Voice feedback: talk, don't type

Why voice beats a feedback form

Written feedback forms get answered by the people who are already angry or already delighted — and the answers are short, because typing is work. Speaking is roughly three times faster than typing, so people say the thing they'd never bother to write: the hesitation, the half-formed idea, the "I couldn't figure out where the button was." That texture is what tells us which tool to build next.

The other half is the follow-up. A static survey asks the same five questions no matter what you answer. A dynamic interview reads your answer first. Say "the score confused me" and the next question digs into which number and why, instead of blindly moving on to "rate us out of ten." That's how one comment turns into a fix we can actually ship.

How this interview works, step by step

1. You talk. Your browser captures the audio locally and packages it as a small WAV clip when you press stop.
2. We transcribe. The clip goes to a speech-to-text model and comes back as editable text — fix any word it mishears before moving on.
3. Ruby writes the next question. A language model reads the running transcript and produces one short, specific follow-up. Never the same question twice.
4. We summarize and send. At the end the transcript is condensed into three bullets — what worked, what frustrated you, and the single highest-value fix — and emailed straight to a human.

The three ways to give us feedback

What we do with what you tell us

Every tool on this site started as somebody's complaint. The background remover, the broken-link checker, and the HEIC converter each came from a single person describing a job they were stuck on. Interviews get read the same week, turned into a short build list, and the fixes show up on What's New. We never publish your name or words without asking, and your email is never added to a list — the full details are in our privacy policy.