KnowVerse

AI-powered knowledge management platform with semantic search, RAG, and MCP integration.

---

Why KnowVerse?

Most document search tools force you to remember exact keywords. KnowVerse understands what you mean.

• Ask questions in plain English - "What was our Q2 revenue?" instead of hunting through files
• Get answers with citations - Every response links back to the exact source chunks
• Connect your existing tools - Google Drive, Notion, or direct uploads
• Use with Claude Desktop - Native MCP server integration for AI assistants
• Keep data isolated - Multi-tenant architecture with organization-level separation

Built for teams who need to find information fast and trust the answers they get.

---

Features

Core Capabilities

Feature	Description
Semantic Search	Find documents by meaning, not just keywords. Powered by sentence embeddings.
Hybrid Search	Combines semantic and keyword search using Reciprocal Rank Fusion for best results.
RAG Chat	Ask questions and get AI-generated answers grounded in your documents. Streaming responses.
Chunk-Level Citations	Every answer includes clickable citations to the exact source passages.
Cross-Encoder Reranking	Optional second-stage reranking for improved relevance on complex queries.

Integrations

Feature	Description
MCP Server	Connect Claude Desktop directly to your knowledge base via Model Context Protocol.
Google Drive	Sync documents from Google Drive with automatic updates.
Notion	Import pages and databases from Notion workspaces.
GitHub	Sync README and markdown files from repositories.
Slack	Import channel messages and discussions for searchable knowledge.
Confluence	Sync pages from Atlassian Confluence spaces.

Enterprise

Feature	Description
SSO/SAML	Single sign-on with SAML 2.0 and OpenID Connect support.
Multi-Tenancy	Organization-based data isolation with role-based access control.
API Access	RESTful API with per-organization API keys.
Stripe Billing	Subscription management with usage-based quotas and Stripe integration.

Billing & Quotas

Plan	Price	Documents	Queries/mo	Team Members
Free	$0	50	500	3
Starter	$49/mo	500	2,500	10
Growth	$149/mo	2,500	15,000	25
Business	$399/mo	10,000	50,000	100
Enterprise	Custom	Unlimited	Custom	Unlimited

Visualization

Feature	Description
3D Vector Space	Interactive visualization of document embeddings with t-SNE/UMAP projections.
Document Clustering	Group similar documents using K-Means, Hierarchical, or DBSCAN algorithms.

---

Architecture

Rails 8 (Web/API)  ─────┬───── PostgreSQL (Metadata + pgvector)
      │                 │
      │                 ├───── Qdrant (Vector Search)
      │                 │
      └─── Haystack ────┴───── Embeddings, Extraction, RAG
           (Python)

---

Quick Start

Prerequisites

• Ruby 3.2+
• Python 3.11+
• PostgreSQL 14+ with pgvector
• Redis 7+
• Docker (optional, for Qdrant)

Development Setup

# Clone and install
git clone https://github.com/noah-ing/knowverse.git
cd knowverse

# Install dependencies
bundle install
pip install -r services/haystack/requirements.txt

# Database setup
bin/rails db:setup

# Start services
docker compose up -d qdrant redis
bin/rails server

# In another terminal
cd services/haystack && uvicorn app.main:app --reload

Docker (Full Stack)

docker compose up -d

---

Configuration

Create a .env file:

DATABASE_URL=postgres://localhost/knowverse_development
REDIS_URL=redis://localhost:6379
QDRANT_URL=http://localhost:6333
HAYSTACK_SERVICE_URL=http://localhost:8000

# AI providers (at least one required for RAG)
OPENAI_API_KEY=sk-...
ANTHROPIC_API_KEY=sk-ant-...

# Connectors (optional - for third-party integrations)
GOOGLE_CLIENT_ID=...
GOOGLE_CLIENT_SECRET=...
NOTION_CLIENT_ID=...
NOTION_CLIENT_SECRET=...
GITHUB_CLIENT_ID=...
GITHUB_CLIENT_SECRET=...
SLACK_CLIENT_ID=...
SLACK_CLIENT_SECRET=...
CONFLUENCE_CLIENT_ID=...
CONFLUENCE_CLIENT_SECRET=...

---

MCP Integration

Connect Claude Desktop to your knowledge base:

{
  "mcpServers": {
    "knowverse": {
      "url": "https://your-instance.com/mcp",
      "headers": {
        "X-MCP-Key": "<YOUR_API_KEY>"
      }
    }
  }
}

Generate your API key from Organization Settings.

---

API Reference

Documents

Method	Endpoint	Description
POST	/api/v1/documents	Upload document
GET	/api/v1/organizations/:id/documents	List documents
DELETE	/api/v1/documents/:id	Delete document

Search

Method	Endpoint	Description
POST	/api/v1/search	Semantic/hybrid search
GET	/search	Web search interface

RAG

Method	Endpoint	Description
POST	/rag/query	Single question/answer
POST	/rag/stream	Streaming response (SSE)

Full API documentation available at /docs/api.

---

Technology Stack

Layer	Technology
Backend	Rails 8, Ruby 3.2
Database	PostgreSQL 14+ with pgvector
Vector Store	Qdrant (HNSW indexing)
ML Services	Python 3.11, Haystack 2.0, FastAPI
Embeddings	SentenceTransformers (all-mpnet-base-v2)
LLMs	OpenAI GPT-4o, Anthropic Claude
Frontend	Hotwire (Turbo + Stimulus), Tailwind CSS
Background Jobs	Solid Queue
Caching	Solid Cache + Redis

---

Testing

# Rails tests
bundle exec rspec

# Python tests
cd services/haystack && pytest

# Linting
bundle exec rubocop

---

License

Copyright 2024-2025 Noah Ingwers. All rights reserved.

This software is proprietary and confidential. Unauthorized copying, distribution, or use of this software is strictly prohibited.

For licensing inquiries, contact: noah.ing@protonmail.com