Collab

Posted on by Aiqwip

Project Overview

  • Project: Multi-Agent Smart Assistant for Collabsuite
  • Domain: Organization productivity and project management
  • Client: US-based startup
  • Duration: 6+ months
  • Scope: Design, Development, Deployment and Ongoing maintenance
  • Use Case: A Multi-Agent framework based Smart Assistant that enables chatting with your project and organisation data.

 

Tech Stack

  • Front End: React + ViteJS, TailwindCSS
  • Backend: Python FastAPI
  • Generative AI: AWS Bedrock, LangChain, LangGraph, Prompt Engineering
  • Database: DynamoDB, OpenSearch, AWS RDS
  • Cloud: AWS Ubuntu VM
  • DevOps: Nginx, Uvicorn, Docker

 

Solution Offered

  • Designed, developed and deployed a multi-agent smart assistant capable of routing user queries to specialised agents that use diverse data sources to draft their response. This response also included insightful visualisations.

 

Features

  • LangGraph multi-agent framework with specialized agents
  • RAG-based Q&A using AWS OpenSearch
  • Natural Language to SQL and Visualisations
  • Conversation management using Redis
  • Agent State management using LangGraph
  • AWS Bedrock to plug and play various LLMs