Capability · Framework — rag
Verba
Verba (The Golden RAGtriever) is Weaviate's opinionated RAG application. It provides end-to-end ingestion, chunking, embedding, retrieval, and a chat UI backed by a Weaviate vector database. You can run it entirely locally with Ollama and local embeddings or use hosted providers (OpenAI, Anthropic, Cohere, Google, HuggingFace). It's a popular starting point for teams prototyping a chat-with-your-docs experience.
Framework facts
- Category
- rag
- Language
- Python / TypeScript
- License
- BSD-3-Clause
- Repository
- https://github.com/weaviate/Verba
Install
pip install goldenverba
# then
# verba startQuickstart
# After pip install goldenverba
export OPENAI_API_KEY=sk-...
export WEAVIATE_URL_VERBA=...
export WEAVIATE_API_KEY_VERBA=...
verba start
# Open http://localhost:8000 and upload documentsAlternatives
- R2R — RAG server, no UI by default
- Chat with LlamaIndex — reference chat UI
- open-webui — chat UI with RAG plugins
- AnythingLLM — similar open-source RAG app
Frequently asked questions
Do I have to use Weaviate Cloud?
No. Verba supports embedded Weaviate, local Docker Weaviate, and Weaviate Cloud. For a zero-infrastructure start you can use embedded mode with local Ollama embeddings.
Is Verba production-ready?
Verba is best treated as a high-quality reference implementation. It's fine for internal tools and demos, and many teams fork it for production. For heavy multi-tenant deployments you'd typically build on Weaviate directly.
Sources
- Verba — GitHub — accessed 2026-04-20
- Weaviate — docs — accessed 2026-04-20