Skip to content

What is TrustRAG?

TrustRAG is a self-contained, multi-platform RAG (Retrieval-Augmented Generation) knowledge workbench. It helps you build a private knowledge base, upload documents, ask questions, and — most importantly — inspect the evidence behind each AI answer.

Why TrustRAG?

Most RAG systems give you an AI-generated answer and maybe a list of "sources." TrustRAG goes further:

  • Every citation is traceable — linked to a specific document, chunk, page number, and heading path
  • Citations are reviewable — approve, reject, or flag each citation for accuracy
  • Review history is preserved — full audit trail for accountability

Two Modes

Desktop Mode (Self-Contained)

Download and run. No servers, no databases to configure. TrustRAG bundles a Rust backend and SQLite database inside the app. Available on Windows, macOS, Linux, Android, and iOS.

Server Mode (Full Stack)

For teams that need PostgreSQL vector search, Redis caching, and MinIO file storage. Deploy with Docker Compose.

Key Features

FeatureDescription
RAG ChatDocument-grounded Q&A with streaming responses
Citation TrackingEvery answer cites specific document chunks
Citation ReviewApprove/reject/flag citations with full audit trail
Review ReportsExport Markdown reports with hallucination rate, approval rate, review coverage
Knowledge GraphInteractive visualization with entity browsing, search, and color-coded graph
Full-Text SearchFTS5-powered search across all documents
Native PDF ParsingDesktop builds parse PDF locally via Rust lopdf (no server needed)
Multi-Language (i18n)Chinese, English, Japanese, Korean UI with persistent language preference
Team CollaborationPersonal & team workspaces, invite codes, role-based access (owner/admin/editor/viewer)
Resizable SidebarDrag to resize, one-click collapse to icon-only mode, persisted preferences
Multi-PlatformWindows, macOS, Linux, Android, iOS, Web

Released under the Apache 2.0 License.