LLM Browser Wars are here

LLM Browser wars

Digiday took a deep dive yesterday into the implications for advertising with AI company browsers – such as the one launched by Perplexity or a rumored one by OpenAI.

“What started out as AI platform wars is about to turn into browser wars as Perplexity’s Comet and OpenAI’s soon-to-be-released browser aim to go head-to-head with Google’s Chrome. But maintaining a browser is an expensive business – and that’s where advertising comes into play.”

Read it (July 14).

Once again, it’s all about the data.

LLMs & CHATBOTS

Zero-click commerce

On the “Adlingo” blog, writer Travis Lusk stepped back from the browser discussion – for a moment – and aggregated an interesting nugget on the state of Perplexity ads today:

“Media buyers describe the platform as ‘still in wait-and-see territory’ with ‘limited scale and focus on awareness’ six months into advertising operations, agencies want more volume and performance options beyond the current sponsored questions format.”

But, Mr. Lusk also noted recent rumors of Apple buying Perplexity and the potential opportunity to be unlocked:

“What Perplexity is building represents a fundamental reimagining of how advertising could work in an AI-native world.”

”Traditional display advertising assumes users browse, click, and convert elsewhere. Perplexity’s model assumes the AI handles the entire journey (research, comparison, and purchase) in a single conversational thread.”

”They call it ‘zero-click commerce.’”

This “zero-click commerce” opportunity doesn’t seem unique to Perplexity. But, it is unique to LLMs as consumers move from search to chatbots (a.k.a. answer engines).

An LLM’s functionality paired with Apple and its device/browser footprint appears to be a juicy “zero-click commerce” proposition.

LLMs & CHATBOTS

Perplexity’s new os.ai domain

Hubspot co-founder and CTO Dharmesh Shah said on Sunday that he sold the “os.ai” domain to Perplexity yesterday and that Perplexity CEO Aravind Srinivas confirmed it – which he did.

Beyond the revealing look at Mr. Shah’s domain sales expertise, clicking the os.ai link takes you to Perplexity’s browser, Comet. Shah suggested that a plan for an operating system (OS) for AI is in the works at Perplexity.

On LinkedIn, Mr. Shah wrote:

“BREAKING NEWS: I have sold the domain os .ai to Perplexity.

Aravind Srinivas shared the news earlier tonight.

I am not at liberty to share details of the deal, but will say this: I was already a shareholder in Perplexity before (I love the product, as do millions of others). Now I’m a somewhat larger shareholder.

I’ve written previously about the idea of an OS for the age of AI.

Decades ago, we saw the PC wars. I was there for that.

Then we saw the battle for the browser.

After that, the fight for the mobile OS.

It makes complete sense for an OS for AI to exist. Every big technology shift brings a new kind of OS.

It is inevitable. And exciting.

I’m here for it.”

Is Perplexity truly building an OS for AI – or – is Perplexity looking to boost its value prior to being acquired? Or both?

Mr. Shah reportedly bought the os.ai domain name for $150,000 back in March. Read that one.

Read: Perplexity buys OS.ai domain to build AI operating system (July 14) – PPC Land

AGENCIES

The New WPP CEO and its AI platform

Adweek’s Alison Weissbrot said WPP CEO Cindy Rose has several challenges ahead when she comes aboard on September 1 – not the least of which is WPP Open.

“Under [outgoing CEO Mark] Read, WPP invested millions of dollars into WPP Open, its internal AI platform. It will be Rose’s job to make that bet pay off—and set WPP up for an AI-powered future. ‘The significant investment into WPP Open hasn’t yielded the growth that was hyped a couple of years ago,” said [former CEO of WPP agency AKQA Ajaz] Ahmed.”
“According to Jay Pattisall, principal analyst at Forrester, WPP chose a tech executive with an AI background for exactly that reason: ‘The selection of Rose is designed, in part, to further commercialize WPP Open,’ he said.”

Read: 5 Challenges Facing New WPP CEO Cindy Rose (July 14) – Adweek

TECH

Soon from Snap: AR targeting + AI

On the most recent Khanversation podcast, Imran Khan invited his former boss, current Snap CEO Evan Spiegel, to discuss the past as well as future plans for “Specs” – his company’s glasses for augmented reality (AR) which is slated for launch next year.

Created entirely in parallel to Snap’s main app business over 11 years, Mr. Spiegel described the Specs’ hardware project as an “AI front end” with its own “proprietary vertical stack.”

If it works, a new data contextual data stream – among other data types – will be available to LLMs, and probably the advertising industry even though advertising was not explicitly mentioned.

Mr. Spiegel explained the augmented reality + AI opportunity his company is pursuing [starts 40:23]:

“The best way to think about it is… augmented reality – or ‘Specs’ – as the user interface for AI.“

“Today, the primary way people interact with AI is through chatbots, which is one way to interact with AI. But, it’s still very high friction. Because the AI doesn’t have a lot of context about where you are, what you’re doing and so on.(…)”

“One of the issues with a chat-based interface is how much work you have to do to describe your problem to AI. So, if you’re trying to get your car fixed, you have to write out all the issues you’re having with your car and what’s going wrong and so on.”

“One of the fascinating things about glasses is that they’re context aware.”

“They know what you’re looking at -and not only that, they’re able to render into the real world tips and instructions from AI.”

“And so one of the really exciting things we just announced at AWE (Augmented World Expo) was the ability for Specs to pass their world understanding – their depth map – to large language models so that large language models can actually annotate the world and help provide instructions.”

“We gave some examples: your car is overheating and you pop the hood and say, ‘Hey, my car is overheating, What should I do?’ And AI can actually annotate the engine compartment by fixing itself to the right places in space, which is a lot more helpful than trying to solve that same problem through like a text-based interface.”

“Depth map” – will it be a new framework for advertisers and, maybe, their brand safety tools?

Hear more on the Apple Podcasts app.

TECH

Ad tech M&A

Given Large Language Model (LLM) startups’ growth and endless access to capital, the untapped market of ads is singing its siren song and it is only a matter of time until one or more jumps at the ads opportunity. So, when ads, LLMs?

To be clear, I don’t think LLMs will initially focus on AI-enabled ads.

(Analyst Eric Seufert made a thoughtful argument against AI ads for LLMs on the Intelligent Marketer podcast a couple of weeks ago.).

In the near term, LLMs are publishers and all about feeding the LLMs’ revenue beast. So let programmatic reign and the big ad dollars flow – with AI ads down the road.

Look at these recent valuations of the largest, private LLM players: xAI, OpenAI and Anthropic:

Elon Musk’s xAI seeks up to $200bn valuation in next fundraising (July 11) – Financial Times
“Today we’re announcing new funding—$40 billion at a $300 billion post-money valuation…” (March 31) – OpenAI
Amazon-backed AI firm Anthropic valued at $61.5 billion after latest round (March 3) – CNBC

Barring a black swan event, I’ll guess valuations double in 6 months given current momentum. So, if true, by the end of this year:

OpenAI valuation will reach $600 billion
xAI will be at $400 billion
Anthropic will reach something close to $200 billion

That’s $1.2 trillion worth of AI companies with no advertising plan, yet. How about some ad tech?

Now acqui-hiring

These acquisitions will ultimately be important from a “people” point-of-view. AI companies need people who understand advertising and its tech – from engineer to executive.

And as I said, my guess is it’s programmatic to start, then AI-enabled ads in the future.

This could mean everything from a “tuck-in” for LLM chatbot ad efforts to multi-billion dollar deals that supercharges the LLM’s revenue and provides access to, say, inventory sources of interest – which in turn could lay the ground work for a wider AI-infused advertising network. (Or AI operating system?)

LLMs could go the route of hiring, too. But I’d argue speed is critical and piecemeal hiring won’t cut it.

Over at X and xAI, X does not seem compelling for ads as it relates to a differentiated AI-enabled product. You don’t hear a lot of advertisers screaming from the rooftops that they love their X’s ads. And there was no “X Performance Marketing Summit” that I’m aware of as there was for Meta in June.

But if Mr. Musk were to feel the need to lean harder toward revenue opportunities with xAI and its associated chatbot Grok in the near term, then more ad tech might make sense.

Still, as Eric Seufert observed last week, Mr. Musk may not care about ads in the end. X and xAI’s efforts are squarely focused on building the world’s best LLM which will have a myriad of revenue implications not to mention enormous influence on the global stage.

TECH

Scope3 AI on Amazon DSP

Head of UK & Northern Europe at Scope3, Tim Collier, announced that his firm was live on the Amazon Ads DSP. Among the features, Mr. Collier listed on LinkedIn:

“Brand standards → AI-driven models that help brands maintain their values by automatically filtering spend away from unsuitable content and reinvesting it into brand-aligned, effective media.”

More here on Scope3’s “Brand Standards” from early June.

Is Scope3 CEO Brian O’Kelley’s “vibe targeting” the next integration?

SELL-SIDE

IAB wrangling the LLM crawlers

IAB Tech Lab CEO Anthony Katsur reviewed his organization’s “LLM Content Ingest API Initiative” with AdExchanger yesterday.

“He explained that there are four major components.”

“The first is access controls, which determine who is allowed to access a publisher’s content in the first place. Once controls are established, access terms come into place, such as licensing models and content tiers. Under the IAB Tech Lab’s guidelines, content will be segregated into tiers based on relevance and value…”

Read the other three on AdExchanger. (July 14)

Also read:

IAB’s press release on the API (June 4).
Visit the API’s hub on the IAB website.

PROMPT

What does an AI advertising operating system look like?

Response from Perplexity.ai:

An AI advertising operating system (AI Ad OS) is a unified, intelligent platform that automates and optimizes the entire digital advertising lifecycle. Unlike traditional ad tech stacks—which are fragmented across many tools—an AI Ad OS brings together campaign planning, creative, targeting, execution, analytics, and optimization within a single, AI-driven environment.

Core Features and Architecture

Unified Platform:
AI Agents and Automation
Natural Language Interface
Real-Time Data Integration
Predictive and Prescriptive Analytics
Personalization at Scale
Creative Generation and Scoring
Fraud Detection and Security