\n<\/p>\n
By <\/em>Peo Persson<\/em>, Co-Founder & Chief Commercial Officer, DanAds<\/em><\/p>\n This year at CES, six of us from DanAds were on the ground in Las Vegas \u2013 meeting with broadcasters, streamers, platform partners, and agencies from morning to night. And if I had to sum up the CTV conversation in one sentence, it would be this: agentic buying is coming fast, but the industry is still struggling with the fundamentals.\u00a0<\/strong><\/p>\n There was a lot of excitement on the floor \u2013 some of it warranted, some of it aspirational. What became clear very quickly is that while AI is accelerating planning and execution, fragmentation, signal loss, and measurement challenges haven\u2019t magically disappeared. In fact, in many cases, they\u2019re becoming more visible.\u00a0<\/p>\n That tension \u2013 between promise and practicality \u2013 is exactly where DanAds sits today.<\/p>\n Digital AI? That\u2019s so 2025. The convention floor at the 2026 Consumer Electronics Show (CES) was all about \u201cphysical AI\u201d<\/em> \u2013 in the form of robots. A highlight was LG\u2019s CLOiD, an AI-enhanced humanoid robot housekeeper that demonstrated its (leisurely paced) cooking, fetching and laundry-folding capabilities for event attendees.\u00a0<\/p>\n While suddenly Rosey from The Jetsons doesn\u2019t seem so far off, it\u2019s also hard to imagine CLOiDs appearing (ominously?) in every household window later this year. The same cannot be said for AI agents in digital advertising, which are proliferating so fast that real-time bidding specs are getting AI upgrades (ARTF) and new protocol standards are being put through their paces.\u00a0<\/p>\n Physical AI may be the next frontier, but the digital media and advertising world was deep in discussion about digital AI execution during CES, particularly with CTV. At the same time, issues with fragmentation, audience targeting, and measurement finally came to a head, with industry buyers and sellers increasingly embracing context.<\/p>\n CES featured no shortage of adtech AI announcements, and a high percentage of them focused on CTV advertising:<\/p>\n What stood out to me and to many of the conversations we had at DanAds, wasn\u2019t just how impressive these announcements were. It was how clearly they reflect a shared industry goal: reduce friction between planning, activation, and reporting.\u00a0<\/p>\n But here\u2019s the catch: AI agents are only as effective as the execution and data frameworks beneath them. In CTV, those foundations are still highly fragmented.<\/p>\n Many of these AI announcements were strikingly substantive. For example, the demo video alongside NBCUniversal\u2019s seller agent launch showed exactly how independent agency RPA would build a buy, using a natural language brief to receive information about inventory availability and then optimize based on desired audiences and outcomes. The video even showed how the approved order would be inserted into NBCU\u2019s order management system.\u00a0<\/p>\n Even more impressive, the demo highlighted the ability to buy ads in NFL\u2019s Sunday Night Football \u2013 CES re-confirmed that live sports inventory is the hottest for brand advertisers. The first campaign transacted via NBCU\u2019s seller agent is booked during American football playoff games in Q1 2026.<\/p>\n That\u2019s not to say brands and agencies were rushing home from Las Vegas to dump their entire advertising budgets into buyer agents. While the releases and demos were impressive, Razorfish\u2019s Chief Social and Innovation Officer Cristina Lawrence commented that AI agents \u201caren\u2019t ready for prime time just yet, so we need to think of this as inevitable but incremental.\u201d\u00a0<\/p>\n Fortunately, the adtech AI conversation seems to have moved past \u201cIt will replace all the humans!\u201d<\/em> to \u201c<\/em>OK, this is pretty different, but it\u2019s going to make a lot of stuff easier and faster<\/em>\u201d<\/em> \u2013 most notably in shortening the time and friction between planning and activation. Still, while the path forward may be clearer, progress will still take time \u2013 observers should keep their eyes on how agencies leverage their new high-level orchestrators.<\/p>\n Streamlining buying and selling via agents sounds particularly appealing in the heavily fragmented landscape of CTV, which seems to grow only more convoluted and difficult to manage by the year. CTV streamers and platforms have long cribbed from the walled garden playbook to demand advertisers to use proprietary identifiers \u2013 part of what TVREV\u2019s Alan Wolk labels \u201cFeudal Media\u201d \u2013 which has created challenges in holistic targeting and reporting.<\/p>\n Identity resolution has been a central tool for targeting CTV campaigns across platforms and measuring outcomes, but conversations at CES conceded an overreliance on IP addresses \u2013 which have become increasingly unreliable. Recent research from the Coalition for Innovative Media Measurement (CIMM) discovered that CTV IP-address-to-email-address linkages are only accurate 16% of the time; IP-to-postal linkages sit at 13%. Identity resolution providers only agree 2.8% of the time on IP-to-email linkages and 6.4% on IP-to-postal linkages.<\/p>\n Further exacerbating IP address reliability, connected TV manufacturers are switching from IP4 to IP6 addresses. \u201cThe era of IPV6 has brought dynamic addressing, carrier-grade NAT, router resets, ISP practices, and IPv6 privacy features that rotate parts of the address,\u201d Amol Waishampayan, Co-Founder at fullthrottle.ai, told Mike Shields of Next in Media. \u201cAll of that to mean IPs change too often to function as a durable household identifier.\u201d<\/em><\/p>\n Perhaps recognizing the challenges of targeting audiences in such a fragmented landscape, Havas Media is stressing \u201cContext-First Planning\u201d<\/em> with its new AVA global AI portal for managing AI tools.\u00a0<\/p>\n \u201cWe are big believers in context-first planning, and that\u2019s important because it allows us to identify growth moments that are more important for a given brand or business throughout a planning cycle,\u201d<\/em> said Jackie Lyons, Chief Planning Officer at Havas Media Network, told Beet.TV.<\/p>\n And even if agentic transactions can help buyers navigate a terribly fragmented landscape for planning and activation, unifying measurement is another massive hurdle. Context is also playing an interesting role here as SSP Index Exchange announced a partnership with Gracenote allowing CTV impressions to be reported at the individual show or program level. With Spectrum Reach as a launch partner, advertisers buying through Index Exchange can now get full details about show context without a GraceNote subscription.<\/p>\n CES offered more clarity as to how agentic transactions will operate, particularly in regards to CTV, even if adoption may lag. However, the event also made clear that CTV fragmentation challenges have become more than the industry can handle. AI may be able to solve some of these fragmentation issues from the planning and execution front, but measurement woes are likely to continue to plague CTV.<\/p>\n At DanAds, that\u2019s exactly what we\u2019re focused on. As agentic buying evolves, incrementally, not overnight, solutions like ours will become increasingly important in helping the industry move from experimental automation to scalable, trustworthy CTV performance.<\/p>\n And based on what we saw and heard at CES, that work has never been more relevant.<\/p>\n<\/p><\/div>\nDigital AI is table stakes. Execution is the real story.<\/strong><\/h2>\n
AI everywhere, all the time (especially in CTV)<\/strong><\/h2>\n
\n
Real applications, real questions\u00a0<\/strong><\/h2>\n
Identity crisis in a fragmented world<\/strong><\/h2>\n
Context to the rescue?<\/strong><\/h2>\n
A sharper picture<\/strong><\/h2>\n