Taken together, these moves might look frantic but the impulse behind them is familiar. 

AdCP provides a shared language for how agents issue instructions to platforms, discovering audiences, activating them, orchestrating buys and curating supply. UCP defines the rules for how compact, privacy-safe signals can be shared so agents know what actions to take and how those actions perform. The Agentic RTB Framework, meanwhile, rewires the auction mechanics so agents can operate within the system rather than sitting on top of it as an afterthought.

The people behind these standards are working from a clear lesson. If the open web waits, AI agents will inherit every flaw programmatic already knows too well — opaque supply, brittle data, uneven governance — and then amplify them at machine velocity. Bad inputs become systemic failures. Weak guardrails metastasize. Bias runs unchecked across the marketplace.

“Everyone is trying to figure out for themselves how to make sure what they’re doing is compatible with the other constituents in the ecosystem,” said Ruben Schreurs, CEO of media management firm Ebiquity at the launch of AdCP. 

After years spent retrofitting RTB to survive privacy shocks, identity erosion and platform consolidation, the companies that live on the open web are trying not to get blindsided again — whether by its own hand or by the next wave of disruption. AI agents are coming, and the scramble now is to build enough structure before they scale in the wild.

That’s why none of the new standards attempt to blow up OpenRTB. They sit alongside it, a relief valve for what’s coming. Demand Supply Platforms don’t need to rebuild their systems overnight. Publishers can test agent workflows without detonating yield. Buyers avoid another abrupt operational overhaul. Supply-side platforms get time to shift more processes into cloud-native environments.

Put another way: the agentic era won’t open with a single switch flip. It will emerge through a long, awkward hybrid phase. Even the groups pushing these standards concede that the rollout will be staggered. The internal politics will slow things down. Data pipelines will need real reconstruction. The tools built on top of these protocols will need to show they are stable, compliant and, most of all, safe for brands.

But if those moving parts line up, the shift could be foundational — even in areas as plodding how long it takes to run an auction. Take the IAB Tech Lab’s Agentic RTB Framework, for instance, which takes the existing RTB process and redeploys it in a containerized environment so the different parts of a programmatic transaction run much closer — and therefore much faster — together.

Index Exchange CEO Andrew Casale captured the shift at Exchangewire’s ATS event in September. The market, he said, is moving from milliseconds to microseconds. That level of resolution offers deeper visibility into auction dynamics, exposes inefficiencies and pushes the industry toward more computationally efficient ad delivery — lower cost and tighter control. 

“We still have a long way to go, because if you look at finance, which is sort of the bar that we hold ourselves to, that industry measures itself in nanoseconds,” said Casale at the event. 

For the first time in years, the open web is trying to architect its future before the platforms dictate it. That ambition rests on several conditions: the standards can’t splinter. Cloud infrastructure stays competitive. Agent governance evolves past white papers. Publishers move from sitting on standards committees to deploying agents that matter to their bottom lines. Buyers move toward outcome planning that doesn’t treat walled gardens as the default answer. 

Should all of that fall into place, the open web could be on the cusp of its biggest reinvention since programmatic took hold. If not, the next decade will have the familiar contours of the last: incremental change and a handful of dominant platforms setting the terms.

“This sort of adaptation [of RTB] is definitely the way to go — at least in the short term,” said Hann. “The speed of evolution will be what dictates what happens afterwards.”

No matter the trajectory, none of this turns the open web around on its own. Infrastructure won’t magically deliver performance. Protocols won’t reconcile the economic interests of buyers and sellers. But this is the most coordinated the industry has been in a long time — and the most motivated to avoid repeating its past mistakes as automation becomes more autonomous. 

The next year will show whether this early push gains momentum or stalls out while companies jockey for advantage. For now, the upside is clear: the shift ahead doesn’t require ripping up the entire floor. Not yet. 

— Seb Joseph

The new discipline behind Unilever’s creator and AI choices 

Earlier this month, Digiday caught up with Ryu Yokoi, chief media and marketing capability officer of Personal Care and North America at Unilever. Below is an excerpt from that conversation, lightly edited for clarity.

When it comes to working with creators, there’s less control over execution for brands. How are you thinking about brand suitability, as you expand your influencer investment?

Casting is super important, and so we’re really careful about which cohorts we want to, who we want to build relationships with… we’re extremely scrutinous of them, of what we choose to amplify. Our brands are our most precious resource that we have as a company.

What about Sora 2? OpenAI’s been inviting brands to use its Cameo feature. Are there brand suitability concerns there?

I have not yet had a chance to to play with it myself. But we are looking at all the tooling. We AI can be an unbelievable accelerant for the creative process, testing, measurement across the gamut. And it’s we think this is like a fantastic opportunity that we have to make our marketing better.

Can AI  work alongside the positioning of your brands? When I spoke to your colleagues earlier this year, they did talk about red lines around AI generating people and faces.

That’s a Dove-specific rule we have [but] we’re very careful with how we use those things. One of the areas we’ve really focused on is on building training sets for our brands… that we can use to make sure that the output we’re getting out of the LLMs is applicable for our brands… [and that is fitting with our brand guidelines.

— Sam Bradley

Numbers to know

$4.3 million: The amount of revenue Disney loses daily thanks to the YouTube TV blackout

$24 billion: The amount Disney plans to spend on entertainment and sports in 2026

62%: Percentage of marketers who described their video workflows as overwhelming, stretched or blocked, even with AI in the mix.

10%: Percentage of Meta’s 2024 revenue that was projected to come from ads for scams and banned goods

What we’ve covered

Amazon rebuilds its ad machine for the market

During Amazon’s unBoxed conference, Amazon introduced a unified Campaign Manager that merges its DSP and the Ads Console into one buying tool.

Buyers aren’t exactly won over by Netflix’s new monthly active viewers (MAV) metric

Just a week since Netflix launched its move from MAUs to MAVs, but marketers aren’t necessarily sold on the idea. Instead, they’re taking a cautious approach, as to them, it feels like a step back against transparency.

Behind Unilever’s creator and social strategy for next year’s World Cup

With next year’s FIFA World Cup expected to be the largest spectator event in history, Unilever is already planning how to capitalize on the billions of viewers watching via paid media and sponsorships.

What we’re reading

Anthropic to outpace OpenAI in server efficiency, internal projections show

The divergence between how much Anthropic and OpenAI forecast to spend on computation resources mean they’ll both reach similar breakthroughs, but their finances will be in very different places, according to The Information.

TikTok Shop is now the size of eBay

TikTok Shop totalled $19 billion in Q3 sales, not far off eBay’s $20.1 billion, according to Wired.

TikTok and iHeartMedia strike partnership for creator podcasts

The two companies unveiled the TikTok Podcast Network, featuring up to 25 creator-driven podcasts and studios in LA, NYC and Atlanta, according to Hollywood Reporter.