Quick start with AI Assistance

Note: This feature is available in Enterprise and AWS editions only.

Table of contents

• Start in the SQL Editor
• Generate directly in the editor with @ai
• Understand what the query is doing
• Fix an error when the query fails
• Describe a database object when you need context

Work through a typical flow in the SQL editor using AI assistant - from writing a query, to fixing an error, to understanding objects.

To get started, activate AI Assistant, open Settings -> Administration -> AI Settings, choose a provider, and paste your API token.

Tip: Different vendors have different limits, models, and regions. Pick what fits your data policy and budget.

Start in the SQL Editor

1. Open the SQL editor.

2. Select Tools -> AI Chat from the top toolbar.

3. Pick a connection.

   

   Optionally set a scope to improve accuracy on large instances.

4. Type what you need in AI chat.

5. Execute or refine.

For details, see AI chat.

Generate directly in the editor with @ai

Want to stay focused in the SQL editor? Use @ai when you don’t need the full chat flow.

1. In the SQL editor, enter @ai your request...
2. Execute the command. The results will appear in the Result panel.

For details, see AI command.

Understand what the query is doing

Need a clear explanation, risks, or edge cases?

1. Select the SQL

2. Click Explain Query with AI icon () located in the left toolbar of the SQL Editor.

   

3. Read the explanation in AI chat, and ask follow-ups

For details, see Explain query using AI.

Fix an error when the query fails

If the SQL fails, use AI to explain and repair it in one pass.

1. Click Explain and Fix

   

2. Apply the fix

   

For the flow, see AI error explanation.

Describe a database object when you need context

Ask AI to describe a table, view, or other object in plain language. The result opens in AI chat.

1. Open Database Navigator or Properties editor

2. Click the cell context menu (three-bar icon)

3. Select Describe object with AI from the context menu.

4. Review the description in AI chat, and ask follow-ups

   

For supported objects and flow, see Describe database object.