Written by:
Naimh
Last updated: 11/08/2026 14:41:30

AI features can understand the map’s structure, including its layers, column names, and data types, but no row data or geometry is provided by default. Map authors can optionally allow limited summaries from specific columns through Attribute config > AI Hints

Alert

## EU-hosted AI with controlled data access AI features use open-weight models hosted within the EU on ISO 27001-certified infrastructure. AI access is deliberately limited: by default, tools receive map structure only. Attribute data is available only where a map author explicitly enables AI Hints.

1

Click Adjust Layer data config.

Adjust popup config on Planning NI Final_Current | Azimap
Alert

What AI can see by default When an AI tool is used, it may receive the map structure, including available layers, column names, and column types. Attribute values, table rows, and geometry are not included by default.

2

Click the 'AI HINTS' button to show controls for sharing summary data with AI tools..

Button labeled 'AI HINTS' in Attribute Configuration section
Tip

What AI Hints can share A map author can explicitly approve individual columns. For approved text columns, up to 50 unique values may be provided. For numeric columns, only summary values such as minimum and maximum/ range are provided. Full rows and geometry are never sent.

Alert

No approved columns = no attribute data if no columns are enabled in AI Hints, the AI receives the map structure only, not the underlying attribute data.

3

Check the values you want to share.

the highlighted button on Planning NI Final_Current | Azimap
Tip

Only approve what is useful Enable AI Hints only for the columns needed to support the AI feature. Columns that remain disabled do not provide values or summaries.

4

Click Save Changes.

Save Changes on Planning NI Final_Current | Azimap
Alert

AI does not directly control the application The AI returns structured JSON only. The application parses and validates that response before running any corresponding application logic. AI proposes; the application decides. The AI itself does not directly run code or perform actions.