Trying to build my own version of retell for personal use. Using twilio numbers some sort of llm.
Key requirements
Help calculate estimated monthly cost of running this system.
Must be significantly cheaper then retell or vapi
Must be able to make outbound calls from leads On a csv.
Use client information in the conversation such as name address ect.
Ask 5 qualifying question then proceed to transfer to human.
Sound as realistic as possible.
Rotate through available DID and select based on area code or state.
Voicemail detection to avoid unnecessary call lengths.
Has a simple front end I can upload csv start stop workflow and Resume where it left off
My understanding of this is that we would use
• Use Twilio Media Streams → WebSocket server (Node/Express).
• Stream audio into STT + LLM + TTS in real time.
Does this sound correct?
Need this built ASAP.