The future of work shouldn’t mean more screens competing for your attention. Teleprompt while staying present. Live subtitles without breaking eye contact. Key information when you need it, without stopping the conversation. That’s the idea behind Even G2: technology that works with your attention, not against it. #EvenG2# #Conversate# #SmartGlasses# #FutureOfWork#

Even G2 v2.2.9 is here. What’s new: • Expanded Contextual Menu controls while features are running • Run features with the display off • End features directly through Even AI • Teleprompt now flags unsupported characters • Export health data from Health Settings • Attach screenshots + videos when reporting issues • Even Hub apps can now integrate with Contextual Menu Available now on the App Store & Google Play.

Controlling my AI agent while riding through the city 👓 With Terminal Mode on the Even G2 glasses, I can keep an eye on my Codex agent running back home, check progress, and even dictate new commands on the go. My computer stays at home. The project keeps moving wherever I go. Learn more: @EvenRealities #CodeInTheWild#

CodeInTheWild is now open! Show the most unexpected real-world way you use Even G2 Terminal Mode to supervise an AI coding agent - and compete for 10 years of Codex Pro 5x. How to enter: 1. Record with the Even Realities' app Terminal mode capture feature 2. Post on X with @EvenRealities & include CodeInTheWild Winners are ranked primarily by public likes.

I'm not sure how useful this is but it certainly would have been useful for me last year... If a chrome trace has events that overlap on the same stream (e.g. event1 ends at 10.1 and event2 begins at 10.05), perfetto's behavior is to not show the events and show an empty gap in your trace >:(

🤖 Turns out robot dexterity scales with data too, and now there's hard evidence for it. Title: GE-Act 2.0: Pretraining and Scaling a World-Action Model for Robotic Manipulation URL: 🧩 Overview AgiBot Research Team built GE-Act 2.0, a world-action model that predicts future visual states to guide robot actions, training every component from scratch on manipulation data instead of inheriting a pretrained video generator. ⚙️ Problem it solves Prior systems reused reconstruction-optimized video generators that don't guarantee action-relevant information, and it was unclear how to pretrain visual generation and inverse dynamics separately before connecting them. 🛠 Methodology Three pieces make it work: a 64x-compressed control-oriented autoencoder (CoAE), a Single-Step Visual Planner that generates a full future frame in one forward pass, and KASO, which fixes the "validity gap" between predicted futures and real actions via top-k selection. 📊 Results Scaling training data from 300 to 30,000 hours lifted the G1-OP robot's success rate from 17.1% to 44.1%, a 27.0-point gain. Even G2-90D, with under 2% of the data, gained 17.7 points, showing real cross-embodiment transfer. Object, color, and position grounding all exceed 90% accuracy. #Robotics# #WorldModels#