It's time to set up your holiday streaming campaign
Are you ready to reach holiday shoppers on CTV? For many streamers, the season is already underway.
Last year, according to Roku internal data, holiday content searches on Roku saw an 86% increase during October. And in November, those searches rise an average of 39% weekly. This surging holiday spirit is mirrored by consumer activity, as about half of Roku streamers shopped Black Friday and Cyber Monday deals in 2025.
Meeting those shoppers is easier than you might think. Begin with the seasonal outcome you hope to drive: awareness, online sales, leads, store visits, or another measurable goal. From there, define your audience, upload your creative, set the budget and campaign dates, and establish measurement before launching. You can even create a shoppable ad unit that lets streamers buy with a click of their remote.
Our holiday playbook offers a month-by-month guide to reaching shoppers during this all-important time of year.
First-time spenders get a $5,000 match on the first $5,000 spent with code BEEHIIV5K. Terms apply.
Beginners in AI
Good morning, and happy Sunday.
This is the weekly catch-up edition. The thread that ran all week was money: AI agents got the power to pay for you, while some stores and airlines started shutting them out. Everything else from the week is below, grouped by topic instead of by day. As always, every story here is human picked and edited.
THE FRONT PAGE
AI Agents Got a Wallet This Week, and Some Stores Locked the Door

TLDR: OpenAI launched ChatGPT Wallet so its agents can pay for what you ask for, the same week United Airlines restricted unapproved AI agents and Visa found shopping agents failed 42.8% of its tests.
The Story:
On Monday, Visa's test of 22 AI shopping agents showed that in more than half of the failed runs, the agent bought something anyway, either the wrong product or before it knew what the shopper needed. The same day, Firmly raised $15 million from backers including J.P. Morgan Payments and Mastercard to help stores sell inside AI assistants. On Tuesday, OpenAI opened a Business Hub where U.S. shops can build an agent that answers shoppers inside ChatGPT.
On Wednesday, United said it's blocking "unauthorized third-party AI agents" from its site, citing exposed logins and purchases people didn't mean to make. On Friday, ChatGPT Wallet arrived: an agent picks the items, you approve with your face or fingerprint, and the store gets a stand-in card number instead of your real one. It's open to a limited group of U.S. Pro subscribers for now, and Google separately let Gemini read your Google Wallet to answer spending questions, without moving any money.
Its Significance:
The parts for an AI that shops for you are now in place: payment networks, stores' own agents and a checkout step that hides your card. What's missing is trust on both sides. Visa's test shows agents still guess when nothing fits, and United's CEO told The Wall Street Journal he doesn't want "another middleman" between the airline and its customers. ChatGPT Wallet has no spending limits yet, so read the review screen every time before you approve. When an agent can't finish a purchase, finish it yourself on the store's own site.
JOIN THE COMMUNITY
Learn AI With Us
Most people try AI once, get a so-so answer, and quit. Our 14-day challenge gets you past that point, one short lesson a day.
Week one, AI starts working for you. It drafts a real email for you, flags which messages in your inbox need a reply, and turns messy notes into next steps. You also learn to catch its mistakes. By day 7 you have a weekly AI routine you'll want to keep.
Week two, you build something of your own. You find a skill people will pay for and turn it into an offer, with a one-page pitch, a short video and your first client emails, all drafted with AI and ready to test. Day 14 hands you a 30-day plan.
10 to 15 minutes a day. No tech background needed. We just released one for older adults, with a gentler progression in the lessons.
QUICK TAKES
AI Did the Hard Work, and Something Else Checked It
The story: A system running an unreleased Anthropic model proved a quantum computing puzzle open since 2008, and the Lean proof checker accepted every step. Google DeepMind's AlphaProof Code wrote over six million lines of machine-checked proof that key parts of the code behind Signal's encryption keep their security promises. Claude helped logician Jan von Plato turn Kurt Gödel's shorthand notebooks into a 650-page guide, and a proof checker confirmed about 400 of the 663 theorems as written. Anthropic says Claude Fable 5.2 designed three gadgets that worked on the first build, and an amateur's 48 AI agents flagged a possible Maya site that Mexico's archaeology institute has no record of.
Your takeaway: Each result leaned on a check outside the AI: Lean, Coq, Diode's engineers, and for the Maya site, a ground survey that hasn't happened yet. If the gadget story made you want to build something yourself, our list of 12 things you can make with Claude starts with no-code projects like a room planner and a dog-catches-treats game.
Nvidia Lost Its CEO in One Week
The story: On Monday, Nvidia's board opened a review of reports that Jensen Huang was the target of a foreign spy effort. By Wednesday it was reportedly weighing whether to oust him, and on Thursday Huang resigned after Nvidia lost about $430 billion in value in two days, with CFO Colette Kress as interim CEO. On Friday, a former OpenAI researcher said export-banned Nvidia chips likely trained China's top open models, though TechCrunch notes the training may have happened outside China.
Your takeaway: Nvidia is one of the biggest holdings in most S&P 500 index funds, so this reaches plenty of 401(k)s. Tuesday's earnings report is the first with Kress in charge.
New Models Got Cheaper, and the Money Got Bigger
The story: Anthropic released Claude Fable 5.2 and Mythos 5.2, with Fable priced 20% below OpenAI's GPT-6 Astra, per Fortune. Meta's Muse Ember costs $0.10 per million input tokens, and Google put its fast Gemini 4 Hydrogen into Search for everyone. Anthropic also set its IPO at $70 to $78 a share, with trading expected Nov. 19, and Bloomberg reports OpenAI topped $10 billion in quarterly revenue for the first time while still losing money.
Your takeaway: Cheaper models usually mean cheaper apps, but nearly every score this week came from the company selling the model. In an outside test, the best coding agent fully finished only 26.4% of real app edits, so test anything AI builds for you before you trust it. None of this is investment advice.
TOOL OF THE WEEK
Thirty-eight tools ran this week. This one wins.
💳 Privacy Freemium: Virtual cards locked to one store with spend limits you set, 12 free a month.
It fits this week's lead. ChatGPT Wallet doesn't have spending limits yet, and a Privacy card gives you one today: lock a card to a single store, cap what it can charge, and pause it whenever you want.
Runner-up: 🕵️ Have I Been Pwned Free: Type in your email to see which hacked websites leaked it, with no account needed.
TRENDING
Google's Gemini Glasses Open for Pre-Order, Starting at $395. The Warby Parker and Gentle Monster frames hit stores Dec. 10 in the U.S. and South Korea, and can explain what you're looking at and translate conversations in over 70 languages. Google describes a camera and open-ear speakers, not a screen.
Reddit Turned Its Threads Into AI Reviews for 10,000 Products. Each page sums up what users like and dislike, with a positive, mixed or negative split. Reddit says brands can't pay to be featured, but open a few original threads before you buy.
The FAA Halted Every U.S. Flight Friday, Days After a New AI Tool Went National. Transportation Secretary Sean Duffy pointed to a communication problem across the country's air route traffic control centers. The maker of SMART, the AI routing system, says the ground stop wasn't caused by its tool.
AI Got Blamed for the Fewest Job Cuts of 2026 in October. Employers tied 1,265 October layoffs to AI, under 3% of the total, says Challenger, Gray & Christmas. AI is still behind about 20% of cuts so far this year.
ChatGPT Retires Its Old Voice Modes on Dec. 11. Advanced and Standard Voice go away and everyone moves to Live voice. Old voice chats stay in your history.
Google's Pixel Phones Can Now Build You a Widget From One Sentence. Describe something like "a trip dashboard" and Gemini builds a home screen widget that pulls from Gmail and Calendar. It's rolling out over the next few weeks in select countries, for adults only.
PROMPT OF THE WEEK (copy and paste into Claude, Grok, ChatGPT, or Gemini)
🔎 Owners Say: Type in a product and see what real owners say after living with it
Six prompts ran this week. This one won because you buy something almost every week, and it checks what owners think before an agent or a sale page talks you into it. The runner-up, Who Can Read It?, is the one to run on the apps you trust with private things.
Build a single-file HTML app called Owners Say in vanilla HTML, CSS and JavaScript. No API key needed.
I type a product (like a specific espresso machine) and pick what matters most to me: reliability, easy for a beginner, worth the price, size and noise, or lifespan.
The app calls the Anthropic Messages API with the web_search tool to read owner reviews, forum threads and long-term reviews. It must never invent quotes, numbers or sources, and says plainly when it can't find enough owner discussion.
Show a two-sentence verdict, an estimated positive, mixed and negative bar, what owners like and dislike, an answer about my priority, any warning signs of fake or paid reviews (like a burst of five-star reviews in one week), and real source links.
Style: near-black background, coral and teal accents, Space Grotesk, Newsreader and JetBrains Mono fonts.What this does: Type in something you're thinking of buying. The app searches owner reviews and forum threads, then shows a quick verdict, what people like and dislike, an answer about the thing you care about most, and any signs the praise might be fake, with links to every source.
WHERE WE STAND (based on this week's news)
✅ AI Can Now: Check out at a store for you with a stand-in card number, so neither the agent nor the shop sees your real one.
❌ Still Can't: Reliably stop when nothing fits. In more than half of Visa's failed test runs, the agent bought something anyway.
✅ AI Can Now: Write millions of lines of proof that a computer checks line by line, as DeepMind did for the code behind Signal's encryption.
❌ Still Can't: Decide what's worth proving. People chose the security rules, and a rule written too narrowly can let a bug slip through.
✅ AI Can Now: Scan 1,200 square kilometers of laser maps overnight and flag a possible Maya site that isn't in Mexico's records.
❌ Still Can't: Confirm it. Only a permitted ground check by archaeologists can show people built it.
How was this week's recap? Tap one below, then tell me what you'd change on the next screen. I read every answer.
How was today's edition?
Thank you for reading. We’re all beginners in something. With that in mind, your questions and feedback are always welcome and I read every single email!
-James
By the way, this is the link if you liked the content and want to share with a friend.
Some * designated product links may be affiliate or referral links. As an Amazon Associate, I earn from qualifying purchases. This helps support the newsletter at no extra cost to you and Amazon makes a tiny hair less.




