1 OpenAI moved to automate the last thing advertisers still did by hand: the creative
OpenAI quietly updated its Ad Tools Terms to allow "AI-powered Creative Tools" that let advertisers generate, modify, optimize, localize, and translate ad creative inside ChatGPT's ad system — the one part of the pilot that, until now, advertisers had to upload themselves. The same update added conversion tracking for app installs and opens, and doubled the daily ad budget cap from $100 to $200, all as the company is reportedly preparing to confidentially file for an IPO.
Once algorithmic optimization ate media buying, creative became the last real variable — and the last manual job. OpenAI closing that gap isn't a feature; it's the platform removing the final reason you needed a person in the loop. Note the fine print: OpenAI says the advertiser is "responsible" for accuracy and compliance, and takes no liability for what the machine generates. The work gets automated; the accountability stays with you. Plan your QC process accordingly.
2 Databricks launched CustomerLake — an agentic CDP that replaces campaigns with "infinity loops"
At its Data + AI Summit, Databricks entered martech with CustomerLake, an agentic Customer Data Platform built natively on the lakehouse that unifies customer data, identity resolution, audience building, and activation — then puts agents on top. Profile Agents turn raw data into Customer 360 profiles; Campaign Agents build audiences, recommend next-best actions, and continuously optimize. The pitch: replace one-off campaigns with "infinity campaigns," continuous agentic loops that react to context and personalize "a billion times a day." It's in private preview with HP, Circle K, AB InBev, and Getnet by Santander, with IAS as a first-party-data launch partner.
The CDP category was already crowded; Databricks' angle is that the data never leaves the lakehouse, so the agents run where the data already lives. That's a direct shot at the "copy your data into our walled CDP" model. If you're an enterprise sitting on a Databricks footprint, the build-vs-buy math for your customer data stack just changed — and the unit of work shifts from "launch a campaign" to "supervise an always-on loop." The new skill isn't briefing a campaign; it's auditing what the loop decided.
3 Salesforce paid $3.6B for Fin (formerly Intercom) — the biggest agentic-CX deal yet
Salesforce signed a definitive agreement to acquire Fin, the company formerly known as Intercom, for roughly $3.6 billion — folding Fin's "customer agent" into Agentforce. Salesforce says Fin resolves, on average, 76% of support volume with no human, powered by a proprietary model called Apex that Fin claims beats frontier models from OpenAI and Anthropic on resolution rates. The deal brings roughly 30,000 customers and is the largest acquisition of an agentic customer-experience provider to date.
Companies that can't build agents fast enough are buying them — and the price of a proven agent is now measured in billions. The tell here is the 76% number: support, long treated as a cost center, is becoming the most automatable customer surface in the company, which makes it the first place leadership will look to convert headcount into software. For marketers, customer service is also where post-purchase experience and retention live; if that conversation is now an agent, your brand voice has to be designed into the bot, not bolted on after.
4 Amazon Ads piped LinkedIn's B2B data into streaming TV
Amazon Ads' streaming-TV inventory can now use first-party data from LinkedIn — routed through the supply-side platform of LinkedIn's owner, Microsoft — letting advertisers on Amazon DSP target business-to-business audiences inside premium Prime Video environments. Amazon paired it with a new Dynamic Creative TV tool that personalizes interactive video ads in real time based on a viewer's shopping behavior and purchase stage, with expansion to live sports and third-party Prime Video channels planned for Q3.
B2B advertising has always struggled on TV because you couldn't tell the decision-makers from everyone else on the couch. Wiring LinkedIn's professional graph into Amazon's screens quietly solves that — and it's a reminder that the Microsoft–Amazon data handshake creates targeting neither could offer alone. If you sell to businesses and wrote off CTV as a brand-only play, that assumption just expired. The interesting part is the alliance: the walls between the big data owners are turning into toll roads, not barriers.
5 Walmart Connect opened its first-party data — and Vizio's screens — to outside buying platforms
Walmart Connect began syndicating its shopper data and Vizio ad inventory through Yahoo DSP, using Magnite's supply-side technology for audience activation, in a closed proof-of-concept that GM Ryan Mayward framed as connecting "strategies that have historically operated in silos." Walmart also turned on self-service campaigns powered by its shopper data on Meta, added an add-to-cart feature for managed Meta/TikTok/Pinterest campaigns, and extended closed-loop measurement through its LiveRamp clean room to Meta.
For two years the retail-media story was "every retailer builds a walled garden." This is the opposite move: Walmart pushing its most valuable asset — purchase data — out across other people's buying environments to capture spend wherever it lives. The strategic bet is that reach plus measurement beats exclusivity. For brands, it means you'll increasingly buy Walmart audiences without ever logging into Walmart Connect — and the question shifts from "which retail network?" to "whose data is inside the DSP I already use?"
6 Comscore reframed AI-cited traffic as the most valuable traffic you can get
Comscore's Custom IQ head, Smriti Sharma, said AI-driven referral traffic can be five to seven times more engaged and more monetizable than traditional traffic, and argued that AI citations are becoming a core visibility and trust metric for publishers and brands as journeys move beyond classic search-and-click. The framing: being named inside an AI answer isn't a vanity metric — it's a higher-intent, higher-value visit.
This is the data point that turns Answer Engine Optimization from a hedge into a priority. If a citation from ChatGPT or Google's AI Mode brings a visitor worth 5–7x a blue-link click, then "are we getting cited?" becomes a revenue question, not an SEO curiosity. The catch is volume: AI traffic is still a fraction of total referrals, so the play is to win the high-value sliver now while it's uncontested, not to wait until everyone's optimizing for the model's one-paragraph answer.
7 Adobe and LinkedIn launched a global AI-training program for marketers
Adobe and LinkedIn rolled out "AI Essentials for Marketers," a global training program of role-based courses across marketing functions, aimed at building AI literacy as demand for those skills climbs. It's a structured, function-by-function curriculum — not a one-off webinar — positioned to certify marketers on applying AI inside their actual jobs.
When the two companies that own much of the marketing stack and the professional network jointly decide to credential AI skills, that's a signal about where job requirements are heading. Read it as the market pricing AI fluency into the baseline of what a marketer is expected to do. The smart move for teams isn't to treat this as HR box-checking; it's to standardize a shared vocabulary so that "use AI for this" stops meaning ten different things across your org. Upskilling the humans is the other half of automating the work.
8 Yum! Brands sold Pizza Hut for $2.7B — splitting a global brand in two
Yum! Brands agreed to sell Pizza Hut in a $2.7 billion deal: the brand outside Mainland China goes to private-equity firm LongRange Capital, while Pizza Hut in Mainland China goes to Yum China Holdings. The move carves a legacy global QSR brand into two separately owned operations and lets Yum concentrate on its faster-growing franchises.
Even the restructuring of a pizza chain is a marketing story now. Splitting a brand across two owners means two divergent marketing strategies, two creative directions, and the real risk of brand drift between markets — the logo stays the same while the playbook forks. For anyone managing a global brand, it's a live case study in how financial engineering quietly reshapes brand consistency. And it's a reminder that while AI dominates the headlines, the biggest dollar figures this week still moved in old-fashioned M&A.
9 Roblox rolled out age-based Kids and Select accounts worldwide
Roblox began globally rolling out new age-based account categories — Kids and Select — introducing age-linked content access, communication settings, and parental controls for users under 16. The change hard-codes age tiers into the platform rather than leaving safety to optional settings.
Roblox is one of the largest youth-reach environments in marketing, and formalizing age gates changes the rules of engagement for any brand trying to reach younger audiences there. Expect tighter constraints on what can be shown and to whom — and treat this as the template, not the exception, as platforms pre-empt regulators on youth safety. If your brand activations live where teens do, the era of "reach everyone on the platform" is ending; brand-safe youth marketing now means designing inside the platform's age tiers from the start.
10 Short-drama app installs jumped 155% — and the top five are eating 90% of the spend
Paid installs for short-drama apps rose 155% globally, with the top five players capturing roughly 90% of user-acquisition spend, according to AppsFlyer. The format — bingeable, vertical, micro-episode mobile series — is pulling a surge of performance budget, concentrated heavily among a handful of dominant apps, with the Indian Subcontinent driving nearly half of global paid-install growth for subscription apps.
Short drama is the fastest-rising new UA channel most Western marketers still aren't watching, and the 90%-to-five concentration is the warning label: the winners are locking in scale and CPMs before the field notices. For performance marketers hunting arbitrage, this is where attention is repricing in real time — and where an entertainment format, not a social platform, is becoming the ad inventory. The throughline of the whole week holds here too: the machine runs the buying, so the edge is spotting where human attention is moving before the algorithm fully prices it in.