Sales Ai Agent: AI-Powered Sales & Email Automation

Overview

aiagentSales is an AI-driven platform designed to automate and enhance sales email generation, customer query replies, and database-driven responses for companies. It leverages advanced LLMs (Gemini/OpenAI), Retrieval-Augmented Generation (RAG), and MongoDB to streamline outbound sales, customer support, and information retrieval workflows.

Features

• Sales Email Generation: Three distinct AI sales agents (professional, engaging, concise) generate cold emails. The best email is selected and sent automatically.
• OPEN AI SDK: LLM integration with OPENAI SDK.
• Automated Query Replies: Uses RAG to answer customer queries by searching a knowledge base.
• Database Query Handling: Converts user requests into MongoDB queries, fetches results, and crafts human-readable responses.
• Agent Handoffs: Intelligent agent handoff system routes tasks between specialized agents (e.g., sales, query, database) for optimal workflow automation.
• Guardrails: Input guardrails for name and spam detection to ensure safe and relevant interactions.
• Email Delivery: Integrates with SendGrid for sending emails.

Project Structure

    SalesAgent/
      app.py                # Main application entry point
      requirements.txt      # Python dependencies
      Agents/               # Sales and DB agent tools
      Guardrails/           # Input/spam guardrails
      Model/                # LLM model setup (Gemini)
      RagSearch/            # RAG utilities (ChromaDB)
      Tools/                # Email, MongoDB, and utility functions

Setup Instructions

1. Clone the repository
2. Install dependencies:

   pip install -r requirements.txt

3. Environment Variables:
   • Create a .env file in ai/SalesAgent/ with the following keys:
     • GEMINI_API_KEY (for Gemini LLM)
     • OPENAI_API_KEY (for traces )
     • SENDGRID_API_KEY (for email delivery)
   • Ensure MongoDB is running locally at mongodb://localhost:27017/ and has a database LPU with a courses collection.
4. Prepare ChromaDB:
   • Place your PDF documents in /docs/ and run create_chroma_db.py to build the knowledge base.

Usage

• Run the main application:

  python app.py

• The system will process either sales or query emails based on the input type.
• Modify email_generator() in app.py to test different scenarios.

Key Components

• Agents:
  • agenticTools.py: Three sales personas as tools.
  • DBtool.py: MongoDB query generator and response formatter.
• Guardrails:
  • inputGuardrail.py: Checks for personal names in messages.
  • spamGuardrail.py: Detects spam or irrelevant queries.
• RAG Search:
  • create_chroma_db.py: Builds ChromaDB from PDFs.
  • query_chroma_db.py: Retrieves relevant info for queries.
• Tools:
  • projectFunctions.py: Email sending and error logging.
  • MongoData.py: Executes MongoDB commands.
• Model:
  • geminiModel.py: Configures Gemini LLM for all agents.

Notes

• Update sender/recipient emails in projectFunctions.py as needed.
• The .env and venv directories are ignored by git.