Claude designs proteins to fight scorpion venom

Hola Decoder😎

If someone forwarded this to you and you want to Decode the power of AI and be limitless, then subscribe now and Join Decode alongside 30k+ code-breakers untangling AI.

🦂 Anthropic's Claude designs 318,000 proteins in 64 hours to fight scorpion venom

Anthropic has published research showing how an internal Claude model designed 318,484 candidate proteins in 64 hours, with independent laboratories confirming that 17 could bind to dangerous scorpion toxins, potentially advancing AI-assisted drug discovery.

The Shift:

Claude Generates 318,484 Protein Designs - Working from one initial prompt and four follow-ups, Claude autonomously generated and screened hundreds of thousands of candidates over 64 hours, using approximately 10,000 GPU hours.

17 Designs Pass Independent Lab Tests - Researchers at Twist Bioscience and Adaptyv Bio tested 211 selected proteins, confirming that 17 successfully bound to individual scorpion neurotoxins under laboratory conditions.

Universal Antivenom Goal Remains Unmet - None of the successful designs could bind multiple toxins. Researchers also have not demonstrated that the proteins neutralize venom, meaning further cellular and animal testing is required.

AI Research Could Reduce Discovery Time - The project involved five researchers from Citrini Research and an estimated $200,000 budget. Anthropic has released all 211 designs and their binding data to support further scientific research.

The experiment demonstrates how AI can accelerate early-stage protein discovery, generating thousands of candidates with limited human guidance. However, binding to a toxin is only the first step toward developing an effective antivenom, and substantial testing remains before any resulting treatment could reach patients.

Together with Omnisend

One Plain-Text Email Outearned a Month of Campaigns 

Before you send everyone the same BFCM discount, check who keeps coming back. Your repeat buyers could justify an offer of their own.

For one skincare brand, 36% of customers had ordered more than six times. Enflow Digital used Omnisend to turn that loyalty into nearly $18,000 from one plain-text email.

Omnisend helped the team:

• Find purchase patterns in minutes, cutting out hours of manual analysis.
• Build a VIP audience of customers with four or more orders, ready for a separate promotion.
• Keep the VIP offer focused on engaged buyers by filtering out contacts inactive for 180 days.

The offer was available only through email: a higher discount plus a free gift above a higher spend threshold. It increased order value while helping clear slower-selling stock.

Co-founder Karolina shares how it came together so you can apply the approach to your own repeat buyers before your BFCM offers are locked in.

✨How to redesign any room and get a real shopping list using ChatGPT

One photo. A professional redesign. A shopping list under $500 to actually build it.

Step 1: Take a good photo - Tidy the room, open the blinds, shoot from the doorway so the whole space is in frame. Good input makes a realistic result. Bad lighting makes a fantasy render.

Step 2: Upload and run the redesign prompt - Attach the photo and paste this:

"Redesign this room like a professional interior designer. Keep the same layout, windows, and proportions. Upgrade the furniture, colors, lighting, and decor. Photorealistic. Give me a few versions."

Pick your favorite.

Step 3: Turn on web search - Tap the globe icon under the message box. This pulls real products with working links instead of invented ones.

Step 4: Get the shopping list - Stay in the same chat and paste this:

"Give me everything in this redesign as a shopping list under $500. Item, price, and a link to buy it. Match the image and show the running total."

Step 5: Check before buying - Click through every link before ordering. Descriptions are reliable. Prices and URLs can go stale.

Renter? Add: "No painting, no drilling, damage-free changes only."

🤖 Sam Altman outlines three possible endgames for the AI race

OpenAI CEO Sam Altman reportedly presented three possible approaches to governing superintelligence during an internal staff meeting, ranging from international regulation to merging major AI labs or allowing one company to dominate the industry.

The Shift:

Altman Favors a Global AI Regulator - According to The Information, Altman appeared to support an international watchdog modeled on the International Atomic Energy Agency, which would establish and enforce safety standards for advanced AI developers.

Major AI Labs Could Merge Into One - The “Demis plan,” named after Google DeepMind's Demis Hassabis, proposes bringing leading AI laboratories together under a CERN-style organization that develops AI models for participating countries.

One AI Company Could Eliminate Competition - The “Eliezer plan” envisions a company using self-improving AI to gain overwhelming economic power and “melt everyone else's GPUs.” Altman reportedly called this scenario “terrible.”

OpenAI Faces a One-Year Decision - Altman reportedly described the next phase as “pacing 2.0,” saying OpenAI needs to choose a direction within a year and establish a speed limit or stopping point for self-improving AI.

Altman's reported comments highlight the growing tension between developing increasingly powerful AI and establishing safeguards before systems become difficult to control. International regulation and shared development would require substantial coordination, while continued competition raises questions about who sets the limits on superintelligence.

Together with Playbook Pro

Your Next 1,000 Customers Are Already Here

Pouring more money into ads won’t fix flat growth. The real leverage is hiding inside the customers you already have, and every day you wait, that momentum leaks away.

Networked Growth Engines is the 26-page field guide built from 200+ hours of research and real brand wins. It gives you ready-to-use systems you can plug in this week, no giant team or budget required. With it, you can:

🔥 Spark self-propelling communities using the Community Building Playbook, exactly how Nyfter built a 500,000-member Discord that drives product innovation.

📈 Design referral programs that fund themselves with the Referral ROI Calculator, the same loop that powered 34% of Blue Apron’s customer acquisition and delivered a 17.3× ROI for Latico Leathers.

🤝 Form trust-based partnerships fast using the Partnership MoU and 50-template Outreach Library to unlock new audiences without paid ads.

Your current customers are the growth engine. 

Download the guide today and start compounding growth before another week of ad spend vanishes.

🏆 Tools you Cannot Miss:

🌸 Scento MCP – Connects AI assistants to a searchable perfume database covering 1,400+ fragrances, 400+ raw materials, and fragrance families, with answers linked to Scento sources.

📱 Wavos – Connects ChatGPT or Claude to social media publishing, allowing users to upload and schedule posts while automating comments and DMs for customer engagement.

🤖 Tabloy AI – An AI agent designed to execute tasks and complete workflows rather than only generate responses.

🎵 Melody Generator – Converts text descriptions into MIDI melodies based on specified genres, moods, tempos, and musical keys for music production.

🎮 Spaleforce – Generates game plugins, 3D items, and complete structures from text prompts or reference images without requiring manual coding.

🚀 Quick Hits

🎵 Apple acquired Samplab, a developer of AI-powered music production tools, according to a European Commission filing dated June 17. The acquisition could support Apple's music creation software.

🎨 Gemini Notebook now suggests topics and outlines before generating slides or videos. Users can customize formatting, visual style, and scope, with an optional review step before final rendering.

🚀 SpaceX reported $14.6 billion in Q3 revenue, with AI contributing $7.5 billion and $431 million in profit. The company posted $4.4 billion in net income despite heavy infrastructure spending.

🤖 Google open-sourced TorchTPU, enabling PyTorch applications to run on its AI chips without rewriting code. Alongside Amazon’s TorchNeuron, the move challenges Nvidia’s dominance in AI development.

🔒 ChatGPT is rolling out Locked Chats, allowing users to protect conversations with a four-digit PIN, Face ID or Touch ID. Locked conversations remain accessible with separate chat history and memory.

👨‍💻 What’s trending on socials

🏭 Elon Musk makes bold Terafab prediction - Elon Musk claims Terafab’s output could rank among the world’s 20 largest economies by GDP. Reports suggest the chipmaking project is approaching multiple South Korean semiconductor suppliers.

💻 Claude can help fix slow computers - A viral post claims Claude’s desktop app can troubleshoot slow computers using seven prompts to identify performance issues, review startup apps, and manage storage.

🚀 Viral post highlights rapid technology breakthroughs - A viral post lists major AI and technology breakthroughs from five weeks, including mathematical research, quantum computing advances, AI-discovered antibiotics and offline AI models. Some claims remain unverified.

🤖 Viral video jokes about AI training in India - 🇮🇳 A viral Reddit post jokes about how AI models are trained in India, featuring a street demonstration with unusual product descriptions and humorously translated subtitles.

🤳AI Nugget of the Day

Thanks for Decoding with us🥳

Your feedback is the key to our code! Help us elevate your Decode experience by hitting reply and sharing your input on our content and style.

Keep deciphering the AI enigma, and we'll be back with more coded mysteries unraveled just for you!