> ## Documentation Index
> Fetch the complete documentation index at: https://netter.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Query in natural language

> Ask questions across the entity graph in plain English. The platform writes the SQL.

The ontology surfaces a **natural-language query** entry point. You type a question; the platform plans the SQL across the entity graph, runs it, and returns the rows.

## How to use it

<Steps>
  <Step title="Open the query tab">
    Go to **Ontology → Query**.
  </Step>

  <Step title="Ask">
    Type a question that references entities and their properties. The platform names what's available; you don't have to remember table names.

    Examples:

    * "Which customers signed up last week and haven't logged in since?"
    * "Top 10 invoices by amount, with the customer name."
    * "Tickets created in the last 30 days, grouped by priority."
  </Step>

  <Step title="Inspect the result">
    The result panel shows the generated SQL and the returned rows. If the answer is wrong, refine the question—the platform retries up to three times if a query fails.
  </Step>
</Steps>

## Scope

Queries run as **SELECT** statements over the merged entity views (source + overlays). Writes go through [row overlays](/docs/ontology/edit-rows) or [Application actions](/docs/applications/actions), not through the query path.

Row-level security is enforced in the generated SQL. A viewer sees only the rows their role grants access to.

## Relationships

Because entities carry typed relationships, questions like "customers' invoices last quarter" resolve without you specifying a join—the platform knows the edge between `Customer` and `Invoice`. Add or fix relationships from the entity detail page if a question can't be answered the way you expect.