Dexora AI⚡ - Enterprise Data Assistant

Our goal is to simplify enterprise data access by enabling secure, role-based, AI-powered natural language queries, ensuring efficiency, compliance, and actionable insights.

Comment 1

⚡ DexoraAI v2 – Enterprise Data Assistant

🏆 Competition Highlights

  • 🥇 Winner at HackHazards 2025 – World’s Largest Community-led Hackathon out of 17,000 participants.
  • 🥉 4th Place at Codeshastra XI – among 100+ cutting-edge AI projects.

🚀 About the Project

ConvulenceAI is an AI-powered Enterprise Data Assistant that transforms how organizations access, analyze, and secure corporate data.

It fuses Retrieval-Augmented Generation (RAG) + Dynamic UI Generation + Role-Based Access Control into one seamless platform.

✨ Instant Natural Language Data Queries ✨ Multi-Modal Insights (documents, images, videos, spreadsheets) ✨ ML-Powered Access Decisions & Security Monitoring ✨ Dynamic Chat-Driven UI Workflows ✨ Enterprise-Grade Compliance & Audit Controls

🔑 Problem We Solve

Problems Faced by Enterprises

Companies today face:

  • ❌ Slow manual retrieval of siloed data (PDFs, images, spreadsheets).
  • ❌ No unified platform for seamless access.
  • ❌ Limited scalability with enterprise tools.
  • ❌ Lack of in-chat analytics or dynamic visualization.

🥊 Competitive Landscape

Competitor Comparison

Unlike Glean or Hebbia, ConvulenceAI is:

  • 🔹 Multimodal-first – Handles text, images, video, structured data.
  • 🔹 Dynamic UI Chat Layer – Instantly builds forms, dashboards, & workflows.
  • 🔹 Enterprise Security Core – Role-based access + ML-driven anomaly detection.

🏗️ System Architecture

User Query → Dynamic Chat Layer (Next.js + Groq AI)  
          → Backend (Flask + LangChain + ChromaDB)  
          → Security Core (RBAC + ML anomaly detection)  
          → Response: Instant Insights + Auto-Generated UI

What it does

ConvulenceAI acts as a secure enterprise assistant, where users can:

  • Ask any question in natural language.
  • Get instant multimodal insights (text, PDF, images, spreadsheets).
  • Generate dynamic dashboards & workflows automatically.
  • Ensure role-secured access with anomaly detection & compliance.

How we built it

  • Frontend → React + TypeScript + TailwindCSS + Chart.js + Framer Motion
  • Chat Layer → Next.js + Groq AI for dynamic UI generation
  • Backend/ML Engine → Flask + LangChain + ChromaDB + RAG embeddings
  • Security Core → Role-based access, Random Forest anomaly detection, encrypted pipelines

Challenges we ran into

  • Handling multimodal retrieval (videos, images, PDFs) in one pipeline.
  • Designing real-time role-based access control without slowing response.
  • Scaling vector embeddings for enterprise-grade performance.
  • Building dynamic UI components that auto-generate based on context.

Accomplishments we’re proud of

  • 🏆 Winning HackHazards 2025 (17k participants) & Codeshastra XI (Top 4).
  • Building a multimodal-first AI assistant that competitors lack.
  • Achieving real-time anomaly detection with role-based secure access.
  • Creating a seamless chat-to-dashboard flow with no manual setup.

What we learned

  • How to integrate AI + security + compliance in one pipeline.
  • Importance of enterprise-ready scalability when working with real-world data.
  • How dynamic UI generation can reduce friction in corporate workflows.

What’s next for ⚡ConvulenceAI

  • 📌 Deploy to enterprise-scale clients with cloud-native scaling.
  • 📌 Add voice-first multimodal search.
  • 📌 Expand compliance support (HIPAA, SOC2).
  • 📌 Launch enterprise partnerships & pilot programs.

🔥 ConvulenceAI isn’t just a chatbot — it’s the future of **Enterprise AI Assistants, redefining secure and intelligent data access.

Built With

  • chromadb
  • flask
  • generative-ui
  • groq
  • langchain
  • multimodal-rag
  • nextjs
  • rag
Share this project:

Updates