Meta is launching a new standalone Meta AI app powered by Llama 4, marking an initial move toward creating a more personalized AI experience. While Meta AI is already integrated daily across WhatsApp, Instagram, Facebook, and Messenger, this app focuses on voice-driven conversations, allowing users to engage with AI in a natural, seamless way, including multitasking with a visible microphone indicator for privacy. This first version aims to gather user feedback for ongoing improvement. Meta AI is designed to better understand users, providing more helpful, socially connected, and context-aware responses by leveraging data shared on Meta platforms like user profiles and content interactions. It supports voice or text commands, integrating features such as image generation and editing through conversational interaction. The app includes a full-duplex speech demo, offering a streamlined, natural voice experience generated directly by the AI, though it lacks real-time web access and may present occasional technical issues as it is experimental. Initially, voice features and the demo are available in the US, Canada, Australia, and New Zealand, with settings and mode management detailed in the help center. Using Llama 4, Meta AI assists with problem-solving, answering questions, and providing personalized recommendations drawn from web searches and users’ Meta data.

The AI remembers user preferences (e. g. , interests like travel or language learning) and can enhance personalized responses by accessing connected Facebook and Instagram accounts through the Accounts Center in the US and Canada. The app also features a Discover feed where users can explore and share AI prompts, with full control over what is posted. Meta AI is accessible across Meta’s products and devices—whether on Facebook, WhatsApp, Instagram, Messenger, or via Ray-Ban Meta glasses. Integrating the Meta AI app with the Meta View companion app for Ray-Ban glasses allows conversations to move seamlessly between the glasses, app, and web interface, though initiation of chats can only occur on glasses or app/web independently. Existing Meta View users will have their data and settings transferred automatically to the updated app. The web version of Meta AI is also being upgraded to include voice interaction and the Discover feed, optimized for desktop use with enhanced image generation featuring new style and lighting options. Meta is testing a rich document editor to create and export documents with text and images, plus importing documents for AI analysis in select regions. Voice interaction is emphasized as the most intuitive way to engage with Meta AI, with settings allowing users to enable a “Ready to talk” mode for hands-free access. Overall, Meta aims to deliver a more personal, social, and integrated AI assistant experience that respects user control and privacy.