April 17th-April 23rd // Estimated Reading Time: 12 minutes
In This Edition!
AI is (finally) getting good at design
Two AI agents walk into a bar…
Introducing The AI Literacy Alliance Explainer Series!
As Deal IDs multiply, AI agents are starting to shoulder some of the work.
In our very first AI Literacy Alliance explainer, Andy Sharkey at Optable breaks down how agent-based systems are emerging to automate increasingly complex programmatic deal workflows.
What Is The AI Literacy Alliance?
An alliance of AI platforms and innovators coming together to help marketing & ads professionals better understand and apply AI.
→ See if your company should join: uof.digital/alliance
Top Headlines🔥
Anthropic debuts Claude Design for building marketing assets, decks, and UIs 🔒
Source: Adweek
April 17th, 2026
Summary: Anthropic's new Claude Design tool lets anyone create marketing assets, pitch decks, prototypes, and web experiences through natural language prompts. Users can upload brand files and Claude Design applies those standards to every project. Canva, Datadog, and Brilliant are part of Claude Design's research beta. Datadog says creating working prototypes used to take a week of back-and-forth but can now happen during a single conversation with Claude Design. The tool is available to Claude Pro, Max, Team, and Enterprise subscribers.
Opinion: Claude Design is the latest component of what looks increasingly like a comprehensive AI productivity suite. Each new tool in Claude works nicely with the others and makes the whole thing harder to leave. It's the Google Workspace playbook, rebuilt for the AI era.
The Trade Desk launches first AI agents with Stagwell as client 🔒
Source: Ad Age
April 21st, 2026
Summary: The Trade Desk launched Koa Agents, an AI suite that helps media buyers plan and run ad campaigns with less manual work. Stagwell is the first agency to use it. The two companies connected their AI agents so Koa can pull in Stagwell's client data and use it to build media plans. Buyers decide how much to hand off to the agents, from research and audience targeting all the way to actually executing the buy.
Opinion: Media buying is becoming a task you delegate to an agent. It started with the walled gardens launching these agentic tools (e.g. Google PMax, Meta Advantage+). Seems like every major DSP is heading this way now, too. It begs the question: What value does the DSP offer anymore if an agent is doing all the work?
A leaked StackAdapt pitch deck sheds light on how ChatGPT ads work 🔒
Source: Adweek
April 20th, 20206
Summary: More details on the mechanics behind ChatGPT ads have emerged via their partnership with independent DSP StackAdapt. Pricing fluctuates based on how many advertisers are competing for the same user prompt, meaning popular queries can cost up to $60 CPM while less competitive niche placements may cost $15 CPM (OpenAI is now also offering CPC pricing and basic measurement 🔒). The minimum spend to join the pilot has dropped to $50K, down from as high as $200K. StackAdapt is positioning the placements as a mid-funnel opportunity, reaching users while they're actively researching and comparing products.
Opinion: Everyone was losing their minds a couple of months ago about $60 CPMs, no performance measurement, no self-serve platform. “ChatGPT Ads will fail!” they said. Chill out. We’re just getting started.
New Products & Features 🚀
What It Does: GPT-5.5 is OpenAI's latest model, stronger than GPT-5.4 at coding, knowledge work, and computer use, and it can take on complex, multi-part tasks with less human direction. It also uses fewer tokens to complete the same tasks, making it more efficient despite being more capable. OpenAI positions it as a step toward a "super app" that combines ChatGPT, Codex, and an AI browser into one unified service. Available now to Plus, Pro, Business, and Enterprise users. Also from OpenAI this week: Workspace Agents let teams build shared AI agents that run in the background on a schedule or inside Slack, handling multi-step workflows like lead outreach, weekly reporting, and vendor risk screening. Workspace Agents are the enterprise successor to Custom GPTs.
Quick Take: Every company’s “AI bill” is getting bigger and bigger. GPT-5.5 is smarter and cheaper to run than its predecessor. That combination matters for marketers building AI into workflows at scale. Hopefully, the numbers will continue to look better: OpenAI has released four models in five months, and its chief scientist said the last two years were "surprisingly slow."
What It Does: Google launched a bunch of products this week at Google Cloud Next. The new Gemini Enterprise Agent Platform lets businesses build, deploy, and manage AI agents at scale across their organization. Tools like Agent Designer and Skills let teams automate repetitive tasks, while a new Workspace Intelligence layer connects Gmail, Docs, Slides, and Sheets so agents can draft content and surface insights without switching apps. WPP says it's already releasing an AI-led campaign every four days using Gemini, and Virgin Voyages cut campaign creation time by 40% with more than 50 agents built on the platform. Also new: 3 safety features in Ads Advisor, an AI agent inside Google Ads, and “auto browse" for Chrome for Workspace users.
Quick Take: Gemini is trying to fend off Claude for enterprise work. If your team already lives in Google Docs and Gmail, agents that work across those applications without requiring new tools or workflows are a much easier sell than standalone AI products.
What It Does: Adobe launched several products at Adobe Summit 2026. Adobe CX Enterprise automates digital marketing workflows, from content creation to customer engagement and campaign execution. Adobe also released a dedicated "Coworker" agent that can coordinate other AI agents, pull business data, build a marketing plan, and execute it. Adobe is also partnering with 30+ companies, including OpenAI, Anthropic, and Microsoft, so customers can run agents across platforms.
Quick Take: Adobe's stock is down 30% this year on fears that AI-native tools will eat its lunch. CX Enterprise is a direct response to that. The bet is that enterprise customers won't abandon a platform that already holds years of their data, brand rules, and campaign history in favor of starting over.
What It Does: AI Max for Search campaigns, available in open pilot in May, uses AI to expand query matching, personalize ad copy, and route users to the most relevant landing page automatically. Microsoft Clarity now shows which of your web pages are cited in AI-generated answers, where competitors are being cited instead of you, and what content gaps to fill. An audience generation tool lets advertisers describe their target customer in plain language and lets AI translate that into targeting settings. Shopify merchants can connect their product catalogs directly to Copilot, with early results showing nearly 90% growth in AI impression share. A new Offer Highlights ad format surfaces relevant product details like free shipping or in-store pickup inside Copilot conversations.
Quick Take: Microsoft shared some crazy stats when announcing all of these new products: "Automated traffic is growing 8x faster than human traffic, AI-driven sessions nearly tripled in 2025 alone, and agentic browser traffic is up roughly 8,000% year over year." That means structured product data and machine-readable content is becoming table stakes for capturing (agentic) demand.
What It Does: Images 2.0 generates better, more precise visuals from text prompts. It handles things that used to break image models, such as small text, icons, UI elements, and dense compositions. In thinking mode, it can search the web, generate up to 10 images from one prompt, and check its own work before delivering. Apparently, it can render text more accurately, which AI tools have been notoriously bad at.
Quick Take: Many AI image tools get you 80% of the way there, and this tool is trying to get you through that last 20%. Readable text in AI-generated images sounds like a small thing, but sometimes it’s the difference between usable output and garbage.
What It Does: Startup CartographAI, launched by digital advertising vets Jay Friedman and Danilo Tauro, rates over 1K mar tech and ad tech vendors across 43 categories using AI and human research. Brands and agencies can filter by category, geography, and company size for free. An AI interface launching soon will let users describe what they’re looking for in natural language. Vendors can't pay to influence their ratings.
Quick Take: Vendor selection can be difficult because many sources of intelligence have a financial relationship with the vendors being rated (*cough* Gartner Magic Quadrant *cough* Forrester Wave *cough*). A neutral, structured way to cut through that could be genuinely useful for marketers drowning in product pitches.
AI Use Case of The Week💡
How Puma used AI to give soccer fans creative control 🔒
The Setup: Puma wanted to build engagement around its partnership with the Manchester City soccer team. Puma decided to use AI to enable fans to design their own Manchester City jerseys.
The AI Solution: The company built the Puma AI Creator, letting fans type in any prompt and generate a custom Manchester City jersey design. Puma worked with Google and Nvidia to reduce the time it took to generate the initial design from 2-5 minutes to 30 seconds. A Tinder-style voting system let users rate each other's designs while waiting. One winning design will become Manchester City's official third kit for the season, making it the first AI-designed jersey to appear on a Premier League pitch.
The Results:
• 54K users and 180K jerseys created in 10 days
• Participants from 206 countries and territories
• Participants voted 1.7 million times, with the average user voting 30+ times
• Average session time of 8 minutes in the follow-up Olympique de Marseille activation
Why It Matters: Brands have handed AI to their internal teams — to write faster, produce more, cut costs. Puma flipped that. They handed the AI to fans. And something interesting happened: 54,000 people in 206 countries became jersey designers!
That's the magic of AI. A teenager in Jakarta could type a prompt and see a Manchester City kit materialize in 30 seconds. That feeling — I made something real — is what drove 1.7 million votes and 8-minute sessions. You can't manufacture that kind of fan experience with an AI-generated banner ad.
Most brands are still deploying AI as utility: smarter chatbots, faster search, better recommendations. Useful, sure. But utility doesn't make marketing magic. Giving every fan the ability to be an artist does.
Your Action: Before launching your next campaign, ask whether AI can be a tool you give to your audience to unlock excitement and creativity, rather than a tool you use on them. Give your preferred AI tool a prompt like this: "Give me 10 creative ways fans or customers could use AI to co-create a [product, design, experience] with our brand, where the output gets made or used in the real world."
Other Notable Headlines📌
Why Mondelez is hiring a global lead to solve for AI-driven shopping bots 🔒 - The Oreo and Cadbury parent company needs to build a global playbook for a world where AI agents handle discovery, decision-making, and transactions on behalf of shoppers.
Meta’s AI push has made its way into ad creative. Not all marketers are happy about it - Marketers and agencies say Meta's AI tools are altering ad creative without their consent, including changing image backgrounds, converting static ads to video, and sometimes generating entirely new visuals.
WPP integrates Google Earth AI into its marketing platform - The holding company is combining Google's satellite and geospatial data with its own marketing intelligence to help clients understand how physical-world factors like weather, foot traffic, and neighborhood conditions influence consumer behavior.
AI-driven traffic declines are fueling a return to live events - Live events offer direct audience access that bypass algorithmic gatekeepers entirely. News publisher Semafor generated roughly $40M in profits last year, with half coming from events.
SpaceX strikes a deal with Cursor, an AI coding tool - SpaceX obtained the right to acquire Cursor for $60 billion later this year, positioning xAI to compete with OpenAI's Codex and Anthropic's Claude in the coding AI race.
EDO is giving ad agencies free TV intelligence data to undercut iSpot 🔒 - The soon-to-launch AI-powered chatbot ChatEDO will offer agencies free access to data on 375M TV ad airings, with the real play being an upsell to EDO's paid outcomes measurement.
Meta employees are up in arms over a mandatory program to train AI on their mouse movements and keystrokes 🔒 - Meta says the tech will help AI models understand how employees actually do their work. Employees aren't into it.
That’s It For This Week 👋
The U of Digital AI Edge Newsletter is intended for AI Edge Pro subscribers, but occasional forwarding is okay!
To subscribe visit Uof.Digital/Newsletters or AI Edge Pro, or contact us directly for group subscriptions.
And remember, U of Digital helps teams drive better outcomes through structured education on critical topics in marketing / advertising like AI, ad tech, mar tech, privacy / identity, CTV, commerce media, and more. Interested in learning more about how we can supercharge your team? Let’s talk!
Thanks for reading!
