
As we move toward NAB '26, where we will formally unveil the AI² Enquire Agent, we invite you to move beyond the flashbulbs. The real work happens in the data, the synthesis, and the courageous decision to overhaul legacy inertia in favor of an AI-integrated future.
Thank you again to the CTVWS team and to everyone who joined us in that packed room. We look forward to seeing you at the next front line.
Fernando & Thierry
The Media League