Or have an a11y standard for MCPs, where they can't show UI elements and have to only respond with text so that Voice Readers could work out of the box. This w

Or have an a11y standard for MCPs, where they can't show UI elements and have to only respond with text so that Voice Readers could work out of the box. This would be a game changer, currenly Voice Readers do not work very well with websites and a11y is a clunky set of tags that you provide to elements and users need to move around elements with back/tab and try to make a mental model of what the website looks like. With MCP and Voice chat, it is like talking to a person.