Skip to content

Built end-to-end AI Sales Development Representative system with AI prospect researcher, AI email generator, and AI follow-up and Meeting Scheduler, boosting engagement by 35%. Created AI meeting assistant that transcribes and extracts action items, with searchable knowledge base.

Notifications You must be signed in to change notification settings

mohdjami/AI-Sales-development-representative

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 

Repository files navigation

AI SDR + Meeting AI + Knowledge Base(code is private)

A comprehensive AI-powered system that integrates multiple AI capabilities for sales development, meeting analysis, and knowledge management. The platform helps sales teams identify prospects, generate personalized outreach, automate follow-ups, and capture meeting insights.

System Architecture

image

Table of Contents

Overview

This system is designed to transform the traditional sales development process by leveraging AI at every stage:

  1. AI Lead Generator & Analyzer: Scrapes LinkedIn and other platforms to find and analyze potential leads.
  2. Personalized AI Email Generator: Creates highly personalized emails based on prospect data and pain points.
  3. AI Follow-up & Meeting Scheduler: Monitors email responses and automates follow-ups.
  4. AI Meeting Notes & Knowledge Base: Captures, summarizes, and makes searchable all meeting knowledge.

Architecture

The system uses a modern microservices architecture:

├── Frontend (Next.js)
│   ├── Dashboard
│   ├── Prospect Management
│   ├── Email Editor
│   ├── Meeting Management
│   └── Knowledge Search
│
├── Backend (FastAPI)
│   ├── AI SDR Services
│   ├── Meeting AI Services
│   ├── Knowledge Base Services
│   └── Authentication/Authorization
│
├── Storage
│   ├── Supabase (Persistent Storage)
│   ├── Redis (Caching)
│   └── Pinecone (Vector Database)
│
└── External Services
    ├── Groq (LLM Provider)
    ├── MeetingBaaS (Meeting Bot)
    └── Jina AI (Vector Embeddings)

Module Details

AI SDR Module

The AI SDR module is responsible for finding, analyzing, and engaging with prospects.

LinkedIn Service (agents/services/linkedin_service.py)

  • Functionality: Scrapes LinkedIn for potential leads, analyzes posts for pain points, and identifies promising prospects.
  • Key Components:
    • LinkedInPublicScraper: Scrapes LinkedIn content
    • analyze_posts(): Extracts insights from posts using LLM
    • get_prospects(): Filters and scores leads based on alignment
  • Technologies: Selenium, BeautifulSoup, Groq LLM

Email Service (agents/services/email_service.py)

  • Functionality: Generates personalized email drafts using a multi-stage LangGraph workflow.
  • Key Components:
    • build_workflow(): Creates the email generation state graph
    • subject_agent(): Generates compelling subject lines
    • content_builder_agent(): Creates initial email content
    • content_refiner_agent(): Refines and improves the draft
    • final_draft_agent(): Formats the final email
  • Technologies: LangGraph, Groq LLM

Reply Tracker (agents/services/reply_tracker.py)

  • Functionality: Analyzes email responses for sentiment and intent, generates follow-ups.
  • Key Components:
    • analyze_sentiment(): Determines email sentiment and intent
    • generate_followup_email(): Creates contextual follow-up emails
    • GmailService: Handles Gmail API integration
  • Technologies: Google Gmail API, Groq LLM

Meeting AI Module

The Meeting AI module handles meeting participation, transcription, and analysis.

Meeting Analyzer (agents/services/meeting_analyzer.py)

  • Functionality: Analyzes meeting transcripts to extract summaries, action items, and insights.
  • Key Components:
    • analyze_meeting(): Processes transcripts to generate structured insights
    • Few-shot examples for consistent output formatting
  • Technologies: Groq LLM, MeetingBaaS API

Meeting Bot Integration (agents/main.py)

  • Functionality: Adds bots to meetings, receives webhooks with transcripts.
  • Key Components:
    • /add-bot: Endpoint to add a bot to a meeting
    • /webhook: Receives meeting completion events
    • Meeting status management
  • Technologies: MeetingBaaS API, FastAPI

Knowledge Base Module

The Knowledge Base module stores and retrieves meeting knowledge.

Vector Service (agents/services/vector_service.py)

  • Functionality: Creates embeddings, stores meeting data, and enables semantic search.
  • Key Components:
    • create_embedding(): Generates vector embeddings using Jina AI
    • store_meeting_data(): Chunks and stores meeting content
    • search_meetings(): Performs vector similarity search
    • generate_rag_response(): Creates RAG-based answers from meetings
  • Technologies: Pinecone, Jina AI Embeddings, RAG

LLM Service (agents/services/llm_service.py)

  • Functionality: Provides a unified interface for LLM interactions.
  • Key Components:
    • get_json_response(): Gets structured JSON responses
    • get_streaming_response(): Streams responses for UI
    • get_completion(): Gets regular text completions
  • Technologies: Groq, LangChain

Frontend Module

The frontend provides a user-friendly interface for the system.

Dashboard (components/dashboard/)

  • Functionality: Displays key metrics, recent activities, and insights.
  • Key Components:
    • DashboardOverview: Shows statistics and activity
    • FollowUps: Manages email responses
    • MeetingNotes: Displays meeting summaries
    • MeetingSearch: Searches knowledge base
  • Technologies: Next.js, Recharts, shadcn/ui

Prospect Management (components/ProspectList.tsx)

  • Functionality: Displays and manages prospects found by the AI.
  • Key Components:
    • ProspectList: Lists all prospects with filtering
    • ProspectCard: Displays prospect details
    • ProspectModal: Shows detailed prospect information
    • EmailDrafts: Manages email generation and sending
  • Technologies: Next.js, shadcn/ui

Authentication (app/(auth)/)

  • Functionality: Handles user authentication and authorization.
  • Key Components:
    • login/page.tsx: Login/signup interface
    • login/actions.ts: Server actions for auth
    • Supabase integration
  • Technologies: Next.js, Supabase Auth

Setup and Installation

Prerequisites

  • Node.js 19+
  • Python 3.9+
  • Docker (optional but recommended)
  • Supabase account
  • Pinecone account
  • Groq API key
  • MeetingBaaS account

Usage

Access the Application

Login with these creds:

Key Workflows

  1. Prospect Generation:

    • Navigate to the Prospects page
    • Set minimum alignment score and click "Generate New Prospects"
    • View and filter prospects based on criteria
  2. Email Generation:

    • Click on a prospect to view details
    • Select "Generate Email Draft"
    • Edit the draft and send
  3. Reply Tracking:

    • Go to the Dashboard and select "Follow-ups" tab
    • Click "Analyze Replies" to process new emails
    • View sentiment analysis and suggested follow-ups
  4. Meeting Management:

    • Navigate to the Dashboard and select "Meeting Notes" tab
    • Add a meeting bot by providing a meeting URL
    • Once a meeting completes, view the transcript, summary, and insights
  5. Knowledge Search:

    • Use the search bar in the Meeting Notes tab
    • Ask questions about past meetings
    • View answers with source information

Adding New Features

  1. New AI Agents:

    • Create a new service in agents/services/
    • Add endpoints in agents/main.py
    • Integrate with existing services as needed
  2. New Frontend Components:

    • Add components in components/
    • Create new pages in app/
    • Update navigation as needed
  3. Database Changes:

    • Update Supabase schema as needed
    • Modify the relevant services that interact with the database

Deployment

Frontend (Vercel)

  1. Push your code to GitHub
  2. Connect your repository to Vercel
  3. Configure environment variables
  4. Deploy

Backend (Railway)

  1. Push your code to GitHub
  2. Connect your repository to Railway
  3. Configure environment variables
  4. Set the start command: uvicorn main:app --host 0.0.0.0 --port $PORT
  5. Deploy

Technologies Used

Backend

  • FastAPI: Web framework for building APIs
  • LangChain: Framework for LLM applications
  • LangGraph: State machines for LLM workflows
  • Groq: LLM provider for text generation
  • Selenium & BeautifulSoup: Web scraping
  • Pinecone: Vector database for embeddings
  • Jina AI: Embedding model provider
  • Redis: Caching and temporary storage
  • Supabase: PostgreSQL database and authentication

Frontend

  • Next.js 15: React framework with App Router
  • TypeScript: Typed JavaScript
  • Tailwind CSS: Utility-first CSS framework
  • shadcn/ui: Component library
  • Recharts: Charting library
  • Sonner: Toast notifications
  • React Hook Form: Form handling

DevOps

  • Docker: Containerization
  • Railway: Backend deployment
  • Vercel: Frontend deployment

License

This project is licensed under the MIT License - see the LICENSE file for details.


For questions or support, please contact mohdjamikhann@gmail.com.

About

Built end-to-end AI Sales Development Representative system with AI prospect researcher, AI email generator, and AI follow-up and Meeting Scheduler, boosting engagement by 35%. Created AI meeting assistant that transcribes and extracts action items, with searchable knowledge base.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published