Getting started

From the first sign-in to the first AI answer, in five steps.

Start in 5 steps

1. Sign in and set up the workspace

On first access you sign in with the admin email defined at installation (see Self-host). In Configurações → Detalhes (Settings → Details) you name the workspace and choose its look. In Configurações → Authentication you decide how people sign in: under Iniciar Sessão (Sign in), turn on E-mail, Passkeys or single sign-on (SSO). The Keycloak address goes in the Keycloak section, right below.

2. Set up the AI

In Configurações → IA → LLM (Settings → AI → LLM), choose the provider (OpenAI, Google Gemini, OpenRouter, litellm or Ollama), enter the key when the provider asks for one (Ollama doesn’t) and the model, and mark it as active. Without an active provider, the chat and the agents don’t answer.

Ollama on the same server, with the wiki in Docker? Use http://host.docker.internal:11434/v1 as the address — inside the container, localhost is the container itself.

For the AI to find documents by the meaning of the question, not just the exact word, there is semantic search. It has to be turned on in the server, with a Qdrant (see Self-host); then, in Configurações → IA → Busca semântica (semantic search), pick the model and click Reindexar tudo (reindex everything).

3. Invite the team

In Configurações → Membros (Members), invite people by email and choose each one’s role. In Configurações → Grupos (Groups) you put people together to grant access in one go. An invitation is valid for 7 days.

4. Create collections and documents

Use Novo projeto (New project) in the sidebar to create a collection for each area or project (e.g. Product, HR, Operations). Inside it, create documents. The editor is block-based: type / to see the options. If you already have content in Markdown or Confluence, bring it all at once through Configurações → Importar (Import).

5. Talk to the AI

Open Chat in the sidebar, or the Chat com IA (Chat with AI) button inside a document. Ask about the content: the answer cites the documents it came from. When the task calls for research and writing, pick an agent. It works in the background: it saves new documents right away and returns changes to existing documents as a proposal — those only go in after you approve them.

Next steps