CallFlowBot is a full-stack simulated AI call agent demo built with Flask, OpenAI API, and a modular web UI. It mimics Twilio webhook interactions, provides LLM-style responses with fallback logic, logs user sessions, and includes analytics + demo tooling — now live and fully deployable.
“Can you tell me your pricing?”
“How do I cancel my subscription?”
“Thank you.”
Or toggle Voice Mode 🎙️ to simulate tone switching.
- Accepts user input via UI or POST to
/twilio-webhook - Responds using OpenAI GPT-3.5 — or scripted fallback logic
- Tracks unique
session_idper user - Logs all messages + bot responses to
.json - Shows analytics dashboard (
/dashboard) - Features a live
/logsviewer - Full interactive demo with scripted buttons and toggles
- Flask for backend + routes
- HTML / JS / CSS frontend
- OpenAI API (or fallback logic)
- JSON logging per session
- Chart.js for live analytics
- Render for deployment
| Phase | Description |
|---|---|
v1.0 |
Scripted backend logic for basic AI simulation |
v1.1 |
OpenAI integration with fallback support |
v2.0 |
Web UI with dynamic input + bot replies |
v3.0 |
.json log output system |
v4.0 |
Live deployment to Render |
v5.0 |
/logs viewer route with interaction table |
v6.0 |
/dashboard with fallback % and top user questions |
v6.5 |
Bar chart + CSV download from dashboard |
v7.0 |
Session filtering + per-user stats |
v8.0 |
Interactive demo page with one-click prompts |
v9.0 |
Voice mode toggle + frontend polish |
git clone https://github.com/AndreaZaragoza/CallFlowBot.git
cd CallFlowBot
# Create a .env file with your OpenAI key
echo "OPENAI_API_KEY=sk-..." > .env
pip install -r requirements.txt
python app.pyVisit http://localhost:5001/chat
curl -X POST http://localhost:5001/twilio-webhook \
-H "Content-Type: application/json" \
-d '{"transcript": "Can you tell me your pricing?"}'- Voice-to-text or Twilio integration (real call flow)
- Agent sentiment tracking
- Admin panel with filters + user activity
- Export logs by session