💜 The AI Receptionist Engine Built for Agencies & Resellers

Train once. Deploy unlimited. Your clients pay monthly. You keep the margin.

---

💰 Turn AI Assistants Into Recurring Revenue

Conny isn't just another chatbot framework. It's a white-label AI receptionist platform that lets you sell branded conversational AI to restaurants, clinics, salons, real estate agencies, e-commerce stores — any business with a phone number.

Your Client's Pain	What They Pay You For
📞 Missed calls = lost revenue	🤖 24/7 AI receptionist that never sleeps
💬 Slow WhatsApp & Telegram response	⚡ Instant replies with context memory
📝 Answering the same questions manually	🧠 AI trained on their FAQ, services, prices
🔄 Hiring, training, managing staff	🚀 Deploy once, forget about it

---

🚀 Your Business Model (The Conny Way)

┌─────────────────────────────────────────────────────────────┐
│  1. Install Conny (5 minutes)                             │
│     npm install -g conny-ai                               │
└──────────────────────────┬──────────────────────────────────┘
                           ↓
┌─────────────────────────────────────────────────────────────┐
│  2. Train ONE agent with a prompt (15 minutes)              │
│     "You are Maria, receptionist for a dental clinic.       │
│      Book appointments, answer FAQ, send price list."       │
└──────────────────────────┬──────────────────────────────────┘
                           ↓
┌─────────────────────────────────────────────────────────────┐
│  3. Clone to unlimited client instances (1 command)         │
│     conny sync --add /opt/conny-client-restaurant       │
│     conny sync --add /opt/conny-client-salon            │
│     conny sync --add /opt/conny-client-realestate       │
└──────────────────────────┬──────────────────────────────────┘
                           ↓
┌─────────────────────────────────────────────────────────────┐
│  4. Each client gets their own isolated instance            │
│     ✅ WhatsApp Business API integration                    │
│     ✅ Telegram bot (optional)                              │
│     ✅ Custom personality via personas/                     │
│     ✅ Isolated database & credentials                      │
│     ✅ Their own .env — you never touch it again            │
└──────────────────────────┬──────────────────────────────────┘
                           ↓
┌─────────────────────────────────────────────────────────────┐
│  5. Charge $97–$497/month per client                        │
│     They manage their WhatsApp number                       │
│     They update their FAQ via simple .txt files             │
│     You maintain the core and sync updates in 1 command     │
└─────────────────────────────────────────────────────────────┘

Clients	Monthly Fee	Your MRR
10 clients	$197/mo	$1,970/mo
50 clients	$197/mo	$9,850/mo
100 clients	$197/mo	$19,700/mo

Your cost: $5–15/mo per client (OpenAI API + server) Your margin: 85–95% 🚀

---

🎯 Why Conny Destroys the Competition

Solution	Their Reality	Conny Reality
Voiceflow / Botpress	❌ $99–$625/mo per bot ❌ Vendor lock-in ❌ You pay per conversation	✅ Open source, MIT license ✅ Unlimited instances ✅ You control pricing
n8n / Make / Zapier	❌ Complex workflow builders ❌ No memory between sessions ❌ Breaks with API changes	✅ Zero external dependencies ✅ Built-in memory engine ✅ 1 command to update all clients
Hiring a Dev Team	❌ $5,000–$20,000 upfront ❌ 3–6 months development ❌ Single-client solution	✅ Install in 5 minutes ✅ Deploy client in 10 minutes ✅ One core, infinite clients
Custom Python Script	❌ No state management ❌ Security nightmare ❌ Dies when you restart	✅ Production-grade FastAPI ✅ Zero-trust architecture ✅ Systemd + Docker ready

---

⚡ Start Selling in 3 Steps

Step 1 — Install Conny

Recommended GitHub installer (always tracks the current main branch):

curl -fsSL https://raw.githubusercontent.com/sxrubyo/conny/latest/install.sh | bash
conny --version

Direct GitHub install with npm:

npm install -g github:sxrubyo/conny#main
conny --version

Registry install:

npm install -g conny-ai
conny --version

Step 2 — Create Your First Agent

conny persona create restaurant-receptionist

Edit personas/restaurant-receptionist.txt:

You are Sofia, the AI receptionist for "La Cucina Bella" Italian restaurant.

Your job:
- Answer questions about menu, hours, location
- Take reservations (collect: name, phone, date, time, party size)
- Send the menu PDF when asked
- Be warm, friendly, professional

Hours: Mon–Sun 11am–10pm
Location: 123 Main St, Miami, FL

When someone wants to book:
"Perfect! I'll need your name, phone number, preferred date & time, and party size."

Never confirm reservations — say "I'll pass this to our manager to confirm."

Step 3 — Deploy to Client

conny sync --add /opt/conny-restaurant-bella
cd /opt/conny-restaurant-bella
cp .env.example .env
nano .env
python3 conny.py

Done. Their WhatsApp now has a 24/7 AI receptionist. You never touch their credentials again.

---

🏗️ Architecture

                  ┌──────────────────────────┐
                  │    Your Development Core  │
                  │      ~/conny-dev/       │
                  └────────────┬─────────────┘
                               │  conny sync
                  ┌────────────┼────────────┐
                  │            │            │
         ┌────────▼───┐ ┌──────▼─────┐ ┌───▼────────┐
         │  Client A  │ │  Client B  │ │  Client C  │
         │ Restaurant │ │   Salon    │ │ Real Estate│
         ├────────────┤ ├────────────┤ ├────────────┤
         │ .env       │ │ .env       │ │ .env       │
         │ persona    │ │ persona    │ │ persona    │
         │ database   │ │ database   │ │ database   │
         │ WhatsApp   │ │ Telegram   │ │ WhatsApp   │
         └────────────┘ └────────────┘ └────────────┘

Module	Purpose
conny.py	FastAPI orchestrator — webhooks, routing, concurrency
conny_brain_v10.py	Memory layer — context normalization, conversation history
conny_domino.py	Quality control — validates responses before delivery
conny_core/	Shared conversation logic and state retention
conny_agents/	Pluggable skills — calendar, CRM, payments, custom functions
personas/	Personality configs — tone, language, brand voice per client

---

🔄 Core vs. Instance State

✅ Synced to all instances conny.py # Main engine conny_brain_v10.py # Memory system conny_domino.py # Quality control conny_core/ # Shared logic conny_agents/ # Skills & integrations personas/ # Personality templates requirements.txt # Dependencies

❌ Never synced (instance-specific) .env # API keys, secrets *.db # Conversation history auth_info_*.txt # WhatsApp sessions logs/ # Instance logs backups/ # Local backups

Updates are safe. Sync the engine without ever touching client credentials or data.

---

🎮 CLI Reference

curl -fsSL https://raw.githubusercontent.com/sxrubyo/conny/latest/install.sh | bash
conny --version                  # Check version

conny sync --list                # List all client instances
conny sync --add /opt/client     # Register new client
conny sync --remove /opt/client  # Remove client
conny sync -y                    # Push updates to all clients

conny persona create <name>      # New personality template
conny agent list                 # Show available agents
conny validate                   # Health check — config, deps, files
conny config                     # Panel interactivo: red, modelos, webhooks, Python/venv
conny doctor --fix               # Diagnóstico con auto-reparación de PM2, deps y webhook

---

🛡️ Security by Design

Risk	Other Platforms	Conny
API keys in version control	❌ Common mistake	✅ .env never synced
Shared database across clients	❌ GDPR violation	✅ Isolated SQLite per instance
Session hijacking	❌ No isolation	✅ Separate auth per client
Cross-client data contamination	❌ Possible	✅ Impossible by design

---

📊 Industry Use Cases

Industry	What the AI Handles	Suggested Price
🍕 Restaurants	Reservations, menu, hours, delivery	$147–$297/mo
💇 Salons & Spas	Bookings, services, prices, upsells	$197–$397/mo
🏠 Real Estate	Lead qualification, listings, viewings	$297–$597/mo
🦷 Medical / Dental	Consultations, forms, reminders	$347–$697/mo
🛒 E-commerce	Product questions, order tracking, returns	$197–$497/mo
🏋️ Gyms & Studios	Class bookings, schedules, memberships	$197–$397/mo

---

🔥 Production Deployment

VPS (DigitalOcean, Linode, Vultr — $6/mo)

npm install -g conny-ai
conny sync --add /opt/client-001
cd /opt/client-001
cp .env.example .env && nano .env

Systemd service (auto-restart on crash):

[Unit]
Description=Conny AI — Client 001
After=network.target

[Service]
WorkingDirectory=/opt/client-001
ExecStart=/usr/bin/python3 /opt/client-001/conny.py
Restart=always

[Install]
WantedBy=multi-user.target

sudo systemctl enable conny-client-001
sudo systemctl start conny-client-001

Docker

docker run -d \
  --name conny-client-001 \
  -p 8000:8000 \
  -v /opt/client-001:/app \
  --env-file /opt/client-001/.env \
  conny-ai:latest

---

🗺️ Roadmap

• Web dashboard for clients (no-code persona editor)
• Multi-language personas (auto-switch by locale)
• Built-in analytics — conversation metrics per instance
• Voice support — Telegram voice → transcription → AI response
• Redis-backed session sharing for horizontal scaling
• Plugin marketplace — community agents and skills
• Edge deployment — Cloudflare Workers / Vercel Edge

---

🎓 What You DON'T Need

❌ n8n workflows    ❌ Zapier subscriptions    ❌ Voiceflow licenses
❌ AWS Lambda complexity    ❌ Kubernetes    ❌ Redis (unless 500+ clients)
❌ PostgreSQL    ❌ Docker Swarm    ❌ A dev team

Conny runs on a $6/mo VPS. One server. Dozens of clients.

---

💜 Ready to Build?

---

Built for agencies. Designed for profit. Engineered for scale.

No vendor lock-in · No recurring SaaS fees · You own the code · You set the price

Created by sxrubyo · MIT License · Open Source