
Siftrio
Web ApplicationAI-powered project memory platform that centralizes scattered project context — meetings, decisions, requirements, and risks — into a searchable knowledge layer with citation-backed AI responses.
Siftrio is an AI-driven platform designed to eliminate context loss across software projects. It ingests project knowledge from meetings, documents, Google Calendar, Fireflies.ai, and Jira into a unified, searchable workspace. Teams can ask natural language questions and receive AI-generated answers grounded in their actual project data — not hallucinations. Features include AI-powered meeting analysis with structured insight extraction, a hybrid RAG pipeline built with LangGraph, an MCP server for IDE-native access, and an approval queue ensuring nothing goes live without human review. Built with a clean architecture backend and a modern responsive frontend for teams managing multiple clients and projects.
Technologies Used
Key Features
- Hybrid RAG pipeline with vector search, keyword search, and reciprocal rank reranking built on LangGraph
- AI meeting analysis extracting summary, outcomes, blockers, requirements, action items, decisions, and risks
- Natural language project memory chat with citation-backed AI responses
- Multi-tenant workspace hierarchy with role-based access (Workspace > Client > Project)
- Knowledge entity management for requirements, action items, decisions, risks, and questions with approval workflows
- Fireflies.ai integration for automatic transcript ingestion and meeting matching
- Google Calendar integration for event creation and Google Meet link generation
- Jira OAuth integration for project mapping, bidirectional action item sync, and issue tracking
- MCP server exposing 16+ tools to Cursor, VS Code, and other IDEs for context-native development
- Approval queue with status tracking ensuring extracted insights are reviewed before going live
- Async SMTP email system for team invitations with HTML templates
- Secure JWT authentication with access/refresh token flow and cookie-based session management