Epsilon AI — InsAI
Enterprise document-AI (RAG) platform with real-time streaming chat
- Year
- 2026
- Role
- Frontend Developer
- Type
- Enterprise AI platform
- Status
- Live
- Company
- Epsilon AI Analytics
The landing page is public; the platform itself requires a login — demo available on request

Overview
Epsilon AI Analytics offers three enterprise AI products in one platform: RAG.INS (chat with your documents and get sourced answers), VOICE.INS (voice queries and spoken AI answers in every language and dialect) and Construct INS (multiple assistants in one workspace, as tabs). I built the production frontend.
What I did
- Built the full Next.js + TypeScript frontend from scratch to production.
- Implemented real-time chat with Server-Sent Events (SSE) streaming responses.
- Built auth flows and role-based access control.
- Integrated backend APIs through JWT authentication and a Backend-for-Frontend (BFF) layer.
- Added full Arabic / RTL support.
- Prepared CI/CD and deployment handoff to Azure.
Key features
- Streaming AI chat over company documents
- Admin panel: channels, users, document ingestion, reports
- Role-based access control
- Product selection landing page
- English and Arabic with full RTL
- Adjustable text size
- Light and dark theme