🤖 NotebookLM adds Data Tables for structured analysis 🤖
Google expands NotebookLM's capabilities with structured data organization and export tools.

Google's NotebookLM has launched Data Tables, a new feature enabling users to organize and analyze information from their sources in structured tabular format. The tool automatically synthesizes uploaded sources into clean, organized tables that can be directly exported to Google Sheets for further analysis. The rollout is now available globally, making it immediately accessible for productivity workflows.

📱 Meta releases ExecuTorch for on-device AI 📱 
PyTorch-powered solution brings LLMs to Instagram, WhatsApp, and Ray-Ban glasses

Meta has officially launched ExecuTorch, an on-device AI deployment framework that currently powers AI features across Instagram, WhatsApp, Quest 3, and Ray-Ban Meta Smart Glasses. Built on PyTorch, ExecuTorch enables developers to deploy LLMs, vision, speech, and multimodal models with familiar APIs while maintaining privacy, performance, and portability.

 👩‍🎓 NotebookLM tests 30-minute AI lectures 👩‍🎓
Google's NotebookLM is testing a new "Lecture" format for its Audio Overviews

The new feature allows to generate comprehensive AI-powered lectures of approximately 30 minutes in length from any uploaded source, including Youtube videos. This rollout expands significantly from existing shorter audio summary formats, with multilingual support based on user settings. This experimental feature could position NotebookLM as the leading tool for education and professional development.

INTERESTING TO KNOW

🎁 OpenAI launches "Your Year with ChatGPT" recap 🎁

OpenAI is rolling out a personalized year-in-review feature similar to Spotify Wrapped, but for your AI chats. The recap analyzes your conversations and offers insights including stats, catchy graphics, awards, and even a custom image and poem based on your most-discussed topics. The feature is available for free, Plus, and Pro users who meet a minimum chat threshold and have Memory enabled, in selected countries.

📊 Karpathy reviews 2025's biggest AI shifts 📊

Andrej Karpathy's influential 2025 year-in-review, which amassed over 2.2 million views, pinpointed several key trends shaping the AI landscape. He highlighted Reinforcement Learning from Verifiable Rewards (RLVR) as a pivotal development, enabling AI to reason with greater autonomy. Karpathy also explored the concept of "jagged intelligence"—the tendency for AI to show exceptional skill in some areas while remaining surprisingly weak in others—and the emergence of "vibe coding," which is lowering the barrier to software development for non-coders. He concluded that Claude and Gemini have distinguished themselves from their competitors, citing the paradigm-shifting Gemini Nano Banana and the sophisticated agentic capabilities of Claude Code as evidence.