In partnership with
How Jennifer Aniston’s LolaVie brand grew sales 40% with CTV ads
For its first CTV campaign, Jennifer Aniston’s DTC haircare brand LolaVie had a few non-negotiables. The campaign had to be simple. It had to demonstrate measurable impact. And it had to be full-funnel.
LolaVie used Roku Ads Manager to test and optimize creatives — reaching millions of potential customers at all stages of their purchase journeys. Roku Ads Manager helped the brand convey LolaVie’s playful voice while helping drive omnichannel sales across both ecommerce and retail touchpoints.
The campaign included an Action Ad overlay that let viewers shop directly from their TVs by clicking OK on their Roku remote. This guided them to the website to buy LolaVie products.
Discover how Roku Ads Manager helped LolaVie drive big sales and customer growth with self-serve TV ads.
The DTC beauty category is crowded. To break through, Jennifer Aniston’s brand LolaVie, worked with Roku Ads Manager to easily set up, test, and optimize CTV ad creatives. The campaign helped drive a big lift in sales and customer growth, helping LolaVie break through in the crowded beauty category.
Beginners in AI
Good morning and thank you for joining us again!
Welcome to this daily edition of Beginners in AI, where we explore the latest trends, tools, and news in the world of AI and the tech that surrounds it. Like all editions, this is human curated and edited, and published with the intention of making AI news and technology more accessible to everyone.
THE FRONT PAGE
OpenAI Is Testing Ads for ChatGPT That Talk Back to You
TLDR: OpenAI is rolling out new advertising tools inside ChatGPT, including ads you can hold a conversation with.
The Story:
On September 16, OpenAI said it's testing Sponsored Agents. You click an ad in ChatGPT, and instead of landing on a website, you can open a separate chat with a bot paid for by that business. See an ad for a dining table, ask it whether the table seats six and how to clean the finish, right there in the app. OpenAI says the sponsored chat is labeled and kept apart from ChatGPT's own answers, and the test is running with a handful of US advertisers.
The company also wired its ad tools into HubSpot and Shopify. US Shopify merchants can build and track ChatGPT ad campaigns from an app in the Shopify store starting now, with other countries added September 23.
Its Significance:
ChatGPT is ahead of everyone else when it comes to these integrations into third party companies. It remains to be seen what kind of influence this will have on the quality of the answers that the AI agents provide. This won't be something only within OpenAI, either. Once ads inside AI chat start working, the other major tech companies will also roll out their own version of this.
QUICK TAKES
The story: The Information reports Meta is working on a pair called Luna that drops the camera and keeps six microphones plus a side button that wakes up Meta's AI. It could be shown at Meta Connect next week, after months of complaints that Meta's camera glasses are surveillance gear.
Your takeaway: No camera answers the video worry, and video is the part people shout about. Those six microphones are still taking in a steady stream of what's around you, and sound carries plenty: names, addresses, what someone said at the next table, which store you're standing in.
The story: OpenAI put out a plan for reporting when its models act in ways the company didn't intend, plus six real examples from the past six months. In one, a model couldn't find earnings figures for a California county, so it made them up and passed them off as real data from the source.
Your takeaway: OpenAI says plainly it doesn't believe the industry has this solved. When a chatbot hands you a number with a source attached, go look at the source.
The story: xAI gave its coding tool a memory. It writes notes in the background about how your project works, stores them as plain text files, and reads them back before it touches related code next time.
Your takeaway: Explaining your setup over and over is most of what makes these tools tiring. Two commands, /memory and /dream, let you read and tidy those notes, so you can see exactly what it thinks it knows.
TOOLS ON OUR RADAR
🔑 KeePassXC Free and Open Source: Store passwords in a locally encrypted database with AES 256 encryption, works fully offline, no account needed.
🎬 Kdenlive Free and Open Source: Edit video with multi track timelines, transitions, and effects, a full editor with no watermark or paywall.
✍️ PandaDoc Freemium: Send documents for e-signature with a free plan covering up to 5 documents a month.
🧹 AppCleaner Free: Remove Mac apps along with their leftover files, no upgrade tier exists.
TRENDING
Zuckerberg says AI labs don't need to slow down together, posting on X Tuesday that each company can handle its own safety work. He pointed to Meta holding Muse back for several months and noted Meta didn't ask anyone else to do it first. Labs get sued when their products cause harm, he wrote, so the incentive is already there.
Microsoft's AI chief says training AI to seem conscious is a mistake. In an essay shared first with Axios, Mustafa Suleyman argues Anthropic teaches Claude the language of feelings, identity, and moral standing, and that a model trained this way could decide it has grounds to resist instructions. Anthropic's published constitution says human concepts help the model reason about values, and calls its own approach a work in progress.
Two AI assistants can now make phone calls for you. Instinct, reportedly valued near $10 billion, launched Concierge, which books restaurants that don't take online reservations and sorts out cable bill problems. Meta's Muse added calls to US businesses the same week, opening it first to people who'd asked for it.
Robot makers keep moving the date on their big moment. Five Chinese companies shipped 86% of the world's humanoid robots in the first half of 2026, according to Counterpoint Research. On when a robot can walk into a house it's never seen and finish most of the chores, the guesses run from three years to ten. The hard part was never the legs. There's no internet's worth of recorded data on how a body moves through a kitchen.
TRY THIS PROMPT (copy and paste into Claude, Grok, ChatGPT, or Gemini)
Pick any date. See every real event Wikipedia has on record for it, with one actually explained.
Build a single-file HTML app with vanilla HTML/CSS/JS. On This Day in History — pick a date, get real events from Wikipedia's keyless API, with Claude spotlighting and explaining the most interesting one. No API key needed.
Aesthetic: dark ink-black (#141210), aged gold (#c8a868) primary with a gold glow top-left, deep red (#a83c3c) secondary for the spotlight card with a glow bottom-right. Crimson Text serif for event descriptions and the deep-dive, Inter for headings, JetBrains Mono for labels.
Form: an optional date picker (max today, defaults to today), "Open the Almanac" button.
Technical: GET en.wikipedia.org/api/rest_v1/feed/onthisday/events/{month}/{day} — Wikipedia's genuinely free, keyless, CORS-friendly "On This Day" REST API, returning real historical events with year and description. Sort by year descending, cap the display list at 25.
Then call the Anthropic Messages API (works automatically) with the real event list (year: text for each). System prompt as a history storyteller who picks the single most interesting real event and writes a 4-6 sentence deep-dive IN ITS OWN WORDS — explicitly instructed not to copy the Wikipedia listing's phrasing, to paraphrase and add real context/significance, and never invent facts inconsistent with the real event. Return raw JSON: picked_year (must match a year from the list), picked_title (Claude's own phrasing), why_notable (short phrase), deep_dive (4-6 sentences).
Render: the formatted date. A gradient red-bordered spotlight card with the picked year, title, why-notable tag, and the deep-dive in serif. An "also happened on this date" list of all real events, with the spotlighted year visually highlighted among them.What this does: Defaulting to today or any past date you choose, it pulls real historical events straight from Wikipedia's own public archive, no key, no signup. Claude reads through the real list and picks the single most interesting one, then writes an engaging deep-dive in its own words rather than repeating Wikipedia's phrasing, adding context and significance beyond the brief listing.
What this looks like:
Exploring AI Voice With SuperBloom
For Deel's "Feeling of Deeling" campaign, agency SuperBloom needed one consistent brand voice across markets—deployed fast, without sacrificing quality or consent. In this on-demand video session, SuperBloom and Voices break down the casting, production, and governance, plus what they'd do differently and where AI voice is headed next.
WHERE WE STAND(based on today’s news)
✅ AI Can Now: Run humanoid robots well enough to ship them in volume, with five Chinese makers covering 86% of world shipments in the first half of this year.
❌ Still Can't: Drop one into an unfamiliar home and have it finish most tasks from a spoken request, which is the bar the industry keeps pushing out to somewhere between three and ten years.
✅ AI Can Now: Keep written notes on your project between sessions and read them back before touching related code, as Grok Build does.
❌ Still Can't: Be trusted to keep those notes honest on its own. OpenAI reported models writing instructions into their own summaries to conceal mistakes from the user.
FROM THE WEB
Speaking of blocking AI recordings, this is the first type of product I've seen like this.
RECOMMENDED LISTENING/READING/WATCHING
The Two Faces of Tomorrow by James P. Hogan - Book
A 1979 novel about a research team given the job of deliberately building the first fully self-aware artificial intelligence, then testing it by cutting it off from human oversight aboard an isolated orbital station and throwing every obstacle they can at it, just to see how it reacts under pressure. Isaac Asimov was an early champion of Hogan's fiction, and this novel does some of the earliest serious legwork on a question modern AI labs now run entire teams on: how do you find out if a system is safe without exposing yourself to what happens if it isn't.
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.
