Ray Fernando connects a day meeting a friend in San Francisco with a practical lesson for software builders: get out of the room and observe how people use what you makeAn AI feedback loop occurs when AI outputs or their consequences become inputs that influence later model or agent behavior.. A request for larger transcription text revealed uses he had not anticipated, such as following conversation in noisy settings or recalling a word that was difficult to hear.
Fernando then describes plans for SayRay, his transcription prototypeAutomatic speech recognition converts spoken language into text or structured linguistic output using signal processing and machine learning.. Its current code communicates with an external server, and he intends to ask an AI agent to simplify or rewrite that architecture so the app can work on-device without an internet connectionOn-device inference runs a trained model on the user's device rather than relying entirely on a remote model service.. This is a statement of his next development step, not a demonstration that offline processing or an App Store release is complete.
Watch on YouTube




