Skip to Content

AI Chat Mode

Prepare data for embedded Bryn

Get a datamart, a workspace and an AI model ready so your end users can ask questions in plain words.

Embedded Bryn lets your end users ask a question in plain words inside an embedded dashboard and get a chart back. This page gets the data side ready. Step 2 turns it on for an embed.

Before you start

  • You are an Admin, or your role can create datamarts and edit LLMs.
  • Your plan includes embedding. Embedded data apps are on Enterprise.

Steps

  1. Create a datamart with the tables your end users can ask about. See Create a datamart. Bryn answers only from this datamart, so leave out tables your customers must not see.
  2. Set up tenancy on the datasource, so each client sees only their own rows. See Choose a tenancy model. Bryn applies the same client filter as every other embedded query.
  3. Add the datamart to a workspace. See Create a workspace. The embed uses this workspace.
  4. Connect an LLM in Settings > LLMs. See Connect an LLM. Embedded Bryn always uses your own model. Databrain AI does not answer for a guest token.
  5. Optional: add a semantic model. Descriptions, synonyms and metrics make answers more accurate. If the datamart has a semantic model, Bryn uses it. Without one, Bryn queries the datamart tables. See Semantic models.

You now have the data, the tenancy and the model. Next, turn on Bryn for an embed.

If something goes wrong

  • Bryn answers “NO DATA FOUND”. The embed’s workspace has no datamart. Add one to the workspace.
  • “No AI model is configured. Add one in LLM settings.” Connect an LLM in Settings > LLMs.