---
title: "Your First Query - Documentation"
description: "Learn how to write effective natural-language questions and understand how AI for Database translates them into SQL."
canonical: "https://www.aifordatabase.com/docs/first-query/"
---

# Your First Query

beginner 6 min read

## Writing Effective Questions

AI for Database uses large language models to translate your natural-language questions into SQL. The better your question, the better the result. This guide teaches you how to get the most out of every query.

### How It Works

When you type a question, AI for Database:

1.  1**Reads your database schema** -- table names, column names, relationships, and sample data types.
2.  2**Translates your question** into a SQL query using an LLM (GPT-4 or Claude, depending on your configuration).
3.  3**Executes the query** against your live database using a read-only connection.
4.  4**Returns results** as a table, chart, or single value depending on the shape of the data.

You can always click **View SQL** to see the generated query and edit it if needed.

### Tips for Better Questions

**Be specific with time ranges.** Instead of "Show me sales", try "Show me total sales for each month in 2024." Time ranges help the AI pick the right date columns and aggregations.

**Name your tables or entities.** If your database has both a `users` and `customers` table, specify which one you mean: "How many customers signed up this quarter?" is clearer than "How many people signed up?"

**Ask for comparisons.** AI for Database handles comparisons well: "Compare revenue this month vs. last month by product category."

**Use follow-up questions.** After an initial result, you can refine: "Now break that down by region" or "Only show rows where the amount is above $1,000."

### Understanding Results

Results are displayed in the format that best fits the data:

*   **Single values** (e.g., "What is the total revenue?") appear as a large number card.
*   **Tables** appear when you ask for lists or detailed records.
*   **Charts** are generated automatically when the data has a natural x-axis (dates, categories). You can switch between bar, line, and pie charts.

### Handling Errors

If the AI generates a query that fails, you will see the error message along with the generated SQL. Common fixes:

*   Check that you are referencing the correct table or column name.
*   Add more context to your question.
*   Try rephrasing: "What is the count of orders grouped by status?" is more explicit than "Show me order statuses."

### Saving Queries

Click **Save** to store the query in your history. You can re-run saved queries at any time, and they will always use live data. Saved queries can also be added to dashboards or used as the basis for alerts.

Source: https://www.aifordatabase.com/docs/first-query/