CorpMind is a multi-agent RAG system that searches Gmail, Drive, Notion and Slack simultaneously — with graph reasoning, contradiction detection, and confidence scoring.
Capabilities
Eleven specialised agents working in concert to surface the right answer from anywhere in your company's knowledge base.
Ask in plain English across Gmail, Drive, Notion and Slack simultaneously
Neo4j maps who knows what, decision trails, and entity relationships
DeBERTa NLI automatically flags when documents conflict with 0.99 confidence
Specialized LangGraph agents coordinate to answer any question intelligently
Every answer comes with confidence score, source citations, and agent trail
Vector similarity + graph traversal combined for maximum retrieval accuracy
From natural language query to confidence-scored answer in seconds.
Ask anything in plain English
→Intent and source routing
→Specialized agents dispatched
→Vector + graph search runs
→DeBERTa NLI cross-checks
→Confidence-scored response
$ corpmind query "What did we decide about the API architecture?"
🧭 supervisor → classified: drive
📁 drive_agent → found 3 files, 18 chunks
🔀 merger → combined → top 10 after reranking
⚠️ contradiction_agent → 0 contradictions found
💡 answer_agent → generated response (312 chars)
Agents: 5 | Chunks: 18 | Confidence: 0.72 (High) | Time: 1.2s
Production-grade open-source tools, running entirely on your machine.
AI / ML
Qdrant
High-performance vector database
Neo4j
Graph database for knowledge mapping
LangGraph
Multi-agent orchestration framework
DeBERTa NLI
Contradiction detection at 0.99 confidence
Groq LLaMA-3.3
Ultra-fast LLM inference
HuggingFace
Embeddings & model hub
Infrastructure
FastAPI
Async Python API backend
Next.js 15
React framework with App Router
PostgreSQL
Relational storage for metadata
Docker
Containerised local infrastructure
Python 3.13
Latest Python runtime
Explore CorpMind with preloaded NovaTech sample data — no signup, no credentials.
Launch Demo →Demo mode · Sample data preloaded · No signup required