Project Astra is a research initiative by Google DeepMind, aimed at developing a universal AI assistant capable of understanding and interacting with the world around its users.
DeepMind is an AI research lab and a subsidiary of Alphabet Inc., Google's parent company.
Project Astra leverages advanced multimodal AI technologies, which allow it to process both voice and visual inputs, giving users a more intuitive AI assistant.
This research prototype is part of Google’s broader vision to create AI systems that can adapt to various environments and contexts to enhance user interactions in everyday life.
- Project Astra is a research prototype by Google DeepMind, aiming to create a universal AI assistant that understands and interacts with the world using advanced multimodal AI.
- As of March 2025, it’s still in development with no public release date or cost, but some features are rolling out in Gemini Live later this month for subscribers.
- It’s accessible on Android phones and prototype glasses, with glasses in testing by a small group; a waitlist is open for potential testers.
- Key features include natural voice and visual interaction, memory of past conversations, integration with Google services, and new live video/screen sharing in Gemini Live, including Android devices.
- From what we've seen, it seems likely that Project Astra will evolve into broader Google product integration.
Key points
Project Astra accessibility
As of March 2025, Project Astra remains in the research prototype phase and has not been released to the general public. There is no announced release date or cost for the full Project Astra experience. However, recent developments including some of its capabilities are being integrated into Google’s existing products.
At the Mobile World Congress (MWC) 2025, Google announced that Project Astra-powered features, such as live video and screen sharing, will roll out in Gemini Live later this month.
These features are currently available to Gemini Advanced subscribers and those with a Google One AI Premium plan on Android devices, marking a significant step toward expanding accessibility to a broader audience.
This integration suggests that while the Project Astra system is still under development, its technology is already rolling out in an effort to give users accces to enhanced AI offerings.
There is currently no public release date for Project Astra as it is currently in ongoing research status, but we can expect future announcements as testing progresses.
Supported devices
Project Astra is designed to be accessible through multiple devices, primarily Android devices and prototype glasses.
The Android device accessibility allows users to interact with the assistant by speaking or using the phone’s camera to show what they see, enabling real-time conversations.
The prototype glasses, will offer users an immersive experience by allowing Project Astra to see the world as the user does, enhancing contextual understanding. However, the prototype glasses are not yet available to the public.
As of December 2024, a small group of testers began trying Project Astra on prototype glasses, with plans to expand this testing. These glasses are part of Google’s long-term vision for augmented reality (AR) and multimodal AI, but they are not yet available for purchase.
Google has opened a waitlist for potential testers, where individuals can share details to join.
Project Astra key features
Project Astra’s key features are designed to provide a natural and context-aware assistance experience. These include:
- Natural interaction: Users can interact with Project Astra through voice commands or by showing visual inputs, such as pointing the phone camera at objects. This multimodal approach enables free-flowing, real-time conversations, making interactions feel intuitive and seamless
- Memory: Project Astra can remember key details from past conversations, refining its responses based on context. It maintains up to 10 minutes of in-session memory and can recall previous interactions, enhancing personalization. For example, it can remember a user’s marathon training regime or dietary preferences, as demonstrated in tester feedback
- Tool integration: The assistant integrates with Google services such as Search, Maps, and Lens to inform its answers, providing comprehensive assistance for everyday tasks. This integration allows it to pull real-time information, such as identifying locations or providing directions, enhancing its utility
- Live video and screen sharing: Users can share their screen with Gemini Live to ask questions about what’s displayed or use a live video feed for real-time assistance.
Project Astra outlook
Project Astra’s outlook indicates a transition from a research prototype to broader integration within Google’s ecosystem.
While it is still in the testing phase with a limited number of private testers, the recent rollout of features in Gemini Live suggests that its technology is being adopted to enhance existing AI products. This integration is expected to expand, potentially making Project Astra’s capabilities available to a wider audience in the coming years.
The prototype glasses, represent an exciting frontier for AR and AI, with potential applications in areas like language translation, location memory, and real-time object recognition. However, specific product release dates for these glasses have not been announced, and they remain a technology demonstration at this stage.
Given Google’s aggressive roadmap for AI in 2025, as seen in recent updates, it seems likely that Project Astra will play a significant role in redefining how we interact with technology.
Final thoughts
Project Astra embodies Google’s vision for the future of AI assistants, focusing on natural, context-aware interactions and seamless integration into daily life.
As it transitions from a research prototype to integrated features in products like Gemini Live, it holds the potential to redefine how we interact with technology.
With ongoing testing, Project Astra is poised for significant developments in 2025 and beyond.
More from Tom's Guide
- ChatGPT's scheduled task feature is a game changer — 5 prompts to try first
- I have ChatGPT Plus — but here's 7 reasons why I use DeepSeek instead
- iOS 18.5 is coming soon with huge Siri upgrades — here’s everything to expect
You must confirm your public display name before commenting
Please logout and then login again, you will then be prompted to enter your display name.
