Rafetus is an AI-integrated unstructured data management application that lets users upload documents, take notes, and query across an entire knowledge base in one workspace. It combines semantic search, document metadata, and explicit note-link relationships so AI answers are grounded in user-approved sources. The platform supports PDF, DOCX, meeting transcripts, and URLs, with AES-256 encryption and a no-training-without-consent policy.
Funding
Funding not disclosed
Founders
Product
Problem
Individuals and organizations often struggle with scattered, unstructured documents—PDFs, meeting transcripts, and notes—that are difficult to search, connect, and query as a whole. Traditional file storage and note-taking tools lack deep cross-document AI-powered question answering, forcing users to manually piece together information across multiple sources.
Solution
Rafetus provides a unified workspace where users upload diverse document types, and the system automatically reads, indexes, and prepares content for semantic search and AI-driven Q&A. Users can open documents, ask questions across the entire knowledge base, and receive answers with cited sources, while also writing notes beside documents and linking them with explicit relationships like supports, supplements, or contradicts. The AI, called Rafai, queries the full knowledge repository to analyze and connect information, using only the documents and notes the user permits. This approach combines document management, note-taking, and AI querying into a single system, unlike tools that focus on one aspect or lack deep multi-document retrieval.
Target Audience
Primary users are individuals or organizations with many scattered documents who need to read, retrieve, annotate, and query within a single knowledge base, such as researchers, analysts, and knowledge workers handling large volumes of unstructured information.
Features
- Supports upload of PDF, DOCX, meeting transcripts, and URLs into a single working repository
- AI-powered semantic search and question answering across multiple documents, with responses citing relevant sources
- Note-taking directly beside documents with explicit link types (supports, supplements, contradicts) to express relationships
- Document metadata management including creation date, author, and document type
- AES-256 encryption for stored data and SSL/TLS for secure HTTPS connections
- ZDR AI policy ensuring no data is used for AI training without explicit user consent
- Optimized for complex queries spanning large document collections, unlike notebook-based or folder-based tools