Overview
We developed Prism AI, a Knowledge Base & AI System as a demonstration platform to showcase our capabilities in RAG, AI agents, knowledge retrieval, and intelligent integrations.
Requirements
- Access information from multiple knowledge sources.
- Search and retrieve information from internal documents
- Interact with business databases using natural language
- Connect AI agents with external APIs and services
- Maintain structured FAQ knowledge
- Retrieve information from the web when required
- Extract information from dynamic websites
- Support different LLM providers without vendor lock-in
- Provide a unified conversational experience across different data sources
Our Solution
We developed Prism AI as a modular AI knowledge platform centered around a single conversational agent:
- RAG from Documents
- RAG from Databases
- External APIs via MCP
- FAQ Knowledge Base
- Web Search
- RAG from Cloud documents
Technologies Used
- Python
- PostgreSQL + pgvector
- Streamlit / FAST API
- LangChain Tools
- LangGraph Agent
- Typed Structured Output
- Guardrail Middleware
- OpenAI
- MCP Server
- Google Drive Sync
