A technically excellent call nobody would make

WebRTC voice between personal devices was feasible, reusable, and already half familiar—which did not give it a reason to exist.

The idea

Add live audio between trusted devices. The transport was interesting, prior WebRTC work reduced the unknowns, and it fit the family resemblance of the existing peer-to-peer tools.

Why it sounded good

The implementation had obvious technical shape: signaling, device identity, media tracks, reconnection, and a tidy self-hosted surface. Those are satisfying engineering problems.

What was actually needed

The use case split in two. Music streaming was already solved by Spotify and Jellyfin. Live voice calls were solved by tools people already had—and there was no recurring moment when calling one personal device from another felt useful.

Decision

Do not build it. Keep the transport knowledge; discard the feature.

Lesson

Reduced implementation risk is not increased product value. “I know how to build this” is evidence about cost, not demand.