#002 · Voice-first CRM In progress
Cadence
Hold one button, talk for 20 seconds, and never lose a sale or an important meeting again.
How it works
from voice to next step
Push to talk→Saved on device→Upload with retry→Local transcription→LLM analysis→Your Briefing
01
Capture never fails
After a conversation you hold one button and talk. The audio lands on the device first and uploads with retry, so a bad signal never costs you a recording.
02
Transcribed on my own server
A self-hosted server transcribes the recording locally with faster-whisper. Your conversations don’t go to a transcription service.
03
The next step, decided for you
An LLM analyzes the transcript and produces the next step: who to call, what to send, what you promised.
04
The daily Briefing
Every morning you see your next steps, together with everything that was discussed, right on your screen.
05
Mobile and web
Follow-ups also happen from the PC, not only on the move, so Cadence works on both.
Stack
Flutter (mobile + web)Python / FastAPIfaster-whisperLLM integrationPostgreSQLSelf-hosted