Pixel Watch 5 Wins by Pointing a Camera at Dumbbells. The AI Does the Thinking. Obviously.
Google's Pixel Watch 5 ships with marginal hardware bumps over its predecessor: extra RAM, double the storage, and a slightly faster processor that makes on-device Gemini responses feel less glacial. The genuinely interesting feature is Coach, which lets you photograph your available weights or name a target muscle group, and it builds a workout accordingly. This launches in the coming weeks and will also appear on the Pixel Watch 4 through software updates.
This is a textbook case of proactive AI replacing procedural input. The mental model is context-to-action compression: instead of the user selecting exercises, sets, and reps through menus, the system ingests physical context via camera and produces a structured plan. The mechanism is multimodal inference running locally, which is why the RAM bump matters. When AI stops waiting for instructions and starts reading the room, the interface shrinks.
Google shipped the Pixel Watch 5 with integrated Gemini AI and a Coach feature built around image recognition and natural language input. PCMag reviewed the device and concluded it is the best smartwatch for Android users, despite only minor hardware changes.
- If you own a Pixel Watch 4, wait for the upcoming software update that brings Coach to your device. No purchase required. The feature is expected within weeks.
- Open the Coach feature when it arrives and try the two input modes. Take a photo of whatever weights you have at home. Then try typing or saying a muscle group like 'shoulders' instead. Compare the workouts it generates.
- Run one of the suggested sessions and note whether the sets, reps, and exercise selection actually match the equipment you photographed. That gap between expectation and output is where you learn what current multimodal AI can and cannot do.