Beta and the actual release could be very different. It is almost impossible to ask volunteers to test every code commit before the actual release, and therefore a regression or bug could happen without being noticed.
A better way is that Wyze can monitor the key performance metrics of their app, like connection failures/latency, latency to retrieve live/recorded video, app response time, so that Wyze can detect regressions/bugs in an automatic fashion, and perform software rollback if a key performance metric regresses.
Wyze should also build such an automation to test their nightly builds, rather than solely relying on volunteer beta testing. Clearly, v3.0 is a failure because of multiple regressions found.