Skip to main content
Back to Pulse
TechCrunch

Memories AI is building the visual memory layer for wearables and robotics

Read the full articleMemories AI is building the visual memory layer for wearables and robotics on TechCrunch

What Happened

Memories.ai is building a large visual memory model that can index and retrieve video-recorded memories for physical AI.

Our Take

This one's actually interesting. Robots and wearables *need* visual memory—if your robot can't remember what it saw 10 seconds ago, it's useless. Same for AR glasses. Building a visual memory model that runs on low-power hardware is hard.

If Memories.ai's cracked the efficiency problem (can you run this on a Snapdragon?), they've got something real. If it requires cloud offload, it's just another API wrapper.

The bet's solid—physical AI's going to need embedded memory. The execution risk is huge, but the problem is real enough that they might pull it off.

What To Do

Find out their model size, latency, and power draw—if those numbers are reasonable for-device inference, it's worth watching.

Cited By

React

Loading comments...