How to Use Notion AI to Generate SQL Query From Plain English

Writing SQL queries from scratch requires remembering syntax, table names, and join conditions. Notion AI can translate plain English descriptions into working SQL code inside any Notion page. This saves time for product managers, data analysts, and developers who need quick database queries without switching tools. This article explains how to set up Notion AI, write effective prompts, and generate accurate SQL queries.

How Notion AI Generates SQL From Natural Language

Notion AI uses a large language model trained on code and documentation. When you type a prompt like “show all customers who bought in the last 30 days,” the AI identifies the key elements: the table (customers), the action (show), the filter (last 30 days), and the column (purchase date). It then constructs a SQL query using standard syntax.

The AI does not have access to your actual database schema. It relies entirely on the information you include in the prompt. If you omit table names or column names, the AI guesses, which may produce incorrect queries. For best results, include the database type (MySQL, PostgreSQL, SQL Server, or SQLite), table names, column names, and the desired output format.

What You Need Before Using Notion AI for SQL

You need a Notion workspace with AI capabilities enabled. Notion AI is available on the Plus, Business, and Enterprise plans, and as a paid add-on for the Free plan. You also need a clear description of the query you want. If you have a specific database schema, write down the table and column names so you can paste them into the prompt.

Steps to Generate a SQL Query Using Notion AI

1. Open a Notion page and place the cursor where you want the SQL code
   Navigate to any existing page or create a new page using the + New Page button in the sidebar. Click inside the page body to position the blinking cursor at the exact location where the generated SQL query should appear.
2. Activate Notion AI
   Press the Space key on your keyboard, then select Ask AI to write from the pop-up menu. Alternatively, type a forward slash / and choose AI > Ask AI to write. The AI input box opens at the cursor position.
3. Write a detailed plain English prompt
   In the AI input box, type a prompt that includes the database type, table name, columns, conditions, and sorting. Example: “Write a MySQL query to select first_name, last_name, and email from the customers table where signup_date is after 2023-01-01. Order by last_name ascending.” Be as specific as possible.
4. Press Enter to generate the SQL query
   After typing the prompt, press Enter. Notion AI processes the text and inserts the generated SQL code onto the page. The code appears in a code block with syntax highlighting.
5. Review and edit the generated SQL
   Read the query line by line. Check table names, column names, JOIN conditions, and WHERE clauses. If the query uses aliases, verify they match your schema. Edit any incorrect parts directly in the code block.
6. Test the query in your database client
   Copy the SQL code from Notion and paste it into your database management tool such as MySQL Workbench, pgAdmin, or DBeaver. Run the query against a test or staging database to confirm it returns the expected results. Never run generated SQL on a production database without testing.

Example Prompts for Common SQL Patterns

The following examples show prompts that produce reliable SQL queries:

• Simple SELECT: “Write a PostgreSQL query to select product_name and price from the products table where price is greater than 50.”
• JOIN between two tables: “Write a SQL Server query to select order_id, customer_name, and order_date from the orders table joined with the customers table on customer_id. Show only orders placed in 2024.”
• GROUP BY with aggregation: “Write a MySQL query to count the number of orders per customer. Show customer_id and order_count. Group by customer_id.”
• Subquery: “Write a PostgreSQL query to select products where the price is higher than the average price of all products. Use a subquery.”

Common Mistakes and Limitations When Using Notion AI for SQL

Generated SQL uses wrong table or column names

Notion AI does not know your database schema. If you write a vague prompt like “show user data,” the AI may invent table names such as “users” or “user_data” that do not exist in your database. To avoid this, always include the exact table and column names from your schema in the prompt. If you have many columns, list the ones you need explicitly.

AI produces syntax errors for specific database dialects

Notion AI may default to generic SQL syntax. For example, it might use TOP instead of LIMIT for SQL Server, or miss PostgreSQL-specific operators like ILIKE. Always specify the database type at the beginning of the prompt. After generation, double-check dialect-specific syntax such as date functions, string concatenation, and limit clauses.

Complex queries with multiple CTEs or window functions fail

Notion AI handles simple and moderately complex queries well, but it struggles with multi-level common table expressions, recursive queries, and advanced window functions. For these cases, break the query into smaller parts. Generate each part separately and combine them manually in your database client.

Notion AI Free vs Plus vs Business: SQL Generation Limits Compared

Item	Free Plan (AI Add-on)	Plus / Business / Enterprise Plans
AI response limit	30 responses per member per month	Unlimited responses
Maximum prompt length	Approximately 2,000 characters	Approximately 2,000 characters
Code block output	Supported — SQL appears in a code block	Supported — SQL appears in a code block
Cost	$10 per member per month	Included in plan cost

If you generate SQL frequently, the Free plan’s monthly limit may be insufficient. Consider upgrading to a Plus or Business plan to remove the cap. All plans support the same AI features, including code generation and syntax highlighting.

After generating a query, you can ask Notion AI to explain the SQL with a follow-up prompt such as “Explain this query step by step.” This helps you understand the generated code and learn SQL patterns. You can also use the Ask AI to edit option to modify an existing query, for example by adding a WHERE clause or changing the sort order.