In partnership with
Show your client a working version in your first meeting
Your client doesn't care how their site was built. They care that it launches on time, runs on their own domain, and keeps working long after you hand it over.
With Lovable, you describe what your client needs and get a working version you can show them in the first meeting. Then you take it all the way to a live site on your client's domain, built to be found on Google. Lovable's project monitoring feature spots bugs and suggests fixes, so your name stays on work you're proud of.
Start today and take $5 off your first month.
Discount of $5 applies to first month of any paid plan. You need to sign up via this webpage in order to claim the discount. The discount applies to new and returning users without an active paid plan. Terms apply.
Beginners in AI
Good morning and thank you for joining us again!
Scorpions sting about 1.2 million people a year, and the antivenom is often pricey or hard to find. Yesterday Anthropic showed Claude designing tiny proteins that grab onto scorpion toxins, and two outside labs confirmed that 17 of them work in a test tube. Today I'll walk through what was tested, what wasn't, and why the next steps matter more than the headline. As always, every story here is human picked and edited.
THE FRONT PAGE
Claude Designed 17 Proteins That Latch Onto Scorpion Toxins in 64 Hours
TLDR: Anthropic says Claude designed proteins that grab onto scorpion toxins, and 17 of them passed tests at two outside labs, an early step toward cheaper antivenom.
The Story:
Anthropic's science team asked Claude to design minibinders, tiny proteins built to stick to one target, against five scorpion toxins from Mexico, Brazil, North Africa and the Middle East. Over about 64 hours, running mostly on its own for the first day, Claude produced 318,484 designs and picked 180 to test. The team added 31 more with Claude's help.
Two contract labs, Adaptyv Bio and Twist Bioscience, tested all 211. Seventeen passed, about 8%. The two best stuck to the Mexican scorpion toxin Cn2 very tightly and seem to grab the same spot as an antibody that has already protected mice from that toxin.
Its Significance:
Scorpion stings hit poorer, hotter regions hardest, which is exactly the kind of problem that drug companies rarely rush to fund. If AI can handle the first round of design in days, scientists get to spend their time on the hard part: proving a molecule is safe and works in a living body. Anthropic is clear this isn't an antivenom yet. The proteins were only tested against lab-made toxins, not in cells or animals, and the run used about 10,000 GPU hours, far more computing than most labs can get. Outside reviewer Timothy Jenkins put it plainly: binding "is no longer the main bottleneck." Anthropic released all 211 designs and the test data, so any lab can pick up where it left off.
QUICK TAKES
The story: Tap "Check for Me" on a restaurant's page in the Yelp app, enter your party size and phone number, and an AI voice agent calls to ask about the wait, then texts you the answer. Yelp's assistant can now also book tables through OpenTable and remember your past chats.
Your takeaway: It's a small, low-risk errand, which makes it a good first agent to trust. One snag TechCrunch points out: if the restaurant has its own AI answering the phone, the two bots may not understand each other.
The story: Spotify says it removed over 50 million slop and spam tracks from recommendations this year and deleted over 75 million more outright. AI-made "artists" now carry an AI Persona badge and stay out of your recommendations unless you follow them.
Your takeaway: Curious how a song was made? Check Song Credits on the Now Playing screen, where artists list the parts AI helped with. Spotify admits its AI detection has limits, so some slop will still slip through.
The story: Orchestra makes AI agents that track drug studies and flag which timelines slip when one part of a program hits trouble. OpenAI is folding it into Rosalind, its life sciences tools used by Moderna and Novo Nordisk, and didn't share the price.
Your takeaway: OpenAI notes a new medicine takes roughly 10 to 15 years to reach approval. Much of that time goes to coordination rather than discovery, and that's the part Orchestra's agents go after.
TOOLS ON OUR RADAR
🎨 Lunacy Free: Design graphics and app screens offline on Windows, Mac or Linux, a free alternative to Figma or Sketch.
🛡️ Quad9 Free: Change one network setting and this Swiss nonprofit blocks known malware and phishing sites on every device.
🌸 drip Free and Open Source: Track your period and cycle with everything kept on your phone, a private swap for cloud-based trackers.
💬 SimpleX Chat Free and Open Source: Message friends without giving out a phone number; SimpleX says it uses no user IDs at all, worth a look after today's Muse leak.
🎬 VEED Freemium: Edit videos right in your browser, trim clips and add subtitles, with a free plan to start.
🖌️ Magnific Paid: Make and edit AI images and video, plus 250 million stock files, from $14.50 a month billed yearly.
TRENDING
Meta Switched Off Muse Chat Sharing After 6,800 Shared Links Turned Up. Security firm Claroc found the links in public web records because each one put the chat's title in its address, and some held personal and business details. Meta told Ars Technica it has disabled sharing; if you shared a Muse chat, delete the link.
Perplexity's Computer Agent Can Now Run Entirely on Your Mac. Pro, Max and Enterprise subscribers with Apple silicon Macs can keep planning, reasoning and file work on the machine, using a version of the open Qwen model. Perplexity recommends 36 GB of memory or more, so most laptops won't qualify.
Gemini Gets a Home Base for Small Business Owners. Owners can upload a brand kit so Gemini makes on-brand graphics, and turn receipts or spreadsheets into charts. It connects to Shopify, Canva, Xero and more, rolling out to all Gemini app users in the coming days.
Eka Raised $150 Million for a Robot Hand That Feels What It Grabs. Started by an MIT professor and a former DeepMind researcher, Eka says its gripper handles about 1,500 objects an hour, versus about 600 for a person. It's now valued at $1.45 billion but hasn't been proven on a real production line yet.
Otter.ai Raised $320 Million Without Selling Any Shares. General Catalyst's fund gets paid back from Otter's marketing results instead of owning a piece of the company. The meeting note-taker says it has 40 million users and is testing a helper that answers your questions during a call.
Canadian News Publishers Sued Anthropic Over Claude's Training. Postmedia, Torstar, The Globe and Mail and The Canadian Press say Claude was trained on their articles without permission and that Anthropic got around their sites' copy protections. Anthropic hasn't commented, and none of the claims have been proven in court.
TRY THIS PROMPT (copy and paste into Claude, Grok, ChatGPT, or Gemini)
🧪 How Far Along? See how close an AI science headline really is to helping you
Build a single-file HTML app called How Far Along? in vanilla HTML, CSS and JavaScript. No API key needed.
I paste an AI science or medicine headline or link. The app calls the Anthropic Messages API with the web_search tool to find the original announcement and independent coverage, then places the work on an 8-step ladder: computer design, lab dish, living cells, animals, early human trials, large human trials, regulator approval, everyday use.
It must never invent facts, numbers or sources. If it can't find the work, it says so instead of guessing. General information only, no medical advice.
Show the title, who did it and when, a one-line verdict, the ladder with the current step glowing, then cards for what it showed, what hasn't been shown yet, what comes next, and source links.
Style: near-black background, lime and coral accents, Space Grotesk and Newsreader fonts.What this does: Paste any AI science or medicine headline, like today's scorpion toxin story, and the app looks up the original work and places it on an eight-step ladder from computer design to everyday use. You see what the research showed, what it hasn't shown yet and what has to happen next, with links to the sources. If it can't find the work, it says so instead of guessing. It's a fast way to tell "promising first step" from "ask your doctor about it."
See what's really happening in every deal
Aligned shows you what buyers are doing between meetings inside your most important deals then surfaces the risks, openings, and next steps. No more guessing where a deal stands or finding out a champion went quiet too late. Just real visibility, so you can act before it's too late.
WHERE WE STAND(based on today’s news)
✅ AI Can Now: Design hundreds of thousands of candidate proteins in under three days and pick ones that stick to a toxin; 17 of 211 passed outside lab tests.
❌ Still Can't: Show that those proteins protect a living body. None has been tested in cells or animals yet, and about 92% of the tested designs failed.
✅ AI Can Now: Phone a restaurant for you, ask how long the wait is, and text you the answer.
❌ Still Can't: Count on understanding another AI on the other end of the line, a problem Yelp's agent may hit as more restaurants use voice bots.
FROM THE WEB
RECOMMENDED LISTENING/READING/WATCHING
Derek Muller explains how DeepMind's AlphaFold learned to predict the shape of proteins, a puzzle that stumped scientists for about 50 years, and why a protein's shape decides what it can do. It's the clearest plain-English primer on proteins I know. Watch it and today's lead makes more sense: Claude was working on the next step, designing brand-new shapes to grab a target.
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.
