Analyze your database with AI
Enriching your data with semantic metadata is a core part of how Toucan AI helps you move from raw data to actionable insights, faster.
This page explains how to analyze your database tables with AI to enrich them with semantic metadata. This metadata helps Toucan AI better understand your data when generating charts, dashboards, and AI-driven responses.
Overview
The Analyze feature enriches the tables in a database schema by generating descriptions and semantic field types.
By adding this context, Toucan AI can:
Interpret columns more accurately
Generate more relevant queries
Produce clearer charts and explanations
Analysis is performed per schema and applies to all tables in that schema.
What the AI Analysis Does
When you click Analyze, Toucan AI:
Analyzes one schema at a time
Generates a description for each table
Generates a description for each column
Infers a semantic field type for each column
Field types are inferred independently from the database’s technical data types.
Generated Metadata
Table Description
For each table, the AI generates a short description explaining what the table contains.
Example:
Provides geographic and regional information about US states
You can edit this description manually at any time.
Column Metadata
For each column, Toucan AI generates:
Display name A human-readable name used in charts and dashboards.
Field type One of the following semantic types:
Category – dimensions or grouping fields
Text – free-text or descriptive content
Date – dates or timestamps
Metric – numerical values used for calculations
Description A short explanation of what the column represents.
You can manually adjust all generated values.
Table Views
Each table provides three main tabs.
Structure
The Structure tab displays:
Table description
Column list
Display names
Data types (from the database)
Semantic field types
Column descriptions
This is where AI-generated metadata is reviewed and edited.
Preview
The Preview tab shows the first rows of the table.
This preview:
Displays a limited sample of data
Helps validate values and column meaning
Is intended only as a visual hint
Toucan AI is not designed for data exploration or full data inspection. The preview is intentionally limited.
RLS (Row-Level Security)
The RLS tab allows you to define access rules on the table.
Using RLS, you can:
Restrict rows based on user attributes
Prepare datasets for secure embedding
Ensure users only see authorized data
RLS configuration relies on the table structure and metadata defined during analysis.
How to Analyze a Schema
Go to Databases.
Select a connected database.
Select a schema from the table list.
Click Analyze in the top-right corner.
Wait for the analysis to complete.
Review the generated metadata.
Click Save to persist your changes.
Limitations and Assumptions
AI analysis is based on schema structure and sampled values.
Field type inference may be incorrect for ambiguous columns.
Generated descriptions are suggestions and may require editing.
Only one schema can be analyzed at a time.
Schema changes in the database require re-running the analysis.
Data previews are limited and not meant for analysis or reporting.
Toucan AI assumes users validate and refine metadata before using it in production dashboards.
Summary
Analyzing your database with AI adds semantic meaning to your tables and columns. This step is essential to improve the quality of AI-generated charts, dashboards, and insights in Toucan AI.
Last updated