In partnership with
Hire anyone, anywhere — compliant in under 3 days
Found the right person, but they’re in a country where you don’t have an entity? Setting one up can take months and significant cost.
Remote removes that barrier by becoming the legal employer through our own entities — handling compliant contracts, local benefits, tax setup, and onboarding for you. In fact, an employee is onboarded to Remote every 7 minutes.
Once they’re hired, the same in-house teams that support employment locally also run payroll — so you’re not bouncing between disconnected providers. Less setup, less complexity, and less time between finding the right person and getting them started.
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
Meta's New AI Agent Wants to Sell Your Car and Argue Down Your Bills
TLDR: Meta launched Muse, a personal AI agent that runs on its own private cloud computer and can send emails, book trips, and buy things for you once you approve each step.
The Story:
Meta announced Muse on September 8. It's freemium for most uses in the US on iPhone, Android, and at muse.ai, with paid plans for heavier use. You talk to it the way you'd text a friend, either in the Muse app or right inside WhatsApp. It runs on Muse Spark, the model Meta calls its most capable so far.
Every person's Muse gets its own virtual computer in the cloud, called Muse Secure VM, and that's where their data and logins live. A separate program named Sentinel sits on that same machine, and nothing Muse does reaches the internet unless Sentinel approves it. Meta says Muse can use your passwords without seeing them, and that it stops to ask before sending an email or making a purchase. Payments go through Link, built by Stripe, which spins up a one-time card number so your real card details stay hidden.
Its Significance:
Meta calls this the first personal AI agent built for everyone, meaning no setup skills required. The company says Muse conversations don't touch its ad systems, and that people can opt out of having their chats train Meta's models. Those are Meta's claims, and reviewers and regulators will spend the next few months testing them. A version where only you hold the encryption key, Muse Confidential VM, is promised later this year. The closest thing to Muse right now is xAI's Grok Bot, which also gives each agent its own cloud computer and its own logins, and it went into beta about a month ago on August 11. Expect to see more human to AI interfaces that feel like a messaging app from the other big companies.
QUICK TAKES
The story: OpenAI released a new image model on September 8 that it says generates up to 50% faster than Images 2.0 and holds earlier edits steady across a long back-and-forth. It's live for every ChatGPT tier, and a feature called Sketch lets you scribble a rough shape in the app and have it turned into a finished picture.
Your takeaway: If you quit on AI image editing because every new request wrecked the last change, it's worth another look. People make more than 3 billion images a week with OpenAI's tools, so small fixes reach a lot of computers.
The story: DeepMind released AlphaGenome Atlas, a free database predicting what happens when any single letter in human DNA gets swapped. That's 9 billion possible changes, pre-calculated into a 1-petabyte dataset, with one simple score called AVI so researchers can sort the interesting ones from the noise.
Your takeaway: Scientists understand about 2% of the genome well. Dr. Gareth Hawkes ran the Atlas against records from 54,000 UK Biobank participants and turned up 22% more links between DNA and traits. The website needs zero coding.
The story: The NSA, FBI, and CISA put out a joint advisory September 8 naming DeepSeek, Moonshot AI, Alibaba, MiniMax, StepFun, and Z.AI. The agencies say those firms pulled billions of tokens from US models since 2024, including Claude, GPT, Gemini, and Grok, using fake accounts and resellers they call transfer stations. Distillation means feeding one model's answers to a smaller model so the smaller one learns to copy it.
Your takeaway: The Treasury Secretary said sanctions and Entity List designations are on the table. Those free Chinese models a lot of people download, like Qwen, could get harder to reach.
TOOLS ON OUR RADAR
⌨️ Alfred Freemium: Replace Spotlight with a faster app launcher, file search, and web search, the free core is fully usable, workflows and automation live behind a one-time Powerpack purchase.
📹 Whereby Freemium: Jump into a video call from a browser with no downloads or installs, a single shared link gets everyone in the room.
🌐 Framer Freemium: Design and publish a full website with a visual, Figma-like editor, the free plan has no time limit but stays on a Framer subdomain with branding.
📄 OnlyOffice Free and Open Source: Edit documents, spreadsheets, and presentations offline with full Word, Excel, and PowerPoint format compatibility, download once and it stays free forever.
TRENDING
Claude subscribers sue Anthropic over Max plan limits, An expanded class action says Anthropic's $100 and $200 Max plans don't deliver the 5x and 20x usage over the $20 Pro plan that the marketing graphics promise. The complaint argues those multiples apply only inside five-hour windows that a weekly cap then limits. Two of the attorneys on it, Monica Vaca and Kati Daffan, used to work at the FTC.
Instacart launches an AI shopping assistant called Clementine, Ask for a week of budget kids' lunches and it builds a cart from what's on the shelf at your store today. It's live across the US and Canada, built on 1.6 billion past orders from roughly 100,000 stores. Snap a photo of a handwritten list and it'll do the typing for you.
Suno rebuilt its music models on licensed songs, The new v6, v6 wild, and v6 mini models were trained on catalogs from Warner Music Group, BMG, and Believe, not the data behind the old versions. Every older model gets retired. Universal and Sony are still suing, and a German court ruled against Suno back in July.
A chip that sees and remembers in the same spot, Researchers at RMIT University in Australia built a prototype that senses light, stores what it saw, and processes it all on one atom-thin material. No shuttling data between a camera, a memory chip, and a processor. It spots numbers, shapes, and motion, keeps only what changed, and the team says a smart bionic eye is the long-term goal. Still early.
The Marine Corps is adding 6,500 positions for drones and AI, The billets spread across 64 separate initiatives over five years, aiming for 4% growth by 2031. Each Marine Expeditionary Force gets its own robotics and autonomous systems company. Software engineering, data science, and AI become a primary job field under the 0673 code, and a second drone school opens at Camp Pendleton.
TRY THIS PROMPT (copy and paste into Claude, Grok, ChatGPT, or Gemini)
🌿 Name a plant. Get its real, documented toxicity status for dogs and cats, side by side.
Build a single-file HTML app with vanilla HTML/CSS/JS. The Pet Plant Toxicity Checker — name a plant, get real toxicity status for dogs and cats via live web search grounded in ASPCA/veterinary sources. No API key needed.
Aesthetic: dark forest green (#10160f), warm gold (#d4a24c) primary with a glow top-left, sage green (#5cb86c) secondary with a glow bottom-right, plus a red/amber/green traffic-light system for toxicity status. Baloo 2 (rounded, friendly, pet-appropriate) for the plant name and status text, Inter for body, JetBrains Mono for labels.
Form: single plant-name text input.
Technical: call the Anthropic Messages API WITH tools:[{type:'web_search_20250305', name:'web_search'}] (works automatically), system prompt as a pet safety researcher who searches for real documented toxicity info, prioritizing ASPCA Animal Poison Control and veterinary sources, reporting dog and cat toxicity SEPARATELY since they often differ, real documented compounds/symptoms/severity, honest when a plant isn't well documented, and always closing with standard non-negotiable guidance to contact a vet or ASPCA Poison Control immediately if ingestion is suspected — explicitly never giving specific treatment or dosing advice. Return raw JSON: plant_name, toxic_to_dogs, toxic_to_cats (each: Yes/No/Mildly toxic/Not well documented), toxic_compounds, symptoms, severity_note, what_to_do.
Render: the plant name as a header. A 2-card side-by-side row (dog/cat icons) each color-coded red/amber/green/neutral by toxicity status. A detail card with what makes it toxic, symptoms, and severity. A prominent red "if ingestion is suspected" action card with the vet/poison-control guidance.What this does: Type a plant's common name and it searches the live web for real toxicity documentation, prioritizing the ASPCA's Animal Poison Control database and veterinary sources, since dogs and cats often react differently to the same plant. You get a clear color-coded status for each species, what makes it toxic, real documented symptoms, and a general severity read, mild GI upset versus something more serious, based on real reported cases.
What this looks like:
The best in influencer marketing. And you’re invited.
Get ready for Return on Influence Festival '26. An entire day dedicated to influencer marketing with speakers running some of the best programs in the world. October 21, 2026, 10 am - 5 pm ET.
WHERE WE STAND(based on today’s news)
✅ AI Can Now: Take a goal you describe in plain words, open a browser, fill out forms, and pay with a one-time card number once you approve it (Meta's Muse).
❌ Still Can't: Be turned loose without that approval. Meta built a second agent, Sentinel, whose only job is checking everything Muse tries to send out, which tells you how often agents still get things wrong.
✅ AI Can Now: Predict the molecular effect of all 9 billion possible single-letter changes in human DNA, and rank them with one score (AlphaGenome Atlas).
❌ Still Can't: Confirm a single one of those predictions. Every hit still has to be tested in a lab before it means anything for a patient.
FROM THE WEB
Here’s Part 2 of the Practical AI Agent Series. You can build along using Grok Bot or if you prefer a different company you can always just copy the prompt that I’m using.
RECOMMENDED LISTENING/READING/WATCHING
Idoru by William Gibson - Book
A 1996 novel about a world-famous rock star who announces plans to marry an idoru, a fully synthetic AI performer generated from massive datasets rather than born, and the two very different people, a corporate data analyst and a teenage superfan, drawn into finding out what that marriage would actually mean. Gibson wrote it nearly three decades before AI-generated pop stars avatars and virtual influencers became a real commercial category, and the novel's central question, whether a relationship with something built entirely from data can be real, has only gotten more relevant.
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.
