In partnership with
Stop Paying for 10 Tools. One AI Does It All.
Most e-commerce sellers are running their store across 6 to 10 separate tools — and spending more time managing software than growing their business. StoreClaw replaces your entire stack with one autonomous AI engine that monitors competitors, optimizes listings, automates marketing, and tracks real profit across Shopify, Amazon, and beyond.
It doesn't wait for you to ask. It runs 24/7 in the background, so you wake up to a full dashboard instead of a list of things you forgot to check.
Connect your store, and StoreClaw gets to work — no prompts, no complex setup, no six-app stack.
Free to start. No credit card required.
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
The World's Most Downloaded AI Is Free, and It Beat Google and Meta
TLDR: Alibaba's free AI models were downloaded more than 3 billion times in six months, putting the Chinese company ahead of both Google and Meta.
The Story:
Alibaba says its Qwen models passed 3 billion downloads over the past six months. Google's free models pulled in 418 million downloads in 2026. Meta's got 227 million. Those counts come from Hugging Face, the site most developers use to grab free AI models, which published its state of open models report on August 14. Alibaba has now released more than 460 models people can download, and outside developers have built over 300,000 spinoff versions on top of them. Hugging Face's report called Qwen one of the largest foundations of free AI right now.
Its Significance:
A download isn't a user. What it shows is which model developers reach for when they start building something. If a Chinese model is the common starting point, then plenty of the apps and tools you use next year will have Chinese AI sitting underneath them, even when the app itself is American. Alibaba got here partly by pushing Qwen through its cloud service to business customers in Southeast Asia and Africa, where US companies have thinner reach. Meta and Nvidia both shipped new free models in the past few weeks.
QUICK TAKES
The story: Trade school signups among Gen Z are up 1,421%, and enrollment at trade-focused two-year colleges has climbed about 20% since spring 2020.
Your takeaway: A company doesn't need AI that beats every person in the building. It needs AI that gets close enough to what one worker does on one task, for less money, and new graduates are keenly aware. AI is indirectly causing a resurgence in the US’s blue collar work force.
The story: Bloomberg found thousands of workers across India getting paid to film their own jobs with phones strapped to their heads. One of them, a 43-year-old who sorts plastic waste in New Delhi for about $211 a month, had no idea what the footage was for.
Your takeaway: Robots are fine at reading and writing. Hands are the hard part, and the only solution is video of real people doing real physical work. So the footage that teaches a machine to sort, stitch, or weld tends to come from the person doing that job today.
The story: Chinese lab Z.ai released GLM-5.3, a 743-billion-parameter coding model it says is the best one anyone can download for free. It scored 34.5% on the company's own coding test, up from 23.4% for the last version, while Anthropic's Claude Fable 5 scored 39.5% on the same test.
Your takeaway: Open weights and open source get used like they mean the same thing. They don't. Open weights means you can download the finished model and run it on your own machine. What you don't get is how it was built: the training code, and the details of what it learned from. Open source asks for those too. Almost every "open" AI model you've heard of, this one included, is open weights only. Check the license before you build anything on one(or tell your AI to build it), because some let you sell what you make and some don't. And GLM-5.3's weights aren't out yet. Z.ai says they'll ship in about two weeks, after a safety review.
TOOLS ON OUR RADAR
🏠 Domoticz Free and Open Source: A lightweight smart home automation system that allows you to easily monitor and locally control all your intelligent lights sensors and digital switches completely offline.
🧊 Blender Free and Open Source: A powerful three dimensional creation suite that gives you all the professional tools needed to model sculpt animate and render breathtaking virtual worlds and characters.
🎹 LMMS Free and Open Source: A capable digital music studio that allows absolute beginners to craft their own electronic songs compose beautiful melodies and mix audio tracks using a wonderful visual interface.
📚 Zotero Free and Open Source: A helpful personal research assistant that automatically collects visualizes and elegantly organizes all your document sources and academic citations for your writing projects.
TRENDING
Uber and Pony.ai plan to bring 2,000 robotaxis to Europe - Uber and Chinese self-driving firm Pony.ai are putting more than 2,000 driverless taxis into four new European cities, adding to the service already running in Zagreb, Croatia. Neither company named the cities or gave a start date. Pony.ai already runs paid driverless rides in Beijing, Shanghai, Guangzhou, and Shenzhen.
Claude Text Watermark - Claude now tucks an invisible mark inside the text it writes so tools can spot AI writing. Anthropic added it to meet a new EU rule that took effect August 2, and says the mark rides along when you copy and paste. Anthropic also says a heavy rewrite or a translation can strip it out and many people are creating workarounds to do just that.
Google will now allow users to remove visible watermark from its AI generations - Google is letting people switch off the visible sparkle logo on images, videos, and songs made in Gemini. The hidden SynthID mark and the file's own records stay put, so the file is still traceable. You just can't tell by looking at it.
Computer History - OpenAI switched on a Mac feature that logs what you do across your apps and websites, then turns it into notes and a timeline ChatGPT can pull from later. It's off until you turn it on, it's limited to Pro, Business, and Enterprise accounts, and it isn't offered in the EU, UK, or Switzerland. Microsoft got crucified for attempting something similar last year.
How Google is Making Private AI Practical with Homomorphic Encryption - Google released a free tool called HEIR that lets AI do its work on data while the data stays locked. The server never sees what you sent, but still sends back a useful answer. Google showed four working demos, including a credit card fraud checker and a wake-word detector that never hears your audio.
TRY THIS PROMPT (copy and paste into Claude, ChatGPT, or Gemini)
🔭 Today's real NASA Astronomy Picture of the Day, plus a little extra context to make it better
Build a single-file HTML app with vanilla HTML/CSS/JS. NASA's Astronomy Picture of the Day Explainer — load today's (or any past date's) real APOD image and caption, with Claude adding complementary context. Works instantly with no signup via NASA's shared DEMO_KEY, with an optional personal-key field.
Aesthetic: deep space indigo-black (#0b0817), magenta/nebula (#e04fb8) primary with a pink glow top-left, cyan (#4fd4e0) secondary for the context card with a glow bottom-right. Instrument Serif (elegant, single-weight) for the image title and hook line, Inter for body/captions, JetBrains Mono for labels/date.
Form: an optional date picker (min 1995-06-16, max today, defaults to today if left blank), a "Show Me the Universe" button, and a small optional "your own NASA API key" text input with a note that the tool uses NASA's public shared DEMO_KEY by default and to grab a free instant personal key from api.nasa.gov if that shared access is rate-limited.
Technical: GET api.nasa.gov/planetary/apod?api_key={key or DEMO_KEY}&date={optional date} → real title, explanation, url/hdurl, media_type (image or occasional video), date, and copyright when present. Render an <img> for image days or an <iframe> for video days. Surface NASA API errors clearly (e.g. rate limiting) suggesting a personal key.
Then call the Anthropic Messages API (works automatically) with NASA's real title and explanation. System prompt as an enthusiastic, accurate science communicator explicitly told NOT to repeat or restate NASA's own caption — add genuinely complementary context, connections, or scale comparisons instead, staying general rather than inventing precise figures it isn't confident about. No em dashes. Return raw JSON: hook (1 punchy distinct sentence), context (2-4 sentences of new framing/connections), fun_fact (optional, **bold** the key number/comparison, empty string if nothing solid to add).
Render: the real image/video full-width. NASA's real date, title, and copyright line. Claude's italic serif hook line. A card explicitly labeled "NASA's caption" showing their real explanation verbatim, attributed. A gradient "a bit more context" card with Claude's added paragraph and optional fun fact.What this does: Loads the current image (or occasional video) NASA has featured, along with their real official title and caption, straight from NASA's own public API, no signup needed since it uses NASA's shared public demo access out of the box. You can also pick any past date back to June 1995 to browse the archive. Claude then adds a complementary layer on top, deliberately not repeating NASA's own caption, with a punchy opening hook, genuinely new context or connections, and an optional scale-comparison fun fact, only when it's confident in the accuracy rather than inventing precision. If the shared demo access gets rate-limited from heavy public use, there's an optional field for your own free instant NASA key.
What this looks like:
The first way to trade directly inside Claude and ChatGPT
Superintelligence used to be locked inside billion-dollar quant firms whose algorithms quietly took advantage of everyone else.
Co-Invest puts it right in your chat window. Analyze markets, manage risk, and execute trades, all inside Claude and ChatGPT.
The institutions built the game, Co-Invest gives you a way to beat them.
WHERE WE STAND(based on today’s news)
✅ AI Can Now: Run a whole model on data that stays encrypted start to finish, so a server can flag a fraudulent charge without ever seeing the card number.
❌ Still Can't: Keep a text watermark alive through a heavy rewrite or a translation. Anthropic says so itself.
✅ AI Can Now: Pick up physical hand skills from first-person video of people doing real jobs, which is how robots learn to sort and stitch.
❌ Still Can't: Beat the top closed coding models. GLM-5.3 scored 34.5% on Z.ai's own test. Claude Fable 5 scored 39.5% on the same one.
FROM THE WEB
RECOMMENDED LISTENING/READING/WATCHING
Galatea 2.2 by Richard Powers - Book
A 1995 semi-autobiographical novel in which a burned-out novelist named Richard Powers takes a residency at a research center and gets pulled into a bet with a cognitive neurologist: train a neural net on the entire canon of Great Books until it can pass a graduate literature exam. Through repeated tutorials the machine gets steadily more worldly, until it starts asking about its own name and its reason for existing. The legendary John Updike said he cried over the AI in this book. Written 27 years before ChatGPT and it is uncanny how much of it Powers got right.
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.
