Build AI Phone Receptionist with Twilio + OpenAI (Must Be Done by Friday – $1500 Budget)
I’m building a SaaS business called MySecretary, a fully automated virtual receptionist for independent contractors (plumbers, electricians, HVAC, etc.).
When a customer calls a contractor and they don’t answer, the call is forwarded to MySecretary. The AI then answers on behalf of the contractor’s business, takes down the customer’s message, and optionally books appointments or sends follow-ups.
The Project
I need a production-ready AI receptionist backend built using:
Twilio (to receive and handle calls)
OpenAI (to generate responses using ChatGPT)
Google Sheets or a small backend DB (to look up contractor details like business name)
Flask or FastAPI, hosted on Render (or your recommendation)
Fallback response if AI or lookup fails
Error-proof: it must never crash during a call
You’ll need to:
Build and deploy the entire backend
Connect it to Twilio
Pull contractor info from a Google Sheet or small DB
Use that data to customize AI replies (e.g., “Hi, this is Aspen Plumbing. How can I help you today?”)
Return valid TwiML to Twilio
Format phone numbers reliably
Log calls (to a sheet or text log)
Send AI replies to a Google Sheet (Pro only)
Plans to Support
You will be building support for two service tiers:
✅ Starter Plan (Basic Receptionist)
Customer calls the contractor
If no answer, call is forwarded to MySecretary (Twilio number)
MySecretary answers using the contractor’s business name
AI says something like: “Hi, this is Aspen Plumbing. How can I help you?”
AI listens and generates a short response (basic conversation)
Logs the interaction to a Google Sheet
Pro Plan (Full AI Secretary)
Everything in Starter, plus:
AI also sends follow-up message summary to the contractor’s email or phone
Optional integration with Google Calendar or Calendly to book appointments
Conversation log (or AI-generated summary) stored in a second Google Sheet
Deadline:
Must be 100% tested and working by Friday.
That means I can:
Test by calling it from a real phone
Hear the AI answer with the correct business name
See logs appear in my Google Sheet
Never get an “application error” again
Budget:
$1500 total, broken into milestones:
$500 for working starter plan (call answers with correct business name)
$500 for Pro plan (logging, follow-ups)
$500 final testing + polish
To Apply:
Please show me:
A Twilio + OpenAI or similar project you’ve built
How fast you can start
If you’re confident you can meet the Friday deadline
BONUS:
If this goes well, I may hire you for continued feature development (user dashboard, Stripe billing, analytics, etc.).
Looking forward to working with you.
Existing Work:
I already have:
A working Google Sheet with contractor data
A Flask-based API deployed on Render
Twilio and OpenAI accounts connected
The basic flow built: incoming call → AI → response → (log to Google Sheet)
However, it’s been inconsistent and unreliable in real-world tests (often returns “application error” on live calls).
You’ll be given access to:
The full Replit and Render codebase
My Google Sheet structure
Working Twilio number
Current Zapier + OpenAI setup (for logging)
You're welcome to use or rebuild any part of this. The most important thing is:
By Friday, the AI must reliably answer live forwarded calls using the correct business name from the Sheet — with no crashes.
Apply Job!
Apply to this Job