The Trade Desk is giving brands a new way to serve contextual sponsored product ads. The ad tech firm announced a new integration with startup Shopsense AI on Thursday to bring sponsored product ads to publisher content across the open internet.

The deal, made through The Trade Desk’s OpenPath supply integration, lets advertisers place shoppable product listings directly next to relevant editorial content—like a Nike shoe ad beside a marathon training article or a cookware link under a recipe video.

“For The Trade Desk, [this integration] creates a bigger surface area on the inventory side, creating a higher relevancy to a different kind of commercial opportunity,” said Will Doherty, svp of inventory development at The Trade Desk. This effectively turns impressions across omni-channel platforms like CTV, web, and audio into shoppable media placements, he added. 

The move follows The Trade Desk’s recent expansion into retail media. ADWEEK previously reported that advertisers will be able to programmatically buy sponsored product ads on delivery app Gopuff through The Trade Desk in a deal facilitated by commerce adtech firm Koddi. The Shopsense integration extends that strategy beyond retailer environments and into premium publisher inventory.

This integration dovetails with an increase in consumer tolerance for commerce-driven media. A December 2024 Bain & Company study found that 45% of consumers don’t mind sponsored ads if they’re relevant, and 40% say well-executed ads can actually help their shopping experience.

Shopsense, founded in 2023 by former Amazon executives, uses a mix of generative AI and machine learning to analyze images, text, and video on publisher pages and extract commerce intent. The system identifies products in content—such as sunglasses on a celebrity or a team jersey in a game recap—and converts them into “targetable parameters” advertisers can buy against, including search queries, Amazon product IDs, or Global Trade Item Numbers (GTINs, a unique number that identifies a product or service), according to Bryan Quinn, president and co-founder of Shopsense.

Shopsense works with media companies including Disney, Fox, Paramount, Vox Media, BuzzFeed, and People Inc. The integration with The Trade Desk allows clients to activate relevant sponsored ads on those publisher sites. 

“We see a great opportunity to merge upper-funnel storytelling with lower-funnel performance via The Trade Desk,” Quinn said. 

Once Shopsense AI identifies product context on a page, it sends a bid request using the OpenRTB protocol. The Trade Desk evaluates whether it has relevant sponsored product ads to serve, and if it wins the auction, the placement appears as native, shoppable media embedded on the page. Brands can layer this inventory alongside existing CTV, display, and audio buys.

“Our clients have CTV running, they have web running, they have audio running, and now they can bring their product listings into the same experience, expanding omni-channel experience,” Doherty said. 

The numbers

$50.1 billion: Meta’s ad revenue, up 26% YoY. Overall revenue was $51.24 billion, up 26% YoY

+14%: YoY increase in ad impressions delivered across Meta’s Family of Apps.

10%: Increase in average price per ad. 

$15.9 billion: a one time, non cash income tax charge on deferred tax assets as a result of Trump’s One Big Beautiful Bill Act

$1.05: Earnings per share, significantly lower than the average analyst estimate of $6.72. (Attributed to the $15.9 billion one-time tax charge)

3 billion: Instagram hit a major milestone with monthly active users 

Watercooler talk

Meta’s accelerating AI investment is now the dominant financial storyline. The company is pouring more than expected into data centers, compute, and cloud capacity, prompting investor concern over whether the return on AI will meaningfully scale beyond ads—and on what timeline.

That increase in AI costs, coupled with a one-time hit to the tune of $15.93 billion tied to tax changes following the implementation of the “One Big Beautiful Bill” signed into law by President Trump in July, sent shares down as much as 10% in extended trading Wednesday.

The social media giant has been on a massive spending spree, investing billions to hire talent and expand the data centers needed to support its growing AI ambitions. Meta now employs more 78,400 people, up 8% YoY, with employee compensation growth accelerating, particularly for technical hires in AI.

Meta CFO Susan Li said Meta’s total expenses in 2026 will grow at a “significantly faster percentage rate in 2026 than 2025,” driven primarily by infrastructure costs, including incremental cloud expenses. She added in a call with analysts that ad revenue will help fund those investments. 

The annual run rate of revenue running through Meta’s end-to-end automated solutions has reached $60 billion, according to Li. Meanwhile, its AI recommendation systems on Facebook, Instagram, and Threads are delivering more relevant content, boosting time spent on Facebook by 5% in Q3 and on Threads by 10%, per CEO Mark Zuckerberg. Reels now has an annualized revenue run rate of over $50 billion, he added. 

Still, Meta signaled that AI infrastructure needs outpace internal forecasts. Li said the company now expects to “invest aggressively” in compute next year—building new data centers while also paying third-party cloud providers—which will push capital spending and overall expenses higher in 2026.

Key quotes

When an investor pressed Meta on whether it’s over-investing in AI infrastructure, Meta insisted the company remains “underbuilt” on AI infrastructure despite swelling capex.

“Almost every week, people come to us from outside the company asking us to stand up an API service or provide compute. We haven’t done that yet, but if we ended up overbuilding, that could be an option,” Zuckerberg said. “The very worst case would be that we’ve essentially prebuilt for a couple of years. There would be some loss in depreciation, but we’d grow into that capacity over time.”

This article is part of ADWEEK’s Marketing Vanguard program and was written in collaboration with one of its members.

For the last two years, every marketing conference has focused on the same topics: ‘AI for Efficiency,’ ‘AI and ROI,’ ‘Will AI Replace Creatives?’ But while the industry obsesses over automation and cost savings, they’re overlooking three huge shifts in how consumers use AI that marketers must understand in order to harness the technology.

AI has rewritten the rules of brand discoverability

Consumers are already discovering brands through AI platforms. According to a recent Forbes study, that traffic converts at 9x the rate of traditional search. Yet most marketers still see AI as a content creation tool, not a discovery engine.

Brands need to focus on Answer Engine Optimization (AEO) in addition to SEO. And unlike SEO, AEO rewards clarity, credibility, and differentiation rather than backlinks. Brands with sharp positioning and clear narratives will surface in LLM responses. Generic brands won’t. 

Marketers also need to understand how Retrieval Augmented Generation (RAG) actually works, including how LLMs pull from structured data and trusted sources. Then, they should prioritize creating content in places where they’re most likely to be cited.

And this shift is accelerating fast. AI referral traffic to news and media sites jumped 770% year over year, according to TechCrunch. The early mover advantage here is real. Unlike SEO, which takes years, startups can win at AEO immediately if they’re strategic about being cited and sourced in the right places. 

What AI can’t do is a marketers’ biggest advantage

Which brings us to the second thing marketers are missing: what actually drives discoverability in the first place. Strong communication skills, like authentic storytelling, are suddenly as valuable as technical skills. The shift to AI discoverability rewards exactly what brand-builders have been defending for the past decade: long-term investment in narrative, trust, credibility, and customer experience. 

Marketers are focused on what AI can do (write copy, analyze data) instead of what it can’t do (original thinking, strategic clarity).

In a world drowning in AI-generated content, the ability to articulate a clear, differentiated strategy is the competitive advantage.  

Take Charli XCX’s brief for Brat Summer. It was bold, specific, and unapologetically clear about the cultural moment she wanted to own. The viral success and her 2024 award season accolades speak for themselves. That’s what a breakthrough brief looks like, and you’re not going to generate that with a prompt.

Curiosity will define the best AI teams

Finally, teams need to actually experiment with these tools—not just sit through vendor presentations. The best way to learn is to play in the dirt. However, most companies have locked teams into one AI platform via enterprise agreements. They think they’ve “solved” AI training by signing a contract. 

The problem with this is that it’s way too early to declare winners in AI. Teams need hands-on experience with multiple LLMs (ChatGPT, Claude, Perplexity, Gemini etc.) to understand their differences. CMOs need to create structured training time. They should not expect people to figure it out on nights and weekends. 

Further, CMOs need to teach teams how to prompt effectively, help them understand how LLMs source and cite information, and identify the content territories where your brand has real authority. 

The AI conversation will keep happening, and if marketers stay focused only on efficiency, they’ll miss the plot entirely. The brands that win won’t be the ones who automated fastest. They’ll be the ones who stayed discoverable, stayed differentiated, and stayed curious.

Lauren Weinberg is a member of Marketing Vanguard, ADWEEK’s exclusive community and intelligence resource for some of the most innovative and accomplished brand-side CMOs. Built on ADWEEK’s editorial foundation and shaped by the needs of today’s marketing leaders, Marketing Vanguard is an application-only program built for learning, connection, and personal and professional growth.

British advertising titan WPP is rolling out a self-serve version of WPP Open, its AI-powered marketing platform. 

Unveiled Thursday, WPP Open Pro is a platform marketers can use to build campaigns, develop creative assets, and activate on their own, with the help of AI. 

The platform is designed to streamline the entire marketing process, from ideation to execution, with the help of automated tools. 

Users can craft campaign strategies with the help of AI agents that pull from a well of proprietary WPP data, partner data, and broader industry insights. Then, the platform can help marketers generate branded content for various channels.

To deploy campaigns created within the system, users can publish directly to major ad platforms. WPP Media clients can also connect to WPP’s Open Media Studio, which offers more advanced audience identification, bid optimization, and programmatic management. 

Image- and video-generating features in WPP Open Pro are powered by a range of generative AI models, including Google’s Imagen and Veo platforms, OpenAI’s DALL-E, Sora, and GPT Image, Adobe Firefly, and others. Text generation, meanwhile, relies on Google’s Gemini, Anthropic’s Claude, and OpenAI’s ChatGPT. 

Courting Small Business

WPP expects the tool to meet the needs of smaller enterprises and performance marketers who don’t need the high-touch managed service version of the program—and might also find uses in the organizations of existing WPP clients.

“We often don’t service the entirety of a client’s business, so it gives us reach into business units and areas that we might not service already,” said Matt McNeany, who helms the Open Pro project at WPP. 

He added that smaller businesses open up a new market for WPP beyond its typical slate of Fortune 500 clients. “Now they can get access to professional-grade tools and AI capabilities, and do that in a pretty straightforward way.”

Users pay a fee to access the tool and are charged by usage, though WPP declined to share specific pricing details. 

WPP Open Pro will be piloted by Google and some other undisclosed WPP clients. Initial feedback from clients has been “tremendous,” according to McNeany.

Google and WPP earlier this month announced a five-year extension of their ongoing partnership focused on advancing AI and cloud technologies for marketing. As part of that arrangement, WPP has pledged $400 million to bake Google AI tools into WPP Open. 

Google declined to comment.

AI push

WPP Open Pro comes as the holding company aims to further align itself with AI to turn around its sinking fortunes.

Since it launched last year, WPP Open got a shiny upgrade in June with the launch of Open Intelligence, a foundational LLM trained specifically for marketing use cases. The platform has also added an integration with TikTok’s generative AI tools. Today, it’s used by brands including L’Oréal, LVMH, and Nestlé. 

The launch of WPP Open Pro is among the first major announcements under new WPP CEO Cindy Rose, the Microsoft and Disney alum who succeeded Mark Read on September 1.

Rose’s appointment came after a grueling few years under Read’s leadership in which WPP suffered significant profit losses, client departures, and felt the squeeze of Trump-era tariffs—all while battling rivals over AI promises. 

The rollout of WPP Open Pro, she hopes, will give the holdco the opportunity to expand into new markets.

In a statement shared with ADWEEK, Rose said WPP Open Pro’s low barrier to entry will put the company’s AI marketing capabilities “directly into the hands of a much wider array of brands and businesses.” If successful, the shift could drive revenues and bolster WPP’s business more broadly. 

A marketing leader at a global brand, speaking on condition of anonymity, told ADWEEK this week that WPP has been pushing the brand aggressively to do more with WPP Open. 

McNeany, for his part, is confident in the positioning; he suggested that the tool surpasses rival agencies’ AI-powered marketing offerings today.

“[We’re] taking our data assets, our partner data assets, our agentic capability, the generative partnerships that we’ve got with Google, OpenAI and so on…content automation, media activation, and bundling that up into one solution,” he said.

Reddit sued AI startup Perplexity and three data-scraping firms Wednesday, alleging they collected and resold Reddit posts without permission.

According to the complaint filed in Manhattan federal court, Oxylabs, AWMProxy, and SerpApi scraped Reddit data via Google search results, and Perplexity purchased it from at least one of the vendors.

The lawsuit claims the defendants masked their identities, hid their locations, and disguised web scrapers to bypass Reddit’s security measures. Reddit says it caught Perplexity “red-handed” using digital markers to confirm the AI startup was accessing scraped content, and that the company ignored a cease-and-desist warning about the commercial use of its data.

“In fact, Perplexity’s citations to Reddit increased forty-fold after Reddit told it to stop,” the complaint states. “As an advertised client of SerpApi, there can be little doubt where and how Perplexity is getting its illicit Reddit data.”

Reddit’s content has become a sought-after asset for AI companies, which rely on massive datasets to train models and surface relevant results. The company has licensed its data to OpenAI and Google but is taking legal action against firms it says are using its assets without permission, following a similar lawsuit against Anthropic earlier this year.

“AI companies are locked in an arms race for quality human content—and that pressure has fueled an industrial-scale ‘data laundering’ economy,” Reddit chief legal officer Ben Lee said in a statement. “Defendants Oxylabs, AWMProxy, and SerpApi—ranging from a Lithuanian scraper to a former Russian botnet—are textbook examples. Perplexity is a willing customer of at least one of these scrapers, choosing to buy stolen data rather than enter into a lawful agreement with Reddit itself.”

Perplexity, in a response posted on Reddit, denied ignoring the company’s inquiries about AI training and pushed back on the lawsuit.

“We won’t be extorted, and we won’t help Reddit extort Google,” the company said, calling the lawsuit “a sad example of what happens when public data becomes a big part of a public company’s business model.”

SerpApi and Oxylabs both learned about the lawsuit through ADWEEK’s request for comment and denied any wrongdoing.

“We strongly disagree with Reddit’s allegations and intend to vigorously defend ourselves in court,” said Ryan Schafer, customer success director at SerpApi.

Oxylabs Chief Governance and Strategy Officer Denas Grybauskas also sent a statement: “Oxylabs has always been and will continue to be a pioneer and an industry leader in public data collection, and it will not hesitate to defend itself against these allegations.

“Oxylabs provides infrastructure for compliant access to publicly available information, and we demand every customer to use our services lawfully.”

AWMProxy could not be reached for comment.

OpenAI’s ChatGPT has been nipping at Google’s search dominance for months. Now, the AI startup is taking Google head-on with an AI-powered browser that can shop, schedule, and search on behalf of users.

Launched globally on Tuesday, Atlas has a ChatGPT agent that can navigate the web and complete tasks—from booking reservations to adding groceries to a cart. The product gives OpenAI a direct foothold in search and online activity and access to the kind of data that powers targeted advertising and commerce.

Atlas is available on macOS, with Windows, iOS, and Android versions coming soon. 

With ChatGPT in every tab, Atlas appears to offer a more personalized browsing experience. Users can ask questions, draft emails, conduct research, shop, or schedule tasks without switching pages.

The browser draws on activity—including open tabs and logged-in sites—to tailor responses, resurface past work, and, in “agent mode,” take actions like adding items to a cart or summarizing content.

By default, Atlas displays a split-screen whenever a user clicks on a search result, showing the webpage it came from alongside a live transcript from ChatGPT.

During a livestream, OpenAI employees said this layout, which users can disable, is designed to provide a constant AI “companion.” The company also demonstrated Atlas’s ability to summarize webpages and its “cursor chat” feature, which lets ChatGPT edit text directly in emails or other fields.

“AI represents a rare, once-in-a-decade opportunity to rethink what a browser can be. The way that we hope people will use the internet in the future, and that we’re starting to see, is that the chat experience and a web browser can be a great analog,” said OpenAI CEO Sam Altman.

Browser wars

Atlas launches as the race to dominate AI-powered search browsers has intensified in recent months.

OpenAI previewed its SearchGPT engine in July 2024, while Perplexity launch AI-powered browser Comet, which can summarize videos, declutter inboxes, and even make Amazon purchases.

Google, meanwhile, has been embedding its Gemini AI assistant into Chrome. It promises Gemini will be able to handle tasks like grocery shopping and scheduling soon, but has not yet confirmed a roll out date.

The moves come as Google faced a softer outcome than expected in its antitrust trial, avoiding a breakup while retaining control of Chrome and Android—though its dominance over search data will be slightly curtailed.

Alphabet shares slipped nearly 5% Tuesday following Atlas’ debut.