Claude adds protein design to its resume
Anthropic's Claude models successfully ran autonomous protein-design campaigns, achieving 22-35% success rates on molecular targets—significantly outperforming industry norms of 10-15%. The newsletter also covers Uber's "Agentic Pods" approach to controlling AI costs, the Loop Method for improving ChatGPT workflows, and Replit's new Free Mode leveraging cheaper efficient models.
Summary
Anthropic published research demonstrating that Claude models can autonomously execute early-stage drug discovery work, specifically protein design campaigns. Using Claude's Mythos Preview and Opus 4.8 models with a single expert-written prompt, internet access, and tools, the system identified molecular candidates with success rates of 22-35% for binding to targets—substantially higher than the typical 10-15% industry baseline. The actual lab validation was conducted by Twist Bioscience and Adaptyv Bio across 15 targets, with 14 showing working molecules. Claude Opus 5 separately demonstrated the ability to read raw instrument files without specialized lab software, measuring sample purity at 96.4% in 19 minutes versus four days for traditional lab reporting. CEO Dario Amodei had recently claimed Anthropic's biology work was months away from "early glimmers," but this research delivered results within days of that statement.
The newsletter features Nate Grahek's analysis of Uber's response to overspending on AI by 300% of their 2026 budget in four months. Uber's CTO Praveen Neppalli Naga created "Agentic Pods"—two-person teams pairing an AI-proficient engineer with domain experts from Finance, Marketing, or Operations—to run 10-day sprints on specific projects. This approach yielded a financial pacing report that previously took two full days to produce but now runs in 10 minutes, demonstrating that enterprise organizations must earn incremental wins before pursuing larger AI initiatives.
The Loop Method is presented as a technique for improving repeated ChatGPT workflows through iterative self-improvement, using adversarial review from sub-agent panels to refine processes until a defined completion state is reached. Replit launched Free Mode on its paid plans, routing routine tasks to OpenAI's cheaper GPT-5.6 Luna model without consuming user credits, allowing 30x usage scaling on lower tiers while reserving premium models for complex work. Additional news includes Z AI's GLM-5.3 ranking fourth globally among AI models, AWS re:Invent offering 2,200+ hands-on sessions, Cerebras releasing a 30x faster AI computer, and OpenAI introducing Private Safety Processing for content moderation without staff visibility.
About this episode
PLUS: Use the Loop Method for better ChatGPT results
Key Insights
- Anthropic's Claude models achieved protein-design success rates of 22-35% on lab-validated targets, exceeding the 10-15% industry standard, while operating autonomously with minimal human intervention beyond an initial prompt.
- Uber discovered that unsubsidized frontier models are prohibitively expensive for unrestricted organizational use, leading them to restrict AI access to specialized two-person Agentic Pods running 10-day sprints on specific efficiency gains rather than broad deployments.
- Claude Opus 5 can interpret raw scientific instrument files and perform measurements without requiring proprietary lab software, completing analysis in 19 minutes versus the four-day timeline of traditional lab reporting methods.
- The economics of AI products are shifting as lower-cost models maintain competitive intelligence levels—Replit's Luna model's 80% price reduction enables 30x usage scaling on basic plans without perceived quality degradation.
- Dario Amodei's statement that Anthropic's biology work was months away from results was contradicted within days by the publication of functional protein-design research validated in external laboratories.
Topics
Transcript
Good morning, {{ first_name | AI enthusiasts }}, and welcome to our 12,567 new readers. Four days ago, Dario Amodei went on X to say Anthropic's biology work was months away from its first "early glimmers." Did he undersell the timeline? Claude just added protein design to its already lengthy resume, with new research showing the models running an early step of the drug discovery pipeline on their own and producing results that held up in the lab. Reminder: Our next live workshop is today at 2 PM EST — Join and learn how to land your first paid AI consulting gig on Upwork. RSVP here . Anthropic's Claude tackles protein design Nate's Notebook: Uber's tiger team playbook Use the…
Full transcript available for MurmurCast members
Sign Up to AccessMore from The Rundown AI
OpenAI’s “generational leap” with GPT-6 Astra
OpenAI released GPT-6 Astra, positioning it as a major advancement in AI with exceptional benchmark performance across multiple domains. The newsletter also covers Google's improved weather forecasting model, the Loop Method for ChatGPT optimization, and a reader's positive-news-only AI app.
Meta, Google join the AI launch party
Meta and Google launched new AI models in early September, with Meta's Muse Spark 1.3 achieving near-frontier performance at low cost while Google's Gemini 3.8 Flash represents a recovery step but still trails the frontier. The newsletter also covers AI safety concerns about reasoning transparency, tech literacy as a career ceiling, and practical AI workflows for professional development.
Fable 5.1 kicks off launch week at the frontier
Anthropic released Claude Fable 5.1, showing significant improvements in coding and research tasks with reduced safety rejections, marking the end of the summer's cautious release period as OpenAI's Astra launch approaches. Bernie Sanders published an op-ed calling for a global AI pause, citing control concerns and societal risks, while ongoing legal battles between Apple and OpenAI center on alleged theft of confidential designs.
Runway's Solaris previews the no-code internet
Runway unveiled Solaris, an AI-powered interface that renders websites and apps as real-time video with no underlying code, while Imperial College researchers developed an AI model that detects heart disease from ECGs in under two seconds with superior accuracy to human doctors.
OpenAI cuts out SpaceX-owned Cursor
OpenAI is removing its models from Cursor coding editor by mid-November following SpaceX's acquisition of the platform, citing Elon Musk's history of contract violations as justification. The move escalates the ongoing feud between Sam Altman and Elon Musk while putting developers in the middle of a high-profile corporate conflict.