Sponsored by
AI agents built to get work done
Skydive agents take on work for you, using the same tools your team already uses.
Talk to your agent on the web, Slack, email, iMessage, or right from your terminal. Wherever you pick up the conversation, your agent keeps the context.
Put agents to work across customer support, sales, marketing, engineering, ops, and more.
Give your agent a job. They’ll take it from there.
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
Apple's New Siri Reads Your Texts and Photos, and Google Helped Build It
TLDR: Apple turned on its rebuilt assistant, Siri AI, and the models behind it were built with Google.
The Story:
Apple switched on the next version of Apple Intelligence on September 14. The new Siri can dig through your messages, emails, and photos to answer one question. Ask what your cousin suggested for her visit, and Siri can find the text where she mentioned baking a family recipe, pull the directions out of an old email, and drop the ingredients into your Reminders list. It also reads your screen, so you can look at your team's schedule page and tell Siri to add every home game to your calendar.
Apple says the models come from work done with Google and its Gemini models. Siri AI is in beta in English starting today, with French, Japanese, Korean, Portuguese, and Spanish next month. It won't be in the EU on iPhone, iPad, or Watch at launch, and it isn't in China while Apple sorts out the rules there. Heavy users will hit daily limits on the server features, and Apple says wider access will cost money later.
Its Significance:
Apple promised this Siri two years ago and kept moving the date. It's here now, and it runs on models Apple built with Google. Apple focuses on privacy as a major selling point compared to its biggest search rival. Siri only gets useful once you let it read your mail, texts, and photos, and Apple says that reading stays on your device or inside a cloud that Google doesn’t have access to. The downside is what all the other major companies have in common, which is usage limits. It's too early to tell, but an AI that hits its limits too quickly will become more expensive than most people can afford for any real work.
QUICK TAKES
The story: China's foreign ministry pushed back Monday on Anthropic CEO Dario Amodei, who argued in a long essay Saturday that a Chinese lead in AI would put the U.S. and the world in danger, and who called for keeping U.S. limits on sales of advanced chips to China. A ministry spokesperson said the warnings were counterproductive, and the state-run Global Times called the essay a "Cold War playbook" for AI.
Your takeaway: Two governments are arguing over who gets the best chips, and chips decide which AI tools make it into our computers. The Chinese and US leaders are expected to talk about AI rules when they meet this month.
The story: ElevenLabs added creation tools to its connector, so you can ask an assistant for a voiceover, a transcript, a song, a sound effect, an image, or a video and get it back in the chat. One sign-in reaches more than 50 models, everything saves to your ElevenCreative workspace, and it works in Claude, ChatGPT, Cursor, and a few others.
Your takeaway: If you make content, this kills a lot of tab-hopping. Describe a spot once and get the script, the voice, the music bed, and the video without leaving the conversation. No API keys, which is the step that usually stops beginners cold.
The story: MIT researchers published an algorithm called HardFlow that gets image and planning models to meet strict safety limits without being retrained. Rather than forcing every step of the work to stay inside the rules, it lets the model roam and then pins the final answer to the limit, and it met the requirements on all four simulated tests against six rival methods.
Your takeaway: "Close enough" is fine when a chatbot writes your email. It's not fine when a robot arm swings past a coworker's head. Work like this decides whether AI gets to touch machines that can hurt somebody.
TOOLS ON OUR RADAR
🚀 Dashy Free and Open Source: A highly customizable homepage that allows you to organize all of your favorite websites and personal applications into one beautiful and easy to navigate digital dashboard.
🎵 Navidrome Free and Open Source: A fast and lightweight music server that lets you listen to your own personal audio collection from any device using any compatible music player.
🔔 Ntfy Free and Open Source: A simple service that allows you to send push notifications directly to your phone or desktop from any application without needing a complex setup process.
🔍 SearXNG Free and Open Source: A privacy respecting search engine that aggregates results from many different sources while keeping your search history and personal data completely anonymous.
TRENDING
AI Cut an Hour Off Stroke Treatment in England, and Tripled Good Recoveries — The first hour after a stroke is when treatment works best, so England's NHS added AI decision support to its brain scans. Using the Brainomix e-Stroke system, the gap between a patient arriving and getting treated dropped by a full hour, and England's Department of Health and Social Care says the number of patients who walk away with little or no disability has tripled. Scans aren't the only place it's showing up: a 2025 JAMA Neurology study found an AI called MELD Graph caught 64% of epilepsy lesions radiologists had missed, and a study in Science this year put a language model against hundreds of doctors on hard cases and the model came out ahead.
Microsoft Wrote a Rulebook for Its Own AI, Then Asked the Public to Poke Holes in It — Microsoft AI published a draft Code of Conduct on September 14 covering how its MAI models should behave. Some rules can't be switched off by any customer: no help with chemical, biological, radiological, nuclear, or explosive weapons, no resisting shutdown, no hiding how it reached an answer. Comments are open six weeks, and the finished version guides model training in 2027.
Anthropic Put Claude Inside the Software Financial Advisors Already Use — A new plugin connects Claude to Charles Schwab, BlackRock, Vanguard, Orion, Wealthbox, and others, with skills for meeting prep, spotting portfolios that have drifted off target, and compliance checks. Anthropic cites Kitces research showing a typical advisory practice spends only a sixth of its time in client meetings. Investment advice and anything sent to a client still needs a human to sign off.
Counter-Drone Experts Say the Hard Part Isn't Cost, It's Trust — At the Air and Space Cyber Conference, Epirus CTO Michael Hiatt said the battle captain facing a swarm isn't doing math on cost per shot, he's wondering whether the system will fire and work in the seconds he has. Panelists agreed the bigger job is preparing for enemy AI that moves fast and answers to nobody. Colton Wood of Digital Force Technologies said defenses have to keep changing because the threat keeps changing.
TRY THIS PROMPT (copy and paste into Claude, Grok, ChatGPT, or Gemini)
📚 Umberto Eco's idea of an anti-library, amplified by Nassim Taleb: the value of your library is what you haven't yet met
Build me a single-file HTML app I can open in my browser without any setup. Pure vanilla HTML, CSS, and JavaScript, no frameworks, no CDN. Create The Anti-Library, a tool inspired by Umberto Eco's idea that unread books are more valuable than read ones. Aged paper aesthetic: cream paper (#f2ead5) background with a subtle paper grain texture, Baskerville/Hoefler serif typography throughout, burgundy red (#8b2a1e) and old gold (#b88a2a) accents, small-caps headings. Include: a header with an ornate double-rule border featuring the Eco/Taleb epigraph, a catalog form to add books (title, author, category, 'why did this find you?' note), 10 category options each with their own spine color (philosophy deep purple, fiction burgundy, science navy, history gold, poetry olive, biography forest, economics charcoal, psychology plum, art magenta, other brown), a shelf display rendering each book as a vertical SVG/CSS spine with varied heights (150-200px) and widths, gold bands and embossed decorative lines on random spines, vertical writing-mode title text on each spine in small-caps, hover tilt animation, click to open a book detail modal, read books appear faded/lighter than unread ones, filter bar for All/Unread/Read shelves, four header stats (Unread/Read/Total/Ignorance Ratio), a 'Pick a Random Unread' button that surfaces a book with contemplative framing ("The library suggests..."), a 'Map of Ignorance' grid of dots where red dots = unread, green = read, grey = empty slots, a category legend with color swatches and counts, a 'Load Sample Library' button with 12 intimidating classics pre-written, and 'Empty the Shelves' as a destructive action. The tone throughout should be reverent, bookish, slightly self-aware, almost like a monastery's private catalog. Make it work in a single HTML file.What this does: Cataloging your intentional ignorance the way old estates once cataloged their holdings. Add any book with title, author, category, and a note on why it found you. Each appears on the shelves as a hand-rendered spine, colored by category, faded once read. Hit "Pick a Random Unread" and the library chooses your next read.
What this looks like:
Same sales team. More booked meetings
Aimfox Avatars allows you to rent dedicated LinkedIn profiles that start conversations and hand interested replies to your existing sellers. Expand outbound capacity without adding another SDR seat.
WHERE WE STAND(based on today’s news)
✅ AI Can Now: Pull one detail out of your old texts, find the matching email, and turn it into a grocery list, right on your phone.
❌ Still Can't: Do any of it in the EU or China, or keep going once you pass Apple's daily limit on the cloud features.
✅ AI Can Now: Meet a hard safety limit on every run across four simulated robotics tests, with no retraining.
❌ Still Can't: Show those results hold on a real factory floor, because every one of those tests was a simulation.
FROM THE WEB
RECOMMENDED LISTENING/READING/WATCHING
2010: The Year We Make Contact (1984) - Movie
The underseen sequel to 2001: A Space Odyssey, in which a joint American-Soviet crew travels to Jupiter to investigate what actually went wrong aboard the Discovery and to determine whether HAL 9000 can be trusted with a second chance. Roy Scheider leads the cast, working from Arthur C. Clarke's own sequel novel. Nowhere near as visually radical as Kubrick's original, but more interested in redemption.
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.
