Document Q&A with RAG
Upload PDFs, images, text files, and more — Open WebUI indexes them locally and answers questions based on their content. No document ever leaves your server for processing.
Applicaties → Open WebUI
Your private AI chat, on your server.
Open WebUI is a private, ChatGPT-style interface hosted on your ReefOffice server. Your accounts, conversations, files and search index stay there. Generation can use a ReefOffice-managed GPU through an encrypted tunnel or a provider you configure.
What Open WebUI does
Open WebUI provides a familiar chat interface connected to your choice of AI models — a ReefOffice-managed GPU on Private AI pilots, or your own provider key on Office plans. Built-in document RAG lets you upload files and ask questions about them. Storage, indexing and SSO run on your server; generation receives the prompt and selected context.
Chat, search, draft and analyse — privately
Multiple model support
Connect to local Ollama models (Mistral, LLaMA, Qwen, and more) or external providers like OpenAI, Anthropic, or OpenRouter. Switch between models mid-conversation.
Conversation management
Organise chats into folders, search past conversations, and pick up where you left off. Your entire chat history stays on your server, searchable and private.
OpenAI-compatible API
Open WebUI exposes an OpenAI-compatible endpoint that ReefRunner and Activepieces can call — enabling automations like "summarise new documents" or "answer questions from your knowledge base".
How it's deployed
A single container, private by design
Open WebUI runs on your ReefOffice server behind nginx and Authentik OIDC. Qdrant performs vector search locally and conversations, documents and settings are stored under /opt/openwebui/data. On Private AI pilots, the model endpoint is reached through an encrypted, fail-closed SSH tunnel.
The chat layer of your workspace
Open WebUI is the conversational interface to your ReefOffice stack. Upload a Nextcloud document, query it, or use the OpenAI-compatible API in Activepieces. Open WebUI and Hermes use separately validated chat and agent models; application data stays on your server while generation uses the configured model endpoint.
Common questions
Can I use my own AI provider (OpenAI, Anthropic, etc.)? +
Yes. On Office plans without local GPU, you can add your own provider key in the Open WebUI settings. On AI plans, local Ollama models are the default, but you can still connect external providers alongside them.
Are my chat conversations private? +
Yes. Everything stays on your server — your conversations, uploaded documents, and the models they're processed by. Nothing is sent to a third party unless you explicitly connect an external provider, and even then only the conversation content you submit is sent.
Can I share a chat with someone else on my server? +
Open WebUI supports conversation sharing within your organisation. You can share chats with other users on your ReefOffice server, making it useful for collaborative research and knowledge sharing.
Ready for private AI chat on your own server?
Book a demo and see how Open WebUI on ReefOffice gives you a private, powerful AI chat interface — with document Q&A, local models, and full SSO, all on your own server.