I wanted to know more about current web speech recognition APIs. I tried three different coding agents to set up a proof of concept app for me. To my surprise,

I wanted to know more about current web speech recognition APIs. I tried three different coding agents to set up a proof of concept app for me. To my surprise, Cursor's composer model beat both Claude and Gemini.