Deploy WeKnora RAG platform
WeKnora document RAG with DocReader parsing, ParadeDB search, and Redis
Web
Just deployed
Application
WeKnora DocReader
Just deployed
WeKnora App
Just deployed
/data/files
Data
Redis
Just deployed
/data
PostgreSQL
Just deployed
/var/lib/postgresql/data
Deploy and Host WeKnora RAG platform
WeKnora is Tencent's open-source platform for document understanding and retrieval-augmented generation. Upload PDFs, Office files, web pages and more into knowledge bases, then ask questions and get answers that cite the retrieved passages.
About Hosting WeKnora RAG platform
This template runs WeKnora v0.9.0 as five services: the public web UI, the private Go API with its background workers, the private DocReader parsing service, ParadeDB (PostgreSQL with pgvector and BM25 search), and password-protected Redis for the document-processing queue. Uploaded files, the database and the queue each have their own volume. Public sign-up is closed, and the first boot creates one administrator account with a generated password, found in the WeKnora App service variables as WEKNORA_ADMIN_PASSWORD (sign in as admin@example.com unless you change WEKNORA_ADMIN_EMAIL before deploying). After signing in, add your own chat and embedding models from any OpenAI-compatible endpoint or supported provider.
Why Deploy WeKnora RAG platform on Railway
Railway provides TLS, private networking, generated secrets and volumes, so the API, parser, database and queue stay off the public internet while documents and indexes survive redeploys. Images are pinned by digest, and the web UI keeps working when the API is redeployed.
Common Use Cases
- Answer questions over internal handbooks, contracts, research papers or support documentation, with citations.
- Give a team a shared knowledge base with per-user accounts and invitations.
- Run agent-style multi-step retrieval over several knowledge bases with your own model provider.
Dependencies for WeKnora RAG platform Hosting
- An API key for a chat model and an embedding model from a provider WeKnora supports, or a self-hosted OpenAI-compatible endpoint.
- Volume space for uploaded documents and the ParadeDB index.
Deployment Dependencies
Template Content
Redis
redis:8.4.0-alpinePostgreSQL
paradedb/paradedb:v0.25.11-pg17WeKnora DocReader
wechatopenai/weknora-docreader:v0.9.0WeKnora App
wechatopenai/weknora-app:v0.9.0
