Posted

We just got back from Marketecture Live, and the energy was different this time. Not the cautious optimism of a year ago, not the breathless hype of two years ago. This felt like an industry that’s starting to see the real shape of what’s coming, and now grappling with the fact that a lot of existing infrastructure wasn’t built for it.
Across two days of sessions covering everything from CTV supply chains to residential proxy botnets to the future of LLM advertising, a few themes kept surfacing. Here’s what stood out to us.
1. The Supply Chain Trust Problem Isn’t Solved… It’s Getting Worse
If there was a single thread running through the conference, it was this: the programmatic supply chain still has deep integrity problems, and CTV has inherited many of them.
Keith Petri from Viant walked through how intermediary hops between publishers and buy-side platforms create incentives to manipulate user agents, device identifiers, and IP addresses, all done to inflate bid density and eCPMs. Recent research from Sincera suggests that only about 13% of IP addresses can be accurately matched to a household. That’s the foundation most CTV identity resolution is built on.
Meanwhile, the session on Bad Box 2.0 and residential proxy botnets was genuinely alarming. Dr. Krzysztof Franaszek from Adalytics and Rob Leathern from InfoHawk laid out how cheap Android TV boxes sold on Amazon and Walmart come pre-loaded with malware that converts home Wi-Fi routers into proxy relay networks. These networks are being used for ad fraud, bank account takeovers, DDoS attacks, and worse. One stat that stuck with us: a major auction site was seeing 60 - 70% of its traffic coming through residential proxies, with bots sniping bids while programmatic ads ran alongside the activity. And these botnets are also why our concert tickets keep getting snagged by expensive scalpers.
For anyone in the verification and brand safety space, the message was clear: the attack surface is expanding, detection is harder than ever, and the industry needs to move faster.
2. Premium CTV Is Being Bought Direct, and the Pipes Are Being Built to Support That
One of the most grounded panels featured Philip Inghelbrecht from Tatari alongside representatives from Mike Reidy NBCUniversal and Bill Murray from Warner Bros. Discovery. The framing was straightforward: of the $30 billion streaming ad market, roughly half is programmatic and biddable, but often includes low-quality inventory and fraud. The other half - the truly premium stuff - is transacted through direct insertion orders or programmatic guarantees.
Tatari’s Upstream platform, which bypasses the programmatic stack entirely by integrating directly into publisher ad servers, represents a bet that for premium CTV, the SSP layer is unnecessary overhead. Philip noted that 90% of their streaming buys come from the same 10 publishers. When supply is that concentrated, programmatic principles don’t always apply.
From NBCU’s perspective, this is about lowering the barrier for performance-focused advertisers (especially small and medium businesses that might currently be social-only) to access premium CTV inventory without meeting traditional minimum spend thresholds. It’s a growth play as much as an efficiency play. Read more on our thoughts about the path to CTV buying here.
3. LLM Advertising Is Coming, but the Business Model Isn’t Settled.
Terry Kawaja’s “10 Explosive Predictions” session set the tone: four major LLMs have raised nearly $300 billion, big tech is investing $700 billion annually in AI infrastructure, and subscription revenue at around 6 - 7% isn’t going to cover the bill. All roads lead to advertising.
But the form factor is genuinely uncertain. OpenAI launched with a CPM model, which Kawaja argued is a smart starting point to build advertiser density on the platform, but he predicted a rapid evolution toward performance and affiliate models. His logic: AI chat prompts deliver intent signals that make search look crude by comparison. An average prompt is 25 words of rich context, followed by conversational refinement. That’s a targeting goldmine.
Ari Paparo’s opening session framed the broader landscape: content marketplaces where publishers license articles to LLMs, agentic commerce that’s somewhere between science fiction and early reality, WebCP protocols that let websites communicate directly with AI agents, and emerging concepts around agentic advertising where buyers and sellers negotiate through AI. None of this is settled. Far from it - all of it is being actively built.
4. Publishers Are Diversifying or Dying, and Brand Equity Is the Differentiator
Neil Vogel, CEO of People Inc., gave one of the most compelling sessions of the conference. His company coined the term “Google Zero” internally three years ago to describe the inevitable collapse of search referral traffic. Since then, Google referrals to their properties have declined 50%.
And yet: nine straight quarters of revenue growth, approaching a billion dollars annually. The strategy is deceptively simple: ruthless diversification anchored by trusted brands. Forty percent of their business is growing at nearly 40%, driven by social video, events, AI licensing deals, Apple News, email, and apps. The other 60% (traditional web) is holding roughly flat thanks to strong monetization discipline.
Vogel’s sharpest point: brands have the privilege of trying to exist, not the right. The old service-publishing brands that people turn to Google or ChatGPT for (parenting advice, basic how-to content) are hollowing out. The brands with permission to do events, licensing, commerce, and premium content - brands like People, Food & Wine, Travel + Leisure - are thriving because they can meet audiences in many different places in many different ways. That’s a thread we heard a lot, that in the age of instant generative everything, authenticity and human connection makes all the difference.
For AI specifically, there’s an inversion happening. The high-volume research content that was most valuable in the search era is becoming commoditized. The fresh, daily content, such as celebrity news, cultural coverage, is what LLMs most need for training and search. People Inc.’s deals with OpenAI, Meta, and Microsoft reflect this shift.
5. Agentic AI in Campaign Operations Is Real, and It’s Moving Fast
The PMG and Freewheel session was the most tangible proof point for AI’s impact on day-to-day advertising operations. Mike Treon from PMG described picking up agentic coding eight months ago and building tools that collapse multi-step manual processes into automated workflows - extracting delivery data by deal ID, analyzing targeting, identifying missed opportunities - all through natural language prompts connected to Freewheel’s MCP server and REST APIs.
The bigger story is organizational. PMG built an internal framework called Alli that gives operators a path to prototype, stage, and productionalize AI tools with proper credentialing and version control. Their chief product officer’s response to the wave of internal AI development wasn’t to shut it down, it was to build rails around it. The result is that experienced campaign managers can now capture institutional knowledge as repeatable, scalable workflows, and the agency is positioning itself as a technology company with media services rather than the reverse.
Jeff Green echoed this from the Trade Desk’s perspective, calling programmatic advertising the most AI-conducive industry in the world: 20 million impression opportunities per second, millions of campaigns, billions of users, all resolved in under 10 milliseconds. He confirmed a closed beta where users can create Trade Desk campaigns through Claude, and described agentic AI as the natural solution to the combinatorial explosion of campaign optimization decisions that UIs have always struggled to represent.
So What Does This All Mean?
From where we sit at Boltive, the conference reinforced something we’ve been seeing in the market: as AI transforms how advertising is bought, sold, and delivered, the trust and safety layer becomes more critical, not less.
Residential proxy botnets are making fraud harder to detect. Supply chain manipulation is inflating CTV metrics. New surfaces such as LLM chat interfaces, agentic browsers, AI-powered commerce are creating attack vectors that didn’t exist a year ago. And the industry’s push toward direct integrations and supply path optimization, while positive for quality, also highlights just how much cleanup remains in the programmatic pipes that most spend still flows through.
The companies that will matter in this next era are the ones that can provide verification, quality assurance, and brand safety across an increasingly complex and fragmented landscape. That’s exactly the problem we’re building for.

The Governance That Failed Us: Why Digital Compliance Must Shift from Intent to Behavior
Go to Post
Text LinkThe Cost of Complacency: What the LinkedIn Privacy Lawsuit Means for Every Digital Brand
Go to Post
Text LinkAvoiding the Next Headline: What Privacy and Security Teams Need to Know Heading Into 2026
Go to Post
Text LinkThe Big Shifts Shaping the Future of Ad Tech: 3 Key Takeaways from the 2025 NAI Summit
Go to Post
Text LinkAI Adoption, Evolving Enforcement & the Convergence of Privacy & Security: 6 Big Takeaways from the 2025 PSF Spring Academy
Go to Post
Text LinkInside the California Law Association Privacy Summit: What Regulators Are Really Watching
Go to Post
Text LinkBoltive Launches New Product Ad Monitor: Revolutionizing Ad Insights with AI-Powered Discovery Engine
Go to Post
Text LinkIAPP GPS - Our Thoughts, Part 2: What Regulators Want Us To Do – The Value of Proactivity
Go to Post
Text LinkIAPP GPS - Our Thoughts, Part 1: What Regulators Want Us To Know – Busting Common Myths
Go to Post
Text LinkPrioritizing Children's Privacy: Strategies for Ethical Advertising and the Use of AI
Go to Post
Text LinkBusiness Ninjas Podcast - Secure Protection for Advertisers Against Invasive Media
Go to Post
Text LinkAutomating Threat Detection: How Boltive is Harnessing Artificial Intelligence to Reshape Ad Security
Go to Post
Text LinkForbes Article - The Privacy Prescription: Rules Restricting Health Data Use And How To Employ More Holistic Security Measures
Go to Post
Text LinkOur CEO, Dan Frechtling, featured on Leadership Live podcast with Daphna Horowitz
Go to Post
Text LinkWhy Data Privacy is Being Overhauled in 2023: Dan Frechtling featured on the Security Weekly Productions podcast
Go to Post
Text LinkOur Director of Product, Christine Desrosiers, discusses the current state of Ad Tech with Brand Safety Institute
Go to Post
Text LinkGeekwire wrote about the heartening link that brought our CEO and CFO to Boltive.
Go to Post
Text Link