Natural Language to AQL in the Web Interface (AQLizer)

Beta

The AQLizer feature is integrated into the Query Editor of the Arango Contextual Data Platform. It uses generative AI to translate your plain language descriptions into AQL queries, so you can explore your data and gain insights without having to learn the query language first.

Access the AQLizer

1. In the left-hand sidebar of the Arango Contextual Data Platform, select the database you want to query.
2. Click Query Editor in the navigation.
3. The Query Editor opens. The AQLizer button is visible at the bottom of each query tab.

Set up AQLizer

Before you can generate queries, set up the AQLizer service. This is a one-time step.

1. In the Query Editor, click the AQLizer button at the bottom of the query tab.
2. The AQLizer panel opens. On first use, you are prompted to provide:
   • OpenAI API Key: Your API key for the LLM provider.
   • OpenAI Model: Select the model to use for query generation (for example, gpt-5.4).
3. Click Start Service to initialize the AQLizer.

The service is now ready to use. You can check the service status or stop it at any time from the Manage Services dialog.

Generate AQL queries

Once the AQLizer service is running:

1. Click the AQLizer button to open the AQLizer panel.
2. Enter a self-contained question or instruction in plain language. Be specific to get accurate results. For example:
   • “List all distinct surnames in the database sorted in descending order”
   • “How many persons are there with the surname ‘Stark’?”
   • “Find the parents of ‘Arya Stark’”
3. Click Ask or press Return .
4. The AQLizer generates an AQL query based on your input and your database schema.
5. Review the generated query in the AQLizer panel.
6. Click Open in Editor to load the query into a new query tab.
7. Verify the query, make any adjustments, and click Run query to execute it.

Tips for better results

• Be specific: The more precise your question, the more accurate the generated query. For example, instead of “show me users”, try “list all users who registered after January 2024, sorted by registration date”.
• Reference your data model: If you know your collection or attribute names, include them in your question.
• Iterate: Use the generated query as a starting point. You can refine it in the Query Editor and ask the AQLizer again with a more focused question.
• Use an AQL-capable model: The quality of the generated AQL depends on the LLM model. A model with strong coding capabilities generally produces better results.

For more information about the Query Editor and its other features, see the Query Editor documentation.

For programmatic access to the same functionality, see the API Reference.