As brands move beyond AI pilots, many are relying on forward-deployed engineers—technical teams that work inside organizations to redesign workflows and build custom AI tools rather than simply installing software.
The approach, popularized by Palantir and now embraced by OpenAI, Anthropic, Microsoft, Amazon and others, reflects a broader shift: enterprise AI increasingly depends on people who understand both the technology and the business processes it is meant to transform.
Ad Age’s Asa Hiken covers a recent example:
“A beauty brand that had been gathering weekly to discuss its media strategy was able to save hours of subjective hair-splitting by turning to an AI-powered system for data-backed takeaways on what was working and what wasn’t. The idea for such a tool did not come from anyone who worked at the brand, but rather from a forward-deployed engineer at agency PMG who was observing the team to find ways that AI could help them work better and faster.”
Read: “How brands get more from AI with forward-deployed engineers” (July 14) – Ad Age (subscription)
Earlier this month, Microsoft announced a forward-deployed engineer strategy of its own with a $2.5B investment in Microsoft Frontier Company, which plans to embed 6,000 industry and engineering experts with customers to accelerate their AI transformation. (tipsheet, July 6)
From tipsheet: Forward-deployed engineers move AI vendors and agencies closer to the decision layer. By redesigning workflows and embedding AI into day-to-day operations, they help shape how business decisions are made—not just how software is deployed. The closer companies get to those decisions, the more durable their competitive position becomes.
Related: “InstaLILY Raises $60 Million Series B and Launches Lily, the World’s First AI Forward Deployed Engineer” (July 14) – press release
SEARCH
Google Images becomes a discovery surface
Google is redesigning Google Images with a personalized homepage that recommends images based on users’ interests rather than waiting for a search query. The update also brings image generation into AI Overviews for certain visual requests.
Google Senior Engineering Director, Search, Brad Kellett writes:
“We’re bringing image generation directly into AI Overviews in Search. Using our latest Nano Banana model, this update transforms a simple text prompt into a high-quality, custom visual made completely from scratch, seamlessly bridging the gap between imagination and reality. Image generation in AI Overviews will start to roll out over the coming weeks in English, for all regions that currently support image creation in AI Mode.”
Read: “Celebrating 25 years of visual search innovation” (July 14) – Google
From tipsheet: Google is extending AI-driven discovery beyond Search into Images. Instead of simply retrieving visuals that match a query, the product begins recommending content before users ask, giving Google’s models a greater role in deciding what users discover.
CONNECTED TV
Measurement moves to the infrastructure layer
Following last week’s launch of his company’s CTV measurement partner program, tvScientific by Pinterest CEO Jason Fairchild argues that AI is changing measurement from a reporting function into part of the decision-making infrastructure:
-
“This is a reflection of the measurement frameworks feeding AI. As more decisions become automated, measurement moves from the reporting layer to the infrastructure layer. It not only tells us what happened after the fact, but it actively shapes how budgets are allocated and how future performance is defined. The quality of those decisions will depend on the quality and diversity of the signals behind them.”
Read: “The Era of the Single-Source Truth Is Over” (July 14) – Jason Fairchild, CEO, tvScientific by Pinterest on the company’s blog
More:
-
tvScientific by Pinterest Launches Certified Measurement Partner Program (July 9) – press release
The program launched with four certified measurement partners. tvScientific’s broader partner network lists 33 organizations.
From tipsheet: Infrastructure is the layer other systems depend on to operate. Fairchild argues measurement is becoming part of that layer, feeding AI with the signals it uses to allocate budgets, optimize campaigns and define success—instead of simply reporting results after the fact.
SELL-SIDE
Compressing the vendor stack
Publisher monetization platform Raptive launched Apex yesterday, a “platform-plus-partnership” offering for larger media companies. The company says Apex combines monetization technology, strategic services and network-scale intelligence into a single offering designed to reduce operational complexity while improving revenue performance.
Chief Business Officer Tina Pautz wrote on LinkedIn: “We built Apex to help media companies improve revenue performance while reducing operational complexity by combining technology, strategic partnership, and the scale of the Raptive network into a single solution.”
Read more. (July 14)
Raptive says Apex is designed around three outcomes: stronger revenue performance, lower operational complexity and network-scale intelligence.
Read: “Raptive Launches Apex as Media Companies Rethink the Future of Ad Monetization” (July 14) – press release
From tipsheet: As AI compresses software categories, publisher platforms are becoming operating systems for revenue. Competitive advantage shifts from individual products to the operating layer, where intelligence coordinates the revenue system.
Related: “Most sales houses talk about transparency. But how much do they really stand behind this?” (July 14) – Jayson Dubin, CEO, Playwire on LinkedIn, after announcing the company had removed all reseller line items from its ads.txt.
LLMs & CHATBOTS
Developments
- “A proposal for a US-based Standards Body for ‘Frontier-class’ AI, modeled after FINRA; labs would share models for review up to 30 days before release” (July 14) – Demis Hassabis, CEO, Google DeepMind on X
- Claude for teachers (July 14) – Anthropic
- Opinion: “You Just Hired a Million Bad Employees” (July 14) – Hebbia CEO George Sivulka on the a16z blog
MEDIA
Publisher bets against scale
Semafor is expanding its video business while “disposing of scale-based thinking” in favor of programming for C-suite executives, arguing that “not all clicks are equal,” CEO Justin Smith tells Adweek.
Consultant Gabriel Dorosz says, “I’m not sure any publisher can be in the scale business anymore. Semafor is not about how many, it’s about who.”
Semafor monetizes its video through long-term sponsorships that extend across its newsletters, events and shows. More than half of the company’s revenue already comes from events.
Read: “Semafor Is Betting Big on ‘Anti-Scale’ Video” (July 14) – Adweek
Semafor has hired Adam Banicki, the former general manager of video at Fortune Media, as its first head of video. He will own both the commercial and editorial sides of the business.
From tipsheet: As AI erodes the economics of scale, some publishers are reorganizing around influence rather than impressions. Unlike display advertising, integrated sponsorships spanning video, newsletters and events are likely to be evaluated across multiple objectives, making measurement both more important and more complex. Success may be judged by outcomes ranging from brand perception and executive engagement to pipeline and revenue.
Related: “Inside the newsroom push to turn print reporters into video talent” (July 14) – Digiday (subscription)
CONNECTED TV
Netflix expands the recommendation engine
Writing on Mobile Dev Memo, analyst Eric Seufert argues that Netflix’s reported interest in live channels, creator content and short-form video reflects a broader strategy than simply adding programming. The goal is to expand the pool of programming available to its recommendation system, giving its AI more opportunities to match viewers with programming they’ll want to watch.
Seufert writes:
“But it would be an oversimplification to describe this merely as a catalog augmentation strategy; it’s a RecSys expansion strategy, in the same vein as Meta’s fundamental transition toward an open graph paired with short-form video… Simply onboarding more content won’t, by default, improve engagement if it’s not easily discoverable or optimally surfaced.”
Read: “Netflix’s engagement problem is a personalization opportunity” (July 14) – Eric Seufert on Mobile Dev Memo (subscription)
More: “Introducing Exciting New Ways to Find and Enjoy Your Next Favorite on Mobile” (April 30) – Netflix
TECH
PubMatic refocuses OpenWrap
Playwire has been selected as the recommended monetization partner for publishers transitioning from PubMatic OpenWrap following a vendor evaluation process.
The move suggests PubMatic may be redefining OpenWrap’s role. While Playwire is becoming the recommended destination for publishers migrating off the web wrapper, PubMatic signaled its investment in the OpenWrap SDK for mobile buyers when it announced in April support for Unity LevelPlay alongside Google AdMob and AppLovin MAX.
Read:
- “Playwire Named Recommended Monetization Partner for Publishers Transitioning from PubMatic OpenWrap Web” (July 13) – Playwire
- “Where Mobile Scale Meets Mediation Power: Unity LevelPlay Now Available with PubMatic” (April 9) – PubMatic
PEOPLE MOVES
Now hiring
- Former Verizon executive Rachel Chan joins adMarketplace as CMO (July 14) – Rachel Chan on LinkedIn
- Former Zefr marketing executive joins CTV platform Strategus as VP of Marketing (July 14) – Elisa Crutchfield on LinkedIn
MORE
- This New Training Framework Gives Publishers A Say In How AI Uses Their Work (July 14) – AdExchanger
- Business Outcomes Are Just the Beginning, According to IAB Digital Video Ad Spend & Strategy Full Report (July 14) – IAB
- “More than 80% of AI ad spending in 2026 will still be traditional search ads shown alongside AI-generated results,” according to eMarketer (July 14) – Forbes