AI Receptionist for VoIP Call Management

I need a simple, self-hosted AI receptionist that picks up every incoming VoIP call, greets the caller as if it were our front desk, and guides the conversation naturally.

Core flow
• The system asks how it can help.
• If the caller asks for a specific person, it checks a predefined employee list (name + direct number).
• It calls that employee’s number immediately. When the employee answers, it announces the caller’s name and reason for calling, then asks whether to accept or reject.
• Accept → the two parties are bridged.
• Reject or no answer → the caller is politely told the person is unavailable and asked for a message.
• Once the caller hangs up, the captured message is sent to the employee by both email and SMS.

Technical notes
• Must work over VoIP (SIP trunk or Twilio Voice are fine; I’m open to similar options).
• Outbound contact to employees happens by phone call only.
• Missed-call notifications go out through email and SMS.
• Please build the conversational logic with a framework you’re comfortable with—Dialogflow, Amazon Lex, Microsoft Bot Framework, or an LLM via an API gateway all work for me—so long as the speech recognition and text-to-speech quality are business-grade.
• I will provide the employee directory as a JSON or CSV file; the solution should let me update it easily.

Deliverables
1. Source code with clear setup instructions (Docker image or step-by-step guide).
2. Configurable dialogue scripts/prompts.
3. README covering how to add employees, change greetings, and connect new VoIP numbers.
4. Short video or live demo proving call flow end-to-end.

Please keep the solution lightweight—speed of deployment is more important than fancy UI. Let me know which stack you propose, how long you need, and any prior work on voice bots or telephony you can share.