Skip to main content

Query Generation

Beyond generating whole applications, ToolJet AI works at the query level: it can create fully configured queries, write JavaScript and Python snippets, and tidy up a query panel that has grown unwieldy.

Generate Query​

ToolJet AI can generate complete, fully-configured queries directly from the AI chat. It creates an actual query in your query panel — with parameters and configuration all set up.

To generate a query:

  1. Add a query on the query panel.
  2. Open the AI chat in the App Builder.
  3. Type a prompt describing the query you need by referencing the specific query.
    tooljet generate query
  4. The AI generates the query with all the parameters and required fields.

You can use @mention in AI Chat to reference an existing query and ask the AI to modify it.

Generating Code​

ToolJet lets you generate JavaScript and Python code snippets directly using AI.

  1. Click the AI icon in the query panel to open the AI code generator.
    tooljet generate code
  2. Enter a prompt describing the logic or query you want to build.
    tooljet generate code
  3. Check the generated code and either insert it directly or click Regenerate to try a different version.
    tooljet generate code

Auto-sort Queries​

As applications grow, queries can accumulate without clear organization. The Auto-sort feature groups your queries into folders automatically based on their names.

To use Auto-sort:

  1. Open the query panel.
  2. Click the Auto-sort unsorted queries into folders option.
    tooljet auto-sort queries
  3. The AI analyzes your query names and organizes them into folders.

How it works:

  • If folders already exist in your app, queries are mapped to the closest matching folder where applicable, and new folders are created for the rest.
  • If no folders exist, new folders are created based on query name groupings.
  • Only query names are sent to the AI - no query content or data source credentials are shared, for compliance reasons.
  • Results depend on descriptive query names. Generic names like postgres_1, postgres_2 will produce less precise groupings.