May 8th-May 14th // Estimated Reading Time: 8 minutes

In This Edition!

  • Rufus, out! Alexa, in.

  • AI ate the upfront

What Human-In-The-Loop Really Means…

Adam Epstein, Co-Founder and CEO of agentic buying platform Gigi, writes in the latest U of Digital AI Literacy Alliance blog post about how the goal isn’t to review every AI decision. It’s to design systems that surface the right moments for human judgment.

That might look like:
→ Letting AI optimize bids continuously
But flagging any major shift (ex: ±50%) for review
→ Letting AI manage campaign pacing in real time
But routing material budget changes to a human approval layer

Not providing constant oversight. Just smart guardrails.

Check it out for yourself! Where “Human in the loop” Safeguards Matter in Agentic Ad Workflows.

Top Headlines🔥

Summary: Amazon is folding its AI shopping assistant Rufus into Alexa for a new experience designed for cross-device continuity. Built into the Amazon app, Amazon.com, and Echo devices, Alexa for Shopping lets you start a conversation on Echo and pick up seamlessly in the app without having to repeat yourself. New agentic features let Alexa automatically buy a product once it hits a target price or add items to a cart for customer review before checkout. Rufus had more than 300M users in 2025 and ran ads. All of it carries over into Alexa for Shopping.

Opinion: Alexa turned out to be a real backstabber. Rufus, you will be missed.

Every bit of information Amazon collects from users from their conversations with Alexa, across multiple devices, feeds Amazon's understanding of what shoppers want, when they want it, and on what terms. Amazon’s building the ultimate demand prediction engine with a built-in buy button. The better Amazon gets at anticipating purchase intent, the more influence it can give to marketers at the moment a decision gets made. The more influence it can give to marketers, the more it can charge for that influence. Feed the Amazon beast…

AI is now part of the upfront pitch
Source: Multiple
May 11th-13th

Summary: During the upfront this week, Amazon announced Dynamic TV Creative, an AI-powered tool that personalizes interactive ads on Prime Video based on a viewer's shopping behavior and prior ad exposure. Netflix🔒 unveiled AI agents that can autonomously manage, optimize, and purchase ads on the platform, alongside a tool that adapts existing creative to different formats without rebuilding from scratch. YouTube🔒 introduced an AI tool that custom-builds ad inventory around cultural events advertisers request, like a specific artist's tour. Warner Bros. Discovery🔒 will let brands use AI agents to deliver ads based on real-time audience interest signals, paired with scene-level contextual targeting. Fox uses AI to analyze every second of videos in real time, extracting topic, talent, mood, and context to match ads to content.

Opinion: You can now buy premium CTV inventory, like Netflix and Disney, through numerous different channels and platforms. And the amount of CTV ad supply is growing as user behavior shifts more towards streaming and streamers push their users more towards ad-supported models. As CTV inventory grows and proliferates, it increasingly becomes a commodity, and data (along with AI’s ability to maximize the value of that data) moves center stage. That's a data + tech arms race Amazon and Google are better positioned to win than Fox or WBD.

Here’s our view of the industry’s new “pecking order”:

New Products & Features 🚀

What It Does: Claude for Small Business connects Claude to tools small business owners already use, including HubSpot, Canva, Google Workspace, Microsoft 365, and QuickBooks. It ships with 15 ready-to-run workflows covering marketing, sales, HR, customer service, finance, and more. Owners approve before anything sends, posts, or pays.

Quick Take: Small businesses have historically been underserved by enterprise tech platforms. Anthropic is running hard at AI productivity for enterprises, and it’s been doing a great job. The jump to small businesses makes a ton of sense as an area for expansion.

What It Does: Built with Gemini, Google Ads' new tool lets advertisers visualize, analyze, and export campaign data without handling any raw data. You can type a prompt directly into the dashboard, like "breakdown my spend by campaign type," and get real-time updates. It’s currently in beta for English-language accounts. More details coming on May 20 at Google Marketing Live.

Quick Take: A prompt-based dashboard is convenient! But it will also train marketers to ask Google questions instead of pulling raw data and manipulating it themselves. Good for Google. 

What It Does: Twilio's new conversation layer is designed to fix a persistent customer service problem: customers having to repeat themselves every time they switch channels or get transferred to a new agent. Three new components work together to fix this. Conversation Memory builds a persistent customer profile that both AI and human agents can pick up mid-thread. Conversation Orchestrator manages handoffs across channels. Conversation Intelligence analyzes live interactions in real time for signals like sentiment and escalation risk. The platform is model-agnostic, so brands can plug in whatever AI they're already using.

Quick Take: When customer service has no memory of your interaction and history with the brand, it’s frustrating af. We’ve all experienced this frustration. This can create a lot of customer churn. AI memory to the rescue!

The Setup: Wix CMO Omer Shai wanted to move his marketing department faster, from idea to production, without the feedback loops that slow traditional teams.

The AI Solution: Shai started with himself, building a personal super agent on Base44 connected to HubSpot, his email, and his calendar. It briefs him before every meeting, monitors leads, sends daily pipeline health reports, and regularly does social listening. From there, he's using AI to rebuild his teams, restructuring around "full-stack marketers" who own projects from end to end.

The Results:

  • AI tools help Shai get to 70-80% of any idea before he involves his team, making briefs and creative direction faster and clearer.

  • Designers no longer wait on content writers, and vice versa. The feedback loop is shrinking.

  • Traditional job titles like "content writer" and "product marketer" are being phased out, replaced by full-stack marketers who can own projects from start to finish. 

Why This Matters: The traditional marketing org was built around scarcity of design skills, writing skills, and technical skills. AI makes all those skills abundant. The new marketing org, in which everyone can do everything, should be built aorund scarcity of good judgement.

Your Action: Are full-stack marketers realistic for your organization? Identify which specialist skills on your team, whether that's copywriting, design, or data analysis, would create the most value if they were embedded across every stage of a project rather than brought in at a single point. Build AI to replicate those skills first and try implementing them across your team.

Other Notable Headlines📌

​​OpenAI adds product feed ads to ChatGPT - Retailers can now connect their product catalogs to ChatGPT to automatically generate product-specific ads at scale, mirroring how shopping campaigns work on Google.

TikTok will let advertisers bring their own third-party AI agents to its ad platform - Marketers can use third-party AI agents to run their TikTok campaigns autonomously. The company also launched TikTok Go, an in-app travel booking experience, with Expedia as an early partner.

Anthropic surpasses OpenAI in US business adoption - About 34% of US businesses now pay for Claude, versus 32.3% for ChatGPT, according to Ramp's May AI Index. Anthropic quadrupled its business adoption over the past year.

Meta launches an AI chat that even Meta can't read - Incognito Chat uses end-to-end encryption and does not store server logs. Messages disappear when you leave the session. It will be rolled out in WhatsApp and the Meta AI app in the coming months.

Google adds inline links and subscription labels to AI search responses - Links now appear within AI responses next to relevant text, and subscription sources get labeled as such. Google says labeled links get significantly more clicks, which has clear implications for branded content and publisher partnerships.

Google is internally testing Remy, a 24/7 personal AI agent for Gemini - Remy is designed to take actions on behalf of a user across Google services, learn preferences over time, and handle complex tasks autonomously. No public release date has been announced.

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, mar tech, programmatic, CTV, commerce media, and more. Interested in learning more about how we can supercharge your team? Let’s talk!

Thanks for reading!

Keep Reading