Ray Fernando describes how repeatedly expanding the scope of his transcription prototype has delayed its release. Requests such as larger caption text expose practical needs, but trying to accommodate every possible use before launch can turn a small product into an unfinished project.
Ray Fernando uses SayRay to distinguish what works now from what he wants to build next. Translation currently requires an internet connection; he hopes newer on-device models will simplify that architecture. His main lesson is to release a focused version, learn from use and avoid treating planned AI capabilities as finished features.
Watch on YouTube
