Skip to content

What a new tenant starts with

After signup, a tenant has a working shell and nothing in it. Knowing the difference between “empty” and “broken” saves a lot of confusion.

  • An admin user — the account you signed up with.
  • A tenant schema — t_<tenant-slug> in the application database. A new tenant’s data never shares rows with another’s.
  • A subdomain — <slug>.semantixapp.com, which determines which tenant a request resolves to.
  • A 14-day trial.
MissingConsequence
LLM providerChat does not work — the most common first blocker
ConnectorsNothing to introspect; the Registry is empty
Context scopesNobody can be granted access, and Home is empty
EntitiesNo tables registered, so no grounding
MetricsNo certified numbers; scope boards are empty
Trust assignmentsEvery entity sits at trust 0
Anomaly profilesFalls back to a built-in default profile

Two defaults that are on:

  • Dashboards — off. The core scope board does not depend on it.
  • Anomaly detection — on, using the built-in default profile, so metrics are still checked while you have no profiles of your own.

The first-run prompt is a shortcut, not a checklist

Section titled “The first-run prompt is a shortcut, not a checklist”

New admins see a prompt offering to create a scope, import from Power BI, or add an LLM provider.

It does not mention connecting a data source — the step that actually populates the Registry. Follow the quickstart instead, or you will end up with a scope, a provider, and no data.

A user with no granted scope sees an empty product and a note to ask an admin. This is correct behaviour, not a failure. Access is granted explicitly under Admin → Users — see context scopes.

The three steps in the quickstart — provider, scope, connector plus introspect — produce a tenant where Chat can answer, the Registry has entities, and someone can be granted access. Everything beyond that is curation, and none of it is required to be useful.