All Projects

MACAS

Completed

Multi-Agentic Conversational AI System + CRM

Solo Developer & Architect2024 – 2025AI System + CRM Platform

Overview

A production-grade multi-agentic conversational AI system with a built-in CRM. Integrates multiple LLMs with RAG for semantic search across documents.

The Problem

Traditional CRM systems require manual data entry, lack intelligent conversation capabilities, and can't process unstructured documents. Businesses need a system that can automatically understand queries, process documents, and qualify leads without human intervention.

What I Built

  • Multi-Agent AI Engine with 3 integrated LLMs working together for different tasks
  • RAG pipeline using ChromaDB for semantic search across PDF, DOCX, CSV documents
  • Full CRM with automated contact extraction, lead scoring, and sentiment analysis
  • 15+ REST APIs with JWT authentication and conversation memory
  • Dockerized deployment with Nginx reverse proxy and Gunicorn WSGI server

Key Highlights

Integrated 3 LLMs (GPT, Claude, Gemini) with RAG-powered semantic search

Designed 15+ REST APIs with JWT auth, conversation memory, and tagging

Achieved 90%+ accuracy in automated lead qualification

Sub-2 second response time in production

Impact & Metrics

90%+
Lead Qualification Accuracy
< 2 seconds
Response Time
15+
REST APIs
3
LLMs Integrated

Tech Stack

FlaskMySQLChromaDBGPTClaudeGeminiDockerNginxGunicornLangChain
AuravestiaKNCRM