🌿 OpenAI Backs AI Health Coach

PLUS: Odyssey Hollywood-Grade AI FX

Welcome back!

OpenAI is funding a new startup, Thrive AI Health, to develop an AI-powered health coach. This digital assistant will be trained on peer-reviewed research and user data. Will this become the next big thing in health tech? Let’s unpack...

Today’s Summary:

  • OpenAI funds AI health coach

  • Odyssey develops AI visual FX

  • Humanoid robots with AI teleoperation

  • Microsoft OpenAI access in China

  • DeepMind maps mouse brains

  • OpenAI fixes ChatGPT Desktop App security flaw

  • 2 new tools


OpenAI funds an AI Health Coach

The Summary: OpenAI Startup Fund and Thrive Global have launched a new venture, Thrive AI Health, to develop an AI-powered health coach. This digital assistant will offer personalized recommendations on sleep, nutrition, fitness, stress management, and social connections.

Trained on peer-reviewed scientific research and user data, the AI coach aims to make behavior change more accessible. The project brings together OpenAI CEO Sam Altman and wellness advocate Arianna Huffington.

Key details:

  • DeCarlos Love, former Google executive, appointed as CEO

  • The AI coach will be available as a mobile app 

  • Partnerships established with Stanford Medicine and other academic institutions

  • The platform will use biometric data and daily habits to provide tailored advice

Why it matters: This collaboration is a new step in applying AI to personal health management. By offering individualized guidance, the AI coach could help users improve their daily habits and prevent chronic diseases. The project also highlights the growing trend of tech companies entering healthcare, promising personalized health advice that is more accessible and affordable. However, it also raises questions about data privacy and the role of AI in healthcare decisions.


Odyssey working on "Hollywood-grade" AI visual FX, trained on real world 3D data

The Summary: Odyssey has emerged from stealth mode with $9M in seed funding. Founded by self-driving car veterans Oliver Cameron and Jeff Hawke, the company aims to create AI-powered video tools for professional filmmakers.

Odyssey technology aims to offer more control and consistency than current video models, targeting Hollywood-grade visual effects. The startup competes with OpenAI's Sora and Runway's Gen 3, focusing on Generative 3D with fully editable layers and longer frame sequences.

Source: Odyssey

Key details:

  • Odyssey AI combines four specialized models for 3D rendering, material effects, motion, and lighting

  • The company aims to reduce film production time and costs dramatically

  • Odyssey gathers its own 3D datasets in the real world to train its models

  • The team includes AI researchers from Cruise, Wayve, Waymo, Tesla, and Meta, as well as artists from major film productions

Why it matters: Odyssey's approach could transform filmmaking by making high-quality controllable 3D visual effects accessible and affordable. This technology has the potential to disrupt traditional studio systems and empower independent creators. As AI continues to advance in the entertainment industry, it raises new questions about the future of film production, visual effects, and creativity in moviemaking.


Humanoid robots get human touch with AI-Powered Teleoperation

The Summary: Researchers at MIT and UCSD have developed Open-TeleVision, an immersive teleoperation system for robots that combines human control with AI.

It allows operators to control humanoid robots remotely using VR headsets, providing stereo vision and intuitive control of the robot's head, arms and hands. The system enables precise manipulation tasks from thousands of miles away, leveraging AI for imitation learning.

Key details:

  • Uses VR headsets to stream stereo video from robot's perspective

  • Allows control of robot's head movements for active perception

  • Enables remote operation over long distances

  • Tested on two humanoid robot platforms: Unitree H1 and Fourier GR-1

  • Incorporates AI-powered imitation learning for task replication

  • Uses a transformer-based visual backbone (DinoV2)

  • Open-source implementation available on Github

Why it matters: This system bridges the gap between human intelligence and AI-enhanced robot capabilities. It allows human operators to merge their skills and dexterity with robots, while AI algorithms learn from these demonstrations. This combination enables adaptable and intuitive robotic manipulation. This could accelerate deployment of intelligent robots in complex environments like disaster response or telesurgery.


Quick news


🥇 New tools

  • Ariglad - Auto create & update your help center with AI

  • Doti - AI copilot for health and habit tracking (iOS)

That’s all for today!

If you liked the newsletter, share it with your friends and colleagues by sending them this link: https://thesummary.ai/