Meta used its 11th annual Connect conference to present a broad set of updates across AI, mixed reality and wearables. The center of the announcement was Meta AI, the company’s assistant, which Mark Zuckerberg said is used by 400 million people every month, and 185 million on a weekly basis.
The message was clear: Meta wants its assistant to move beyond typed prompts and become part of everyday interactions across apps, glasses and future headsets. Voice, images, video translation and wearable controls are now all part of that direction.
Meta AI Gets A Voice Across Major Apps
The most direct change is voice input for Meta AI. The feature is coming to Facebook, Messenger, WhatsApp and Instagram, with an initial rollout in Australia, Canada, New Zealand and the USA. At first, it will work only in English.
That matters because voice input changes the way people use an assistant. Instead of treating Meta AI as a box for typed questions, users can speak to it inside the social and messaging apps they already use. For Meta, this makes the assistant feel less like a separate product and more like a layer running across its existing platforms.
Meta is also giving the assistant recognizable voices. Awkwafina, Dame Judi Dench and John Cena are among the well-known voices being used to give Meta AI more personality. The goal, based on the announcement, is to make interaction with the assistant feel more natural.
The scale also gives the update weight. An assistant used by 400 million people every month is not a small experiment. If voice input becomes a regular way to interact with Meta AI, the company could quickly learn what people expect from spoken AI inside social apps.
Images And Reels Become Part Of The AI Interface
Meta is also expanding AI-supported editing tools. Users will be able to ask questions about uploaded photos and use voice commands to add, remove or change elements in real photos. These capabilities rely on the vision abilities of the new Llama 3.2 models.
This is a shift from simple image generation toward AI that can work with a user’s own media. A person can start with a real photo, ask about it, and then request changes by voice. The source does not describe every editing option, but the direction is clear: Meta wants image understanding and image editing to be part of the same assistant experience.
Video is also part of the plan. Meta is experimenting with automatic video and lip-syncing for reels on Instagram and Facebook. The stated aim is to make content in different languages accessible to a wider audience.
That experiment could affect companies already focused on AI video translation and dubbing. The source specifically notes dedicated startups like HeyGen as a possible example. If automatic dubbing and lip-syncing become native features inside Instagram and Facebook, creators may have fewer reasons to leave those platforms for separate tools.
Smaller Llama Models Point To On-Device AI
Alongside the user-facing features, Meta has released smaller models with 1B and 3B parameters. These models are optimized for devices such as smartphones or future AR headsets.
That detail is important because many of the experiences Meta showed depend on AI being available close to the user. Smart glasses, phones and headsets all benefit from models that can run in device-focused environments. The source does not state exactly where each model will run, but it does connect the smaller models to smartphones and future AR headsets.
The broader picture is Meta trying to close the gap with commercial AI competitors such as OpenAI. The source says the performance gap to models like GPT-4 is shrinking, and points to multimodality and voice interfaces as useful features for ordinary users.
In plain terms, Meta is focusing on AI that can listen, see and respond in formats people already use. That includes voice messages, photos, reels and eventually wearable displays.
Ray-Ban Meta Smart Glasses Add Practical AI Features
The Ray-Ban Meta Smart Glasses are also getting more AI support. New updates include voice command reminders, QR code and phone number scanning, and a personal city tour in real-time.
The glasses will also soon translate Spanish, French and Italian in real-time into English. This connects the wearable strategy directly to Meta AI: the glasses are not only a camera or audio device, but a way to put the assistant into a person’s immediate surroundings.
Meta also highlighted a cooperation with Be My Eyes. Through that cooperation, users can take on the perspective of visually impaired people and describe their surroundings. The source presents this as part of the expanded support around the smart glasses.
The hardware collection is expanding too. Meta is adding new transition lenses and a limited edition of transparent Wayfarer frames. Those additions keep the product tied to familiar eyewear while the AI features become more capable.
Orion Shows The Longer Wearable Roadmap
One of the biggest Connect announcements was Orion, Meta’s first AR headset prototype. Meta says it has been working on the device for ten years.
The wireless headset weighs less than 100 grams. It includes ten customized silicon chips and bright displays with a wide field of view of 70 degrees. It supports voice control, hand tracking and eye tracking for navigating the user interface.
Meta also showed a new interaction method using an EMG wristband. The company described that option as "socially acceptable", suggesting that future AR controls need to work in public without feeling awkward or intrusive.
Orion is still a prototype, but Meta emphasized that it could be an end customer product. Even if this version never reaches the market, the company says an improved version could be developed more quickly. Zuckerberg also noted that the AI elements are ready for the market earlier than the actual hardware, something he had predicted a few years ago.
Taken together, the announcements show how Meta is connecting its assistant, social apps, smart glasses and AR research. Meta AI is no longer only a chatbot inside an app. It is becoming the software layer Meta wants to place across cameras, microphones, displays and everyday communication.