Alice is an AI-powered recording and transcription tool. Its users — professionals, researchers, students — rely on it for meetings, interviews, and long-form recordings where losing content isn't just inconvenient, it's costly.
When the Alice team embedded Voicebox directly into the product, the intent was straightforward: make it easy for users to report problems as they happen, in their own words, without friction. No forms. No support tickets. Just talk.
"We're not guessing what our users want anymore. We're hearing it from them, in their own words, while the frustration is still fresh."
Over four to five months, Voicebox captured a steady stream of voice messages from Alice users. The messages covered a wide range of issue types — giving the team an unusually clear picture of where the product was working, and where it wasn't.
The messages clustered naturally into four categories — each representing a different kind of product intelligence.
Users flagged data integrity issues in real time — often before the team detected them internally.
"It's supposed to be 1 hour and 45 minutes long... it's only giving me the transcript for 47 minutes. I want all of that retrieved."
Time-sensitive outages surfaced immediately, with full context about what the user was trying to do.
"My app is not working today and I can't access or record any of my meetings that I desperately need."
Navigation changes that confused users were visible within days of a release — not weeks.
"I am having a harder time navigating your new program... how can I download a recording without the transcript?"
Unprompted feature ideas arrived with natural-language context no survey would have surfaced.
"It would be helpful to split a quote between two speakers — sometimes their sentences get merged together."
The team attributes the quality of signal to one thing: voice removes the friction of writing. A user who wouldn't open a support ticket will talk. A user who wouldn't describe a bug in text will explain it out loud — with context, with emotion, and with the kind of specificity that's actionable.
Because Voicebox is browser-based and embedded directly in the app, users could report issues at the exact moment they occurred. The team received messages mid-session, mid-recording, and mid-frustration — which meant context was complete, not reconstructed after the fact.
The VBX AI team didn't just build Voicebox — they use it. Embedding it into Alice before recommending it to customers was a deliberate choice: if it wasn't going to work for their own product, it wasn't ready for anyone else's.
Four months in, the signal has been consistent. Users talk. The product team listens. And the gap between what customers experience and what the team knows has narrowed considerably.
Voicebox doesn't replace your support team. It tells you what your support team was never going to hear.
support@vbx.ai
© 2026 VBX AI, Inc.
