In partnership with
Dear Readers,
Today, we are dealing with nothing less than the question of how close we are to a future in which machines not only work for us, but think with us—and could even extend our lives. Whether it's code that understands itself, robots that can walk even with broken limbs, or algorithms that fold proteins on a laptop, a new era is dawning in which AI not only provides tools, but also changes fundamentals.
In this issue, you'll find insights into Meta's Code World Model, SAP's plans for a sovereign OpenAI cloud in Germany, and the lean revolution in protein folding. We also look at new AI agents in longevity research, Aubrey de Grey's radical combination strategies, and the search for anti-aging ingredients in existing drugs. Stay tuned—the most exciting stories are right here in front of you.
In Today’s Issue:
💻 Meta's CWM makes AI understand code execution
🧪 K-Dense Beta can now accelerate longevity research
💊 Researchers using network models to discover existing drugs that target aging mechanisms
🧬 Aubrey de Grey proposes combining multiple interventions to combat aging
✨ And more AI goodness…
All the best,
Robots That Really Adapt
Skild AI proposes an “omni-bodied brain” model — one AI trained across a vast diversity of robot morphologies — so it cannot overfit to one body and must learn generalizable strategies. They show that the same model, without fine-tuning, can adapt zero-shot to robots with broken limbs, jammed wheels, altered structure, etc. The core takeaway: real-world robotics demands adaptation, not memorization — robustness comes from diversity in training.
SAP and OpenAI Launch Sovereign AI
SAP and OpenAI are teaming up to roll out OpenAI for Germany, a sovereign-cloud AI offering tailored to German public institutions. It will run on SAP’s Delos Cloud (built on Azure), ensuring compliance with data sovereignty, security, and legal standards. The goal: speed up administrative workflows while maintaining trust and control over AI infrastructure.
SimpleFold Simplifies Protein Folding
A new preprint introduces SimpleFold, a 3B-parameter protein folding model that drops complex domain-specific modules in favor of general-purpose transformers and flow matching. Trained on 9M structures, it achieves efficient, scalable performance and even runs on consumer hardware like MacBooks. This could democratize access to protein structure prediction.
OpenAI is developing a “Daily Pulse” feature that delivers personalized insights each morning via ChatGPT. The tool appears tied to Memory functions and may connect with the previously hinted “Golden Hour” concept. This signals OpenAI’s push toward more habitual, daily-use AI experiences.
The Takeaway
👉 CWM gives you a model that reasons about code execution, not just syntax.
👉 Its open-weights release invites broad scrutiny, extension, and hybrid modeling.
👉 Researchers can use it to explore agents that plan, predict, and self-debug.
👉 This is a stepping stone toward coding AIs that anticipate consequences — not just generate lines.
From the moment we fed machines code, we’ve asked: can they understand it — not just spit it out? Enter Meta’s CWM (Code World Model), a 32-billion-parameter open-weights LLM designed to bridge that gap.
Unlike traditional code-generation models that learn only from static code samples, CWM is trained on execution traces and interaction trajectories — think of it as not just reading a recipe but mentally cooking the dish step by step. This lets it reason about how code changes state, predict behaviors before running, and catch errors proactively.
For the AI community, CWM signals a shift: from surface pattern matching to semantic reasoning in code. It offers researchers a sandbox to explore hybrid models that combine planning, simulation, and symbolic thinking.
Why it matters: It pushes the frontier from “code generation” to “code understanding,” enabling smarter debugging, safer transformations, and a new class of AI-augmented development tools.
Sources:
Where to find high-intent holiday shoppers
Let’s be real: most brands are targeting the same people this holiday season. “Shoppers 25–54 interested in gifts.” Sound familiar? That's why CPMs spike and conversion rates tank.
Speedeon’s Holiday Audience Guide breaks down the specific digital audience segments that actually perform—from early-bird deal hunters actively comparing prices to last-minute panic buyers with high purchase intent. These aren't demographic guesses.
And our behavioral audiences are built on actual shopping signals and real-world data —the same approach we use for clients like FanDuel and HelloFresh."
You'll get the exact audiences, when to deploy them, which platforms work best, and what kind of performance to expect.
Download the guide and get smarter about your holiday targeting before the holiday rush hits.
New AI tool demonstrates potential to accelerate longevity research
An AI system called K-Dense Beta, developed by Biostate AI, orchestrates complete research cycles (experiment design, data analysis, publication) — reducing processes that often take years to just days. In cooperation with David Sinclair's laboratory, this agent architecture aims to accelerate aging research and produce new drug candidates more quickly. For the future: AI as a collaborator in gerontology, not just a support tool.
Network-driven discovery of repurposable drugs targeting hallmarks of aging
This preprint presents a network model that embeds 2,358 longevity-associated genes into the human interactome to identify existing drugs that specifically modulate aging mechanisms. The method uses both “proximity” and a novel transcription index (pAGE) to evaluate drug effects—a structured bridge between molecular mapping and clinical repositioning.
Aubrey de Grey on the Future of Longevity and Aging Research
At the Longevity Summit Dublin 2025, Aubrey de Grey presented a combination experiment: four age interventions (mTERT, rapamycin, senolytic, stem cell transplantation) added up to a greater life extension in mice than any single strategy. His proposal: eight or more parallel interventions in the future to combat aging synergistically—a bold reversal of the “one magic bullet” paradigm.
Frase AI: Content That Gets Discovered.
Create and optimize content with AI to drive traffic from Google, ChatGPT, and beyond.
