Why 'Voice Everywhere' Is Bad Design and How to Use Voice Contextually

A design analysis published on DEV Community argues that enabling voice assistants across all pages and contexts is a fundamental UX mistake. The piece, drawing on production experience at Scenaro, explains that voice is cognitively demanding and unsuitable for tasks like comparing spec sheets or for use in public spaces such as offices and transit. Instead, the author proposes four context-driven modes — full duplex, discreet, push-to-talk, and text-to-audio — each activated based on the user's environment and task. Real-world products like Urbansider, a travel concierge app, already ship a discreet mode that responds in text to avoid forcing users to speak aloud in public. The core argument is that voice should be a contextual design policy, not a blanket toggle, and that transitions between modes should be defined at the scenario level rather than the page level.