Key Takeaways
- Mark Zuckerberg predicts smart glasses will not feature a traditional app store; instead, models will generate interfaces on the fly inside the user's field of vision.
- Humans hit an input-output asymmetry: voice offers high-bandwidth output from human to machine, but audio feedback is too slow for consuming complex information compared to visual feeds.
- Alexandr Wang points out that consumer behavior is already multimodal: people talk, listen, and watch rather than interact through raw text boxes.
- Meta shipped HomeLink, a nickel-sized USB hardware dongle that connects the Muse agentic model to local hardware across a house.
The Death of the App Store
Smartphones trapped software inside discrete icons and approval queues. Every service fought for home screen real estate, built custom user interfaces, and paid a 30 percent tax to platform owners. Zuckerberg plans to erase that entire software model with smart glasses.
“I don't even know if we're going to have an app store on the glasses,” Zuckerberg told John Collison on Cheeky Pint. “It'll just be like a multimodal generation that kind of creates whatever you need right in your vision.”
When an interface generates dynamically on demand, static software binaries stop making sense. If you glance at a broken appliance, the glasses do not ask you to install a repair app from a store. The model looks through the camera, inspects the motor, retrieves the schematic, and renders an interactive visual repair path right over the machine. The concept of an app dissolves into context-driven generation. Founders building standalone mobile apps with rigid menus are optimizing for an interface model that is rapidly expiring.
Voice for Input, Vision for Output
Most builders treat voice assistants as audio-in, audio-out channels. You talk to Siri or Alexa, and the voice reads a paragraph back to you. Zuckerberg points out that this setup contradicts human biology.
Speaking your intent takes two seconds. Listening to an AI read out a four-paragraph answer takes forty seconds. That mismatch kills agent utility. The human eye processes visual information orders of magnitude faster than the ear absorbs spoken language. Smart glasses solve this asymmetry by pairing low-friction voice input with high-bandwidth visual projection. You speak a prompt, and the model projects a graphic, a table, or spatial arrows directly into your sightline. As Wang noted, “Most people in the world really experience the world in a highly multimodal way. They like talking to other people and listening to things and seeing videos and filming things of their own.” The winning interfaces will not rely on audio monologues.
Anchoring Ambient Models to the Physical World
Ambient intelligence falls apart if it cannot interact with physical objects. Alongside Meta Superintelligence Labs and frontier models like Muse Wonder, Meta quietly tested tiny physical bridge devices.
Zuckerberg revealed one recent example: “One of the gadgets that we just shipped was the little HomeLink. It's like the size of a little, like, nickel or something, but with a USB, right? So you can basically connect that, and then your Muse can connect to everything in your home.”
Hardware interfaces do not need to replace all existing household electronics. A tiny USB bridge allows an agentic model to control local appliances while the user walks around wearing sensory hardware. Glasses provide visual and auditory context: “The glasses basically will eventually just be a pure multimodal product,” Zuckerberg said. “You can let it hear what you are listening to, and it can talk to you, and it can see what you see if you let it.” Paired with physical anchors like HomeLink, the model connects live visual perception to local hardware action.
What to Do With This
Audit your product's user workflow this week. Strip out two clicks from your core user loop by accepting speech input, but return the result as a dense visual card rather than audio or text paragraphs. Test this interface with five customers to see if task completion speeds double.