Google DeepMind
Try Google AI Studio Try Gemini

Project Astra

A research prototype exploring breakthrough capabilities for Google products — on the way to building a universal AI assistant

Join trusted tester waitlist

We’re working to bring Project Astra’s capabilities to Gemini Live, new experiences in Search, as well as new form factors like glasses.

Enhancing Gemini Live

Some of the latest features in Gemini Live were first explored using Project Astra.

Over the past year, we’ve been integrating capabilities like screen sharing and video understanding into Gemini Live for more people to experience.

Try Gemini Live

Discover what’s possible

Natural interaction

Improved audio input and output for smooth communication across languages — without interruptions.

Native audio dialogue

Detects different accents, languages and emotions, and gives fluid responses in 24 languages.

Proactive responses

Project Astra can intuitively start conversations, and respond in the moment — without interrupting or time lag.

Context aware dialogue

Ignores distractions, like background conversation and irrelevant speech.

Action intelligence

Helps users by reacting to their needs and taking actions on their behalf.

Agent highlighting

Understands objects in context, using on-screen highlights to show you what’s important.

Tool use

Uses tools like Search, Gmail, Calendar Maps, and interface control to complete tasks for you.

Intelligent personalization

Project Astra learns and retains user preferences to give personalized answers — and can explain the thinking that guides its answers.

Personalized reasoning

Uses deep reasoning and memory to build a rich understanding of your preferences, and tailor shopping recommendations to your needs.

Content retrieval

Retrieves content — like a PDF user manual or a recipe you’ve shared — to provide tailored help for tasks.

Multimodal memory

Integrates different types of data to perform tasks, and remembers key details from past interactions.

Supporting the blind and low-vision community

We’re developing a version of Project Astra in collaboration with the blind and low-vision community.

Project Astra’s Visual Interpreter research prototype is able to understand objects and unfamiliar spaces.

  • It reacts to changes in the visual world, like describing what it sees as the camera view moves.

  • The prototype also works with Google products like Maps, Photos and Lens to accurately identify objects and the environment they appear in.

We’ve partnered with visual interpreting service Aira, to develop and refine Project Astra for blind and low-vision users. Aira users and interpreters helped us tailor the prototype for their community.

Now we’re offering a Trusted Tester program for early access to the prototype — with live, professional, human oversight. If you’d benefit from this experience, sign up for the waitlist.

Join the waitlist

Meet Dorsey

Dorsey Parker is a musician. He also has 8% vision, and is losing the sight he has.

Dorsey’s using Project Astra to adapt to this change in his life. Using his phone, Project Astra describes his surroundings, and accesses apps like Lens and Maps to help him explore new places.

Assistance across devices

Project Astra works on Android phones and prototype glasses. Cross-device memory means you can switch devices and carry on the same conversation.

  • Mobile

    Point your phone camera at what you’re interested in, and start a conversation. Or share your screen to unlock a new level of interactive assistance.

  • Glasses

    Project Astra integrates seamlessly with prototype glasses to see the world as you see it — for an even more immersive and helpful experience.

Building responsibly in the agentic era

We recognize the responsibility it entails to develop these new technologies, and aim to prioritize safety and security in all our efforts.

Learn more

Experience Project Astra

Project Astra is a research prototype, being used and refined by a limited number of trusted testers. If you're interested in becoming a tester, please share a few details to join the waitlist.

Join the trusted tester waitlist

Acknowledgements

This work was made possible by the contributions of: A Theeraphol, Aaron Donsbach, Aaron Schneider, Ace Kulshrestha, Adam Shafi, Adrian Bolton, Agustin Dal Lago, Alejandra Molina, Alessio Tonioni, Alex Balabanov, Alex Chen, Alex David Norton, Alex Greve, Alexandre Moufarek, Alistair Muldal, Allison Garcia, Ana Hosseini, Anand Rao, Andrea Colaco, Andrew Rhee, Andrew Rosenberg, Andy Bodart, Andy Chen, Angad Chandorkar, Angela Sun, Angelica Schwartz, Anirudh Baddepudi, Ankesh Anand, Anna Galusza, Antonia Paterson, Anu Sinha, Anurag Ajay, Anurag Kumar, Aparna Joshi, Artem Dementyev, Arun Ahuja, Arun Narayanan, Ashyana Kachra, Aveek Purohit, Bakary Diarrassouba, Benedict Odai, Bethanie Brownfield, Bhanu Guda, Bibo Xu, Bo Li, Bob Muniz, Boon Panichprecha, Brendan Jou, Brian Patton, Caitlin Sikora, Canoee Liu, Cecilia Abadie, Charles Yoon, Charn Kobkuachaiyapong, Chen Cai, Chen Liu, Chen Sun, Chen Yan, Chris Shaheen, Christina Butterfield, Christopher Diamond-Jones, Christopher Haire, Cindy Fan, CJ Zheng, Claudiu Musat, Clement Farabet Dana Kurniawan, Cody Wild, Connie Huang, Dan Motzenbecker, Dan Walker, Daniel Padgett, Darren Carikas, David Wood, Dawn Bloxwich, Dayson Pais, Deepali Jain, DeLesley Hutchins, Demetra Brady, Derek Pang, Dev Valladares, Diane Wan, Dimple Vijaykumar, Dome Seelapun, Don Barnett, Duncan Williams, Ed Chi, Elahe Dabir, Elisabeth Chauncey, Emanuel Taropa, Emily Kaplan, Eric Lai, Evalyne Muiruri, Evan Hardesty Parker, Evan Palmer, Evan Senter, Ewa Dominowska, Fadi Biadsy, Federico Carnevale, Federico Tombari, Fedir Zubach, Fei Zheng, Ferjad Naeem, Florian Dubost, Frankie Garcia, Fred Alcober, Gaby Pearl, Gan Song, Gary Vondran, Genzhi Ye, Georgi Stephanov, Golan Pundak, Grant Yoshida, Gregory Wayne, Hadas Natalie Vogel, Haiguang Li, Haokai Lu, Haoting Wang, Haoting Wang, Haozhen Wu, Harrison Lee, Harsh Lara, Hsiu Wang, Igor Mordatch, Ik Aiempongpaiboon, Itay Inbar, Jack Huo, Jake Varley, Jason Lin, JD Chen, Jeff Gray, Jenny Brennan, Jens Heitkaemper, Jesper Andersen, Jiaxi Tang, Jihan Li, Jilin Chen, Jilin Tu, Joe Richey, Joe Stanton, Johann Rocholl, John Penington, Johnny Lee, Jon Blanton, Jon Buckley, Jonas Degrave, Jonathan Fildes, Jordan Griffith, Joseph Chen, Julian Salazar, Julie Anne Seguin, Julie Sohn, Juliette Love, Junwen Bai, Justin Snyder, Justin Taft, Juston Payne, Kai Zhou, Kandarp Joshi, Keerthana Kumar, Ken Mixter, Keran Rong, Khe Chai Sim, Kim Nomrak, Kitty Stacpoole, Komal Jalan, Krishna Shastry, Krishna Somandepalli, Kristen Olson, Kyle Kastner, Laurent Simon, Lavisha Aggarwal, LeHan Meng, Lei Shi, Leslie Baker, Liangliang Cao, Liangzhe Yuan, Lichan Hong, Lihao Liang, Lin Li, Liviu Panait, Lucia Rossazza, Lucy Abramyan, Lucy Kim, Lukas Hoyer, Luke Friedman, Lynn Nguyen, Maciej Kula, Mahesh Ramachandran, Manish Jain, Marc Seger, Mark Lieu, Mark White, Marke Hallowell, Mathangi Venkatesan, Matt Barnes, Maxine Deines, Meethu Malu, Mehdi Mirza, Mei Lu, Meiqi Guo, Melissa Ferrari, Michael Han, Michael Severino, Michael Siliski, Michal Lahav, Michal Prussak, Mike Dory, Mikel Rodriguez, Mingqiu Wang, Misha Bilenko, Misha Dashevskiy, Naina Raisinghani, Nathalie Rauschmayr, Nathan Howard, Nathan Waters, Nathaniel Fruchter, Navaneeta Dorbala, Neel Joshi, Neil Chen, Nicholas Ambur, Nicolas Serrano, Nikhil Siddhartha, Nikita Kodali, Nikolai Grigorev, Nimer Jaber, Nishtha Bhatia, Noah Murad, Omar Estrada, Oscar Chang, Oskar Bunyan, Palash Nandy, Pankil Botadra, Patrik Sundberg, Paul Pucciarelli, Paul Rohde, Pavel Dubov, Pengfei Xing, Peter Chen, Piermaria Mendolicchio, Piotr Swigon, Pooja Agarwal, Prakash Shroff, Pramod Gupta, Pranav Khaitan, Praveen Srinivasan, Pu-Chin Chen, Qiao Yang, Raghav Gupta, Ran Xin, Reed Roberts, Remi Crocker, Renhao Liu, RJ Skerry-Ryan, Rob Martin, Rob Willoughby, Ronny Huang, Ross West, Ruihan Zhang, Sadh MNM Khan, Sadhana Narasimhamurthy, Sage Stevens, Saket Joshi, Sam Castle, Sam Lawton, Sami Lachgar, Scott Adams, Sean Csukas, Seth Welna, Shaan Bijwadia, Shashwath Santosh, Shefali Garg, Shishir Ayalasomayajula, Shuo-yiin Chang, Siggi Orn, Silvano Bonacina, Simral Chaudhary, Sofía Sánchez-Zárate, Srinivas Sunkara, Stav Ginzburg, Stephanie Debats, Stephanie Zhang, Steven M. Hernandez, Stiven Morvan, Susan Xu, Sushil Nath, Suz Chambers, Tanay Sharma, Tanuj Bhatia, Tanya Kraljic, Tara Sainath, Tara Thomas, Taylor Bos, Terry Tai, Tim McConnell, Timothee Cour, Tom Ouyang, Tony (Tuấn) Nguyễn, Tony Bruguier, Tony Lu, Toshiyuki Fukuzawa, Uli Rueckert, Václav Novák, Vahid Ghadakchi, Valeria Wu, Vikas Bahirwani, Vikas Sindhwani, Virat Shejwalkar, Vivek Kesarwani, Warren Morningstar, Wei Huang, Will Truong, Wing Wat, Woohyun Han, Xi Xiong, Xi Yi, Xiaotong Yang, Xingyu Cai, Xinyang Yi, Yana Lunts, Yanzhang (Ryan) He, Yao Fu, Yaqing Wang, Yiling Huang, Yufei Wang, Zack Ontiveros, Zeina Oweis, Zhouyuan Huo, Zilla Gunawardana, Zoe Mongan.

We also acknowledge the many other individuals who contributed across Google DeepMind and our partners at Google.