Deploy and Host Bot_MultiAgente(5) on Railway

About Hosting

This repository is designed for easy deployment on Railway, Docker, or any Node.js-compatible cloud platform. It is optimized for cloud hosting and scalable multiagent operations.

Why Deploy

• Automate WhatsApp conversations with AI-powered multiagent logic.
• Centralize customer support, sales, and information flows in a single bot.
• Reduce manual intervention and improve response times.

Common Use Cases

• Customer support automation for businesses.
• Sales funnel and lead qualification via WhatsApp.
• Automated follow-ups and reminders for clients.
• Integration with Google Sheets for CRM or reporting.

Dependencies for

This project requires Node.js (v18+), pnpm, and access to OpenAI API. For Google Sheets integration, a valid Google service account is needed.

Deployment Dependencies

• Node.js (v18 or higher)
• pnpm (latest)
• Docker (for containerized deployment)
• Railway (for cloud deployment)
• OpenAI API Key
• Google Service Account credentials (for Sheets integration)

Repository Data: WhatsApp Multiagent AI Bot (BuilderBot.app)

Name

Bot WhatsApp Multiagente AI Bot (BuilderBot.app)

Description

This project implements a multi-agent WhatsApp bot using BuilderBot and OpenAI Assistants. The system allows a receptionist assistant to route conversations to specialized assistants, maintaining context and thread continuity.

Main Features

• Multi-agent architecture: a receptionist identifies intent and routes to expert assistants.
• Integration with OpenAI Assistants for intelligent responses.
• Customizable and scalable conversational flows.
• Automated follow-ups and conversation closure configurable via environment variables.
• Support for Google Sheets integration and data storage.
• Easy deployment on Railway, Docker, or locally.

Agent Structure

• Receptionist: First point of contact, classifies user intent.
• Specialized assistants: Handle specific queries (sales, reservations, support, etc.).
• Automatic routing: The receptionist decides which assistant to route to based on detected intent.

Required Environment Variables

• ASSISTANT_1, ASSISTANT_2, ASSISTANT_3, ASSISTANT_4, ASSISTANT_5
• OPENAI_API_KEY
• ID_GRUPO_RESUMEN
• msjCierre
• msjSeguimiento1
• msjSeguimiento2
• msjSeguimiento3
• timeOutCierre
• timeOutSeguimiento2
• timeOutSeguimiento3
• PORT

Installation & Usage

1. Clone this repository.
2. Install dependencies: pnpm install
3. Configure your .env file with the required values.
4. Run the bot in development: pnpm run dev
5. (Optional) Deploy on Railway or Docker.

Multiagent Workflow

1. The user writes to the bot.
2. The receptionist (ASSISTANT_1) analyzes the intent.
3. If necessary, the conversation is routed to a specialized assistant (ASSISTANT_2, ASSISTANT_3, etc.).
4. Context and thread are maintained throughout the conversation.
5. If the user does not respond, follow-up and closure messages are triggered as configured.

Customization

• Modify messages and timeouts in the .env file to adapt the bot to your conversational flow.
• Main flows are in src/Flows/.
• The file src/app.ts orchestrates the multiagent logic and routing.

Credits

Developed with BuilderBot and OpenAI. Custom for Pereyra Hugo - DusckCodes.

Contributing

Contributions are welcome! Please submit a Pull Request.

License

This project is open-source and available under the MIT license.

Contact

• Website: https://clientesneurolinks.com
• Email: [email protected]