💥 Contextual AI for CTV: Savior, Snake Oil, or Just the Best Bad Idea We’ve Got?

👋 Welcome back to the unregulated theme park of nonsense known as adtech—where acronyms outnumber working products, and everyone is "AI-powered" until you check under the hood and find a tired keyword engine duct-taped to a Google Sheet.

Today’s headliner:

Contextual AI in Connected TV (CTV).

It’s having a moment. The kind of moment where investors are leaning in, CMOs are nodding solemnly in keynotes, and every startup with a pulse is retrofitting their deck like it’s 2019 and blockchain is still cool.

But is this actually the answer to adtech’s midlife crisis? Or just another buzzword dressed in a hoodie pretending to be a revolution?

Let’s pull the curtain.

🔥 ACT I: The Sudden Sex Appeal of Contextual AI

Once upon a time (read: the early 2000s), contextual targeting was the nerdy kid in the cafeteria—reliable, underappreciated, and always available to do your homework. Then behavioral targeting waltzed in like the quarterback with tracking cookies and promised us conversion rates to die for.

Fast forward:

• 🍪 Cookies are crumbling.

• 🏛️ Regulators are foaming at the mouth.

• 🤳 Consumers are side-eyeing every creepy retargeted ad like it’s a stalker ex.

And suddenly, the quiet kid with the contextual chops?
He’s the only one not under GDPR indictment.

Contextual AI is getting dragged back onto the dance floor, now with glow sticks and a machine-learning makeover. The pitch? Smarter. Safer. Scalable. Also… allegedly sexy. But we’ve heard that one before, haven’t we?

📈 ACT II: The Makeover Montage—Contextual Goes to Tech Camp

Let’s talk about what “Contextual AI” is supposed to mean now.

It’s not just keywords and genre matching anymore. This is sentiment analysis meets computer vision meets “trust us, it’s AI.”

✅ It reads the room.
✅ It listens to the tone.
✅ It claims to know that an ad for antidepressants might not belong next to a scene where someone’s dog dies.

We’re now pretending this is all very new—because we added words like “multimodal,” “taxonomy,” and “Plutchik’s Wheel of Emotions.” (Yes, seriously.)

But under the hood, it’s contextual v2.5, now with more feelings.

🎬 ACT III: The Cast of Characters Actually Doing the Work

Let’s give credit where it’s due—some companies aren’t just slapping “AI” on their pitch decks and hoping no one asks questions.

🎭 Anoki

Their tech doesn’t just watch the show, it feels it. A breakup scene? No chalupa ads. Maybe a box of Kleenex or whiskey instead. Anoki's plugged into Google TV and Amagi, which means this isn’t just vaporware—it’s already lurking inside your favorite FAST apps.

📚 Gracenote (Nielsen)

Remember metadata? Gracenote does. They’re playing librarian in the land of chaos, pushing a standard taxonomy for CTV while the rest of the industry plays Mad Libs with genre labels. 70% of FAST channels? Already on board. (Probably because they were tired of getting toothpaste ads in horror marathons.)

🧠 IRIS dot TV

These guys treat video like URLs. Their IRIS_ID system lets DSPs and SSPs know what’s actually on screen—down to the frame. That means no more luxury car ads before “World’s Cheapest Cars” clips. They swear Carl’s Jr. saw triple-digit incremental sales from this tech. Fries with that?

🛒 KERV

If Anoki feels the scene, KERV buys it gifts. They turn every object on screen into a shoppable item: sneakers, coffee mugs, weird cat-shaped lamps. It’s QVC on steroids, and it works—because yes, viewers will absolutely buy a throw pillow in the middle of a murder mystery if it’s clickable.

😬 Wurl

They’ve gone full Freud. Wurl’s BrandDiscovery tool uses emotional intelligence (yes, we’re in therapy now) to match ad tone with content. No more upbeat jingles next to grim true-crime docs. Plutchik’s Wheel of Emotions ensures the ad knows when to sit down, shut up, and stop smiling.

⚖️ ACT IV: The Case For Contextual AI (The Optimist's Pitch)

Let’s be generous. There are real reasons to care about contextual AI, especially in CTV:

🔐 Identity is shattered.
No cookies, shaky IP targeting, and clean rooms that don’t clean up much. Context is one of the few things left that’s universally observable.

📉 Behavioral creepiness is canceled.
Nobody wants their hangover Googling to show up in their ad experience for the next month. Context lets ads stay relevant without dragging your digital skeletons out of the closet.

📺 FAST is booming.
And with boom comes chaos. Contextual AI is promising to tame the wild, overgrown garden of FAST channels and give advertisers actual control.

🧼 Brand safety, finally handled.
No more Applebee’s spots right after breaking war coverage. Contextual AI sees the mismatch coming and slams the brakes.

🧪 ACT V: The Skeptic’s Guide to the Galaxy

Of course, this is adtech. For every promising pitch, there’s a “yeah, but…” waiting in the wings.

👴 Same old trick, new trench coat.
We’ve done contextual before. This version is smarter, sure—but is it a revolution? Or just the same dog, new algorithm?

🌀 AI-washing everywhere.
“Multimodal inference.” “Cognitive scoring.” “Emotion-graph-based intelligence.” Please. Half these tools are just keyword engines with a thesaurus addiction.

🧩 No real standards.
Gracenote’s trying, but everyone else is still playing Calvinball. Until the industry agrees on how to define “context,” it’s just duct tape on a leaky boat.

📊 Performance is… chaotic.
For every Carl’s Jr. success story, there’s someone getting pitched a minivan at 2 a.m. during a slasher film.

🧠 ACT VI: The Power Move Beneath the Buzz

Let’s be honest—this isn’t just about technology. It’s about optics.

Contextual AI is the perfect boardroom safe word:

• ✅ It’s privacy-compliant.

• ✅ It’s buzzword-friendly.

• ✅ It won’t get you dragged into a GDPR audit or subpoenaed by Congress.

In other words: It’s adtech’s legal department-approved Hail Mary.
Not because it’s perfect—but because everything else is radioactive.

🎤 ACT VII: Final Scene—The Real Talk

Let’s land this.

📌 Is Contextual AI a scam? No.
📌 Is it overhyped? Absolutely.
📌 Is it the least embarrassing thing we’ve got right now? Without question.

It’s the one playable hand we’re holding while the rest of the deck catches fire.
It’s not “the future,” but it’s a future—and that’s better than the slow-motion behavioral data implosion currently unfolding.

So yes, you’re going to buy it. Probably this quarter.
Just don’t expect it to stop that toothpaste ad from landing smack in the middle of your late-night true crime binge.

Henry Innis to Adtech: Stop Chasing IDs Like It’s Black Friday at a Data Breach

Henry Innis from Mutinex just handed out the kind of blunt clarity this industry needs, minus the spin and hand-wringing.

“I think, broadly, that people chasing ‘identity’ are really chasing workarounds to consumer privacy,” he told ADOTAT. Translation: the ad world’s obsession with duct-taping together identity graphs and shadowy data pipes isn’t innovation—it’s desperation.

Henry didn’t stop there. He sees regulators sharpening their knives, ready to carve up the Frankenstein monsters built on non-consensual data. “Targeting and measurement really should not be reliant on tracking consumers unless the company has explicit consent.” Radical, right? Actually asking first.

And here’s the kicker: “Contextual targeting is a far more robust and privacy-safe way to target and probably will do the job better than cobbling together identity using data from all sorts of dubious places.”

In other words, stop looting the back alleys of the internet and start using your brain. Contextual isn’t just the clean option—it might be the smarter one.

From Hype to Hit: How AI Just Gave CTV Its AdSense Moment

Abbey Thomas, CEO of Anoki AI, laid it out bluntly to ADOTAT: CTV is finally catching up to its hype.

“Multimodal AI is the unlock,” Thomas said. “A few years ago, this wasn’t just difficult—it was impossible at the scale brands actually need.” Now, they’re mapping advertisers' queries to specific scenes inside vast content libraries. We’re talking emotions, sentiment, tone, objects, and every little thing happening in a scene—tagged and indexed at lightning speed.

This isn’t the usual AI-washing fluff, Thomas insists. It’s a full-blown shift, putting CTV on the same disruption tier as what AdSense did to the web.

Oh, and the cherry on top? “We built a query tool so brands can immediately get scene-level recommendations across FAST channels,” Thomas adds.

Translation: marketers can finally ditch the guesswork and plug this tech into their CTV strategy—right now.

🧠 CONTEXTUAL AI: IT’S NOT A BUZZWORD, IT’S A FLEX
Let’s stop pretending it’s just a nice-to-have.

The adtech world loves a shiny object. But here’s the uncomfortable reality: contextual AI isn’t some polite fallback because cookies are crumbling—it’s a full-blown revolution. And unlike your third-party ID graph that stopped working six Apple updates ago, this one’s already delivering.

🍕 Domino’s, Railways, and the Secret Life of Beekeepers

Ken Weiner, CTO of GumGum, isn’t here to play nice. He’s seen what happens when AI gets trained to understand context—not just keywords.

Take Domino’s: their campaign didn’t just perform—it obliterated expectations, raking in 135% to 398% ROAS depending on the creative unit. Then there’s Great Western Rail, which clocked a 134% better ROI than their benchmark. That’s not a rounding error. That’s a mic drop.

And then things got weirder—in a good way. One luxury retailer discovered their contextual sweet spot was beekeeping content. No, seriously. Turns out their fancy bags were buzzing (sorry) in videos about hives and honey.

Behavioral targeting would’ve missed that entirely. Why? Because it’s obsessed with who you are, not where your attention actually is. Contextual AI flips the script: forget the stalker vibes. Serve the ad when the mindset is right.

And if you think that’s just cute anecdotal fluff, Marika Roque, Chief Innovation Officer at KERV, has news: contextual isn’t just a vibe—it’s a revenue generator. “AI-powered contextual drives measurable ROI while optimizing the consumer’s viewing experience,” she says. KERV’s tracked increases in brand lift, engagement, purchase intent, and—yes—even lower funnel ROAS.

But here’s the nuance: broad contextual? Kinda like throwing spaghetti at the wall. Moment-based contextual? That’s the truffle oil. It doesn’t replace behavioral—it enhances it, surgically placing brands in the right moment without needing to stalk users like a jealous ex.

🧼 AI-Washing Is Real—And It’s Gross

According to Ken, everyone and their keyword parser is now claiming to do “contextual AI.” Spoiler: most of them aren’t. There’s the lightweight stuff—scraping metadata, URLs, and calling it a day. Then there’s the good stuff: true content-level analysis that reads the actual video, audio, text, and images.

And that’s the catch. Buyers are being sold “AI” when it’s really Excel with a wig. If you’re evaluating a partner, ask the real questions:

• Can it analyze everything in the content, or just keywords?

• Is it your tech, or did you borrow someone else’s API and slap your logo on it?

• Who’s verifying this isn’t snake oil?

Not all “contextual” is created equal, and if it’s not analyzing at the content level, it’s not contextual. It’s just cosplay.

📺 CTV: Contextual’s Biggest Blind Spot (And Opportunity)

Here’s where things get thorny. Contextual in display is one thing. But in CTV? That’s where marketers keep tripping over their own shoelaces.

Ken breaks it down: linear TV has decades of standardization. CTV? It’s the Wild West. Incomplete metadata, conflicting taxonomies, and platforms hoarding their own standards like Gollum with a spreadsheet.

No shared language = no scale. Unless...

🔗 Enter Richie Hyden: Building the Rosetta Stone of CTV

Richie Hyden, Co-Founder and President of IRIS.TV, knows contextual AI doesn’t work without a translation layer. His team isn’t analyzing the content themselves—but they are making sure everyone can speak the same language.

IRIS.TV created the IRIS_ID, a persistent content ID that lets buyers, sellers, and data providers all speak in one standardized voice. No, it won’t make HBO Max and Tubi start hugging—but it means content metadata doesn’t get lost in the shuffle.

And here’s the kicker: IRIS.TV doesn’t touch PII. They don’t even want it. Everything is server-to-server, with zero IP addresses, hashed emails, or device IDs flying around. Because let’s be honest—VPPA lawsuits are no joke, and they’ve already seen the future. It’s consent-optional content data, not third-party guessing games.

Meanwhile, Marika flags another operational blind spot: marketers want laser-focused context in CTV—the exact moment, the perfect vibe—but CTV was built for reach, not precision. That disconnect is where scale dies. “The next several months will be very interesting,” she says, hinting at fresh methodologies from the KERV brain trust to finally crack moment-based targeting at scale.

🛋️ The Couch, Not Just the Door

Richie put it perfectly: “The household ID gets you through the front door, but the IRIS_ID gets you to the couch.” You might know who is watching, but if you don’t know what they’re watching, you’re just throwing darts in the dark.

When PMG used this setup for Carl’s Jr., they didn’t just “optimize.” They crushed fragmentation and actually improved performance. Because let’s be real—no one’s watching a pre-roll for burgers during a documentary about famine. Context matters.

🔮 So… Is Contextual AI Just a Temporary Fix? Nope. It’s the Plan.

Ken’s not having the “identity 2.0” debate. For him, contextual AI isn’t a band-aid—it’s the future. Consumers are over being tracked like lost puppies. Ad blockers are up. Trust is down. But contextual doesn’t need your browser history to deliver a great ad. It just needs to know what you’re paying attention to right now.

Marika takes it one step further: contextual alone isn’t enough. It needs to be paired with dynamic creatives and measured outcomes. That combo—moment-based context, smart creative, real performance metrics—that’s the actual future. Not some mythical identity unicorn that may never show up.

GumGum’s research says nearly 80% of people prefer transparency and hate hyper-personalized ads. Contextual respects that. And it performs.

🔮 TL;DR: Contextual AI Is What Comes After the Cookie Funeral

• It works. Like… actual ROAS, not vibes.

• It’s privacy-friendly. Not just “trust us” friendly.

• It’s already fixing CTV fragmentation. Slowly, but surely.

• And if you’re still calling it a fallback plan? You’re not paying attention.

Welcome to the contextual era. The cookies are stale. The IDs are broken. And the bees? They’re still buzzing—with ROI.

📣 Why You Need ADOTAT+ (Unless You Enjoy Being Duped)

🧠 Everyone says they know AI. Most are just tossing buzzwords like pasta at the wall to see what sticks. Meanwhile, we're breaking down who’s selling snake oil 🐍 and who’s quietly rewriting the playbook 🎯—like Richard Howe, who was coding neural networks when you were still using dial-up.

💥 ADOTAT+ gets you the stuff no one else will print.
No fluff. No paid vendor spin EVER. Just savage insights, spicy receipts 🧾, and unfiltered commentary on the AI hype machine, the CTV context circus 📺, and the adtech arms race running on fumes and Mountain Dew.

🔓 What’s inside this week’s Insider Report:
— Roku’s "AI" that might accidentally show Psycho ads to toddlers 😱
— FreeWheel’s new duct-tape marketplace 🚧
— KERV vs. Anoki: The contextual showdown you didn’t know you needed ⚔️

💡 If you’re still living off free newsletters, you’re playing checkers in a chess match.

👉 Upgrade to ADOTAT+ — because the real players already have.

Become a paying subscriber to get access to this post and other subscriber-only content.

Already a paying subscriber? Sign In.