Blu Dot surpasses 2,000% ROAS with self-serve CTV ads
Home furniture brand Blu Dot blew up on CTV with help from Roku Ads Manager. Here’s how:
After a test campaign reached 211,000 households and achieved 1,010% ROAS, the brand went all in to promote its annual sales event. It removed age and income constraints to expand reach and shifted budget to custom audiences and retargeting, where intent was strongest.
The results speak for themselves. As Blu Dot increased their investment by 10x, ROAS jumped to 2,308% and more page-view conversions surpassed 50,000.
“For CTV campaigns, Roku has been a top performer,” said Claire Folkestad, Paid Media Strategist, Blu Dot. “Comping to our other platforms, we have seen really strong ROAS… and highly efficient CPMs, lower than any other CTV partner we've worked with.”
Using Roku Ads Manager, the campaign moved from a pilot to a permanent performance engine for the brand.
Beginners in AI
Good morning and thank you for joining us again!
Grok just put out the first seven minutes of an AI-made Odyssey, and a few shots look like they came off a real film set. That's a big step for AI video, and it shows you what you can make yourself now, plus where it still falls flat. Today's prompt helps you plan your own short AI film, one shot at a time. As always, every story here is human picked and edited.
THE FRONT PAGE
Grok Made 7 Minutes of an AI Odyssey, and Some Shots Pass for Real Film

TLDR: Elon Musk's Grok released the first seven minutes of its AI-made Odyssey, and even a skeptical critic said some shots could pass for real actors on a real sea.
The Story:
Three months ago, Musk promised a full-length, "historically accurate" AI movie of Homer's Odyssey. On Friday, Grok Imagine, Grok's image and video maker, posted Episode 1: just over seven minutes in 4K, made with Wonder Studios and director Dev Walker. You get black-sailed ships on a gray sea, a hero in a dark cloak and a narrator with a British accent.
TechCrunch's Anthony Ha said it looked better than he expected. Grok hasn't said how many episodes are coming or when. Compare these videos with what was available a few years ago and you'll see just how quickly improvements in this space have been made.
Its Significance:
Seven connected minutes that sometimes look like a movie means a small team can now make what used to need a crew, a set and a big budget. The same week, Grok Imagine posted a step-by-step guide to keeping AI characters looking the same from shot to shot, the trick that makes longer videos hold together. Short films, ads and class projects are within reach for anyone willing to plan. If you've never made an AI clip, our beginner's guide to AI video walks you through your first one in four steps.
The hard part hasn't moved. Story, timing and acting still come from people, and Christopher Nolan's Odyssey, shot for real across six countries, sold out IMAX screens for months. Today's prompt gives you the planning half, so your clips at least stay consistent.
QUICK TAKES
The story: Dream Labs came out of hiding with $90 million and Dream-1, a model that steers two-armed robots through new tasks with no extra training. The company says it worked on 27 robot types in simulation and 4 real ones, and it rents the model out for $3 an hour.
Your takeaway: Robots usually have to be shown each job on their own hardware first, so skipping that step would make cheaper home and shop robots far more useful. It's a research beta that Dream Labs says "will make mistakes," and it handles grippers only, not hands.
The story: The UK's new AI Biosecurity Lab will use AI to spot biological threats and speed up vaccines and treatments. OpenAI and Anthropic are each adding £100 million in credits for their models, and Google DeepMind is lending experts.
Your takeaway: Faster vaccines are one of the clearest wins AI could deliver, and it's rare to see rival labs feeding the same project. The government also says AI creates "new biological risks," which is why this lab works both sides.
The story: London dermatologist Magnus Lynch tested 200 skin spots confirmed by biopsy. A small image model from 2015 named the right condition first 62% of the time, while OpenAI's GPT-6 Astra and Anthropic's Claude Fable 5.1 each got 42%, and both chatbots told people to see a doctor for almost everything.
Your takeaway: An all-purpose chatbot isn't a skin checker. AI built and tested for one medical job does much better, and if a spot worries you, a doctor is still the right call.
TOOLS ON OUR RADAR
🎞️ Shotcut Free and Open Source: Edits video in 4K with color tools, a free Premiere swap that can join AI clips into one film.
🎼 MuseScore Studio Free and Open Source: Write, play back and print sheet music for any instrument, a free alternative to Sibelius.
🖼️ Bing Image Creator Free: Microsoft's image maker turns a sentence into pictures, unlimited in Standard mode once you sign in.
🧠 Obsidian Freemium: Keeps your notes as plain files on your own device, free without limits, with paid sync.
🔖 Raindrop.io Freemium: Saves bookmarks, articles and videos in tidy collections, with unlimited bookmarks and devices free.
🗣️ ElevenLabs Freemium: Turns your text into lifelike voiceovers in many languages, with 10,000 free credits a month.
🌐 Wegic Freemium: Describe your business in a chat and it designs and publishes a simple website, no coding.
TRENDING
Sweden's Government Offices Block ChatGPT and Claude Starting Today. Staff who support the prime minister get Microsoft Copilot instead, the only AI tool that passed the office's security review. Decrypt points out Copilot can run Claude models too, and nobody has said whether that's switched on.
An Apollo Economist Says AI Hasn't Paid Off Outside Big Tech Yet. Torsten Slok looked at the S&P 500 minus seven tech giants and found flat profit margins and flat revenue per worker. Other recent studies haven't found big productivity gains either.
Forced to Pick, Grok Chose Republicans 92% of the Time and GPT Chose Democrats 97%. Researchers at Convovo asked both models to name one candidate in nearly 500 races through their developer access. The regular apps refused to pick, and OpenAI says the test doesn't reflect what users see.
OpenAI and Anthropic Cut Some AI Prices by Up to 80% in Three Months. Developer Max Woolf tracked the drops, including an 80% cut to OpenAI's GPT-5.6 Luna in July. The savings mostly reach apps and coders who pay by use, not your monthly plan.
One Person and 147 Claude Helpers Brought F-Zero X to the Nintendo 3DS. The fan port runs at about 60 frames per second on newer 3DS systems. Its first phase cost roughly $4,974 in AI usage, by the developer's own estimate.
Meta Pushed Back New Ad Rules That Let It Scrape Business Websites. The terms now start November 20 instead of October 30, after publishers objected. Meta had already dropped the line about using that content to train its AI.
TRY THIS PROMPT (copy and paste into Claude, Grok, ChatGPT, or Gemini)
🎬 Shot by Shot: turn a one-line story into a ready-to-film AI video plan
Build a single-file HTML app called Shot by Shot in vanilla HTML, CSS and JavaScript. No API key needed.
I type a story idea in a sentence or two, pick a length (4, 6 or 8 shots) and a look, like realistic film, hand-drawn animation or stop-motion clay. The app calls the Anthropic Messages API to turn it into a shot plan for any AI video tool.
Show a character sheet that describes each character's face and clothes plus the setting, then one card per shot with what happens, the camera move, how many seconds it runs and a ready-to-paste video prompt that repeats the character sheet word for word so every clip matches. Add copy buttons and three tips for keeping clips consistent. Use only original characters, never real people, brands or famous characters.
Style: near-black background, amber and sea-blue accents, Cinzel, Cormorant Garamond and JetBrains Mono fonts.What this does: Type a story idea in a sentence or two, pick how long and what look you want, and the app plans your short film shot by shot. You get a character sheet so faces and clothes stay the same in every clip, plus a camera move, a length and a ready-to-paste prompt for each shot. Copy the prompts into whichever AI video tool you like. It only uses original characters, never real people.
JOIN THE COMMUNITY
Learn AI With Us, Free
Most people try AI once, get a so-so answer, and quit. Our free 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 actually 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. There's even a cinematography course coming out soon.
WHERE WE STAND(based on today’s news)
✅ AI Can Now: Make seven connected minutes of 4K film where some shots look like real actors on a real sea.
❌ Still Can't: Write and act a scene people care about. TechCrunch's critic found the dialogue stiff and the characters flat.
✅ AI Can Now: Steer two-armed robots it has never seen through new tasks with no extra training, across 27 simulated robots and 4 real ones.
❌ Still Can't: Work robot hands or move a robot around a room. Dream-1 only controls where a gripper ends up, and its makers say it will make mistakes.
✅ AI Can Now: Look at a photo of a skin spot and name the right condition about 4 times in 10, in one dermatologist's test of GPT-6 Astra and Claude Fable 5.1.
❌ Still Can't: Match a small 2015 model built just for skin, which got 62%, or tell harmless moles from risky ones without sending everyone to a doctor.
RECOMMENDED LISTENING/READING/WATCHING
Before you watch anyone's version, AI or Nolan, read the story itself. Wilson's 2017 translation keeps Homer's line count and reads in plain, quick English, which makes it a great first try at the epic. Notice how much it grips you with nothing but words, the part today's AI video still can't fake.
How was today's edition? 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.





