Designing for AI Means Designing Beyond Screens

Something shifted in the last few weeks, and it did not happen in a keynote. It happened in the small print of product updates. Cameras in your ears. AI in your TV whether you asked or not. A chatbot that reshapes your news feed when you tell it to. The place where people first touch your product is moving, and it is moving fast. Let me catch you up.

The deep cut

• Voice and vision are becoming the front door. Apple's AirPods and Amazon's Alexa+ both aim to replace the app you open with a thing you talk to.
• Free is a distribution play, not a gift. Amazon dropped the $19.99 Alexa+ fee on Fire TV to put AI in front of everyone at once.
• A leaked demo is a shipped promise. Apple's "your world becomes savable" video means the camera AirPods are close enough to plan for.

The device that never leaves your ears

Apple is close to shipping AirPods with cameras. A demo video buried in the macOS 26.7 beta shows a man holding up a book while a voiceover says, "With Visual Intelligence, your world becomes savable". The cameras act as eyes for Siri. Point your head at something, ask a question, save it. No phone in hand.

This is a real change in where your product lives. If the entry point moves from a screen to a voice in someone's ear, the tap and the swipe stop mattering. What matters is whether your product can answer a spoken question. The AirPods reportedly capture only low-resolution visuals and cannot record video, which is Apple's way of dodging the "pervert glasses" fight Meta walked into. Note the tradeoff: discreet is exactly what creeps people out.

Free is the whole strategy

Amazon just made Alexa+ free on all Fire TV devices, no Prime needed. It used to cost $19.99 a month. Now everyone gets upgraded automatically, no app to download, no signup. The price went to zero because the goal was never the fee. The goal is to make talking to your TV the default way you find something to watch.

Amazon says Alexa+ users have nearly twice as many conversations on Fire TV as they did with old Alexa. Read that carefully. They are counting conversations, not satisfaction. Google did the same thing pushing Gemini onto Google TV. When a company hands you a metric like "time spent," ask what it is standing in for. More talking is not proof people wanted to talk.

The feed that answers to a prompt

Google is adding a chatbot to the Discover feed. You tap the three-dot menu and tell it what you want, like "show me kitchen renovation tips," and it reshapes the feed and remembers for next time. Meta shipped a Mac app that watches your screen and pulls from Instagram, Facebook, and Google Workspace to draft decks and weekly reports.

Both moves point the same direction. The user stops browsing and starts asking. If your content or product shows up inside someone else's feed, an AI now decides whether you make the cut based on a spoken sentence. Google's refreshed Preferred Sources button is the one lever publishers get. That should tell you how little control is left on your side.

What this means for your roadmap

Your product has entry points you did not design and cannot see. Someone asks Siri, Alexa, or Gemini a question, and an assistant answers instead of your interface. If the answer is good, the user never opens your app. If the answer is wrong, you eat the blame without ever being in the room.

So the work is not building your own AirPods. It is making sure your product is legible to a voice assistant and a prompt-driven feed. Can it be found, described, and acted on without a screen? That is the question these updates are quietly forcing, and it belongs on your roadmap before the hardware ships in September.

Three questions for your team

• If a user reaches our product by asking an assistant instead of opening our app, does the experience still work, and who owns that path today?
• We measure engagement. Are we counting conversations and time spent the way Amazon is, or do we know whether people actually got what they wanted?
• When our content lives inside Google Discover or a Meta app, what is the one lever we control, and are we using it?