SQL Keywords

AI_CLASSIFY

Feb 23, 2026

min read

AI_CLASSIFY

Overview

Classifies text or images into user-defined categories. Supports both single-label and multi-label classification.

Syntax

AI_CLASSIFY(
  input,
  categories,
  [options]
)

AI_CLASSIFY(
  input,
  categories,
  [options]
)

AI_CLASSIFY(
  input,
  categories,
  [options]
)

Parameters

input (VARCHAR or FILE): Text string or image file reference
categories (ARRAY): List of category names to classify into
options (OBJECT): Optional settings like multi_label, examples, descriptions

Use Cases

Customer feedback categorization
Content moderation
Email routing and prioritization
Product categorization
Image classification
Sentiment classification

Code Examples

Example 1: Simple Text Classification

SELECT AI_CLASSIFY(
  'I dream of backpacking across South America.',
  ['travel', 'cooking', 'technology', 'sports']
) AS

SELECT AI_CLASSIFY(
  'I dream of backpacking across South America.',
  ['travel', 'cooking', 'technology', 'sports']
) AS

SELECT AI_CLASSIFY(
  'I dream of backpacking across South America.',
  ['travel', 'cooking', 'technology', 'sports']
) AS

Output:

travel

travel

travel

Example 2: Multi-Label Classification

SELECT AI_CLASSIFY(
  'This laptop has amazing battery life and a beautiful display.',
  ['performance', 'design', 'battery', 'price'],
  {'multi_label': true}
) AS

SELECT AI_CLASSIFY(
  'This laptop has amazing battery life and a beautiful display.',
  ['performance', 'design', 'battery', 'price'],
  {'multi_label': true}
) AS

SELECT AI_CLASSIFY(
  'This laptop has amazing battery life and a beautiful display.',
  ['performance', 'design', 'battery', 'price'],
  {'multi_label': true}
) AS

Output:

["performance", "design", "battery"]

["performance", "design", "battery"]

["performance", "design", "battery"]

Example 3: Classification with Descriptions

SELECT AI_CLASSIFY(
  customer_feedback,
  ['bug', 'feature_request', 'complaint', 'praise'],
  {
    'descriptions': {
      'bug': 'Technical issue or error',
      'feature_request': 'Request for new functionality',
      'complaint': 'Negative feedback about existing features',
      'praise': 'Positive feedback'
    }
  }
) AS feedback_type
FROM

SELECT AI_CLASSIFY(
  customer_feedback,
  ['bug', 'feature_request', 'complaint', 'praise'],
  {
    'descriptions': {
      'bug': 'Technical issue or error',
      'feature_request': 'Request for new functionality',
      'complaint': 'Negative feedback about existing features',
      'praise': 'Positive feedback'
    }
  }
) AS feedback_type
FROM

SELECT AI_CLASSIFY(
  customer_feedback,
  ['bug', 'feature_request', 'complaint', 'praise'],
  {
    'descriptions': {
      'bug': 'Technical issue or error',
      'feature_request': 'Request for new functionality',
      'complaint': 'Negative feedback about existing features',
      'praise': 'Positive feedback'
    }
  }
) AS feedback_type
FROM

Example 4: Classification with Examples

SELECT AI_CLASSIFY(
  review_text,
  ['positive', 'negative', 'neutral'],
  {
    'examples': {
      'positive': ['Great product!', 'Love it!'],
      'negative': ['Terrible experience', 'Do not buy'],
      'neutral': ['It works', 'As expected']
    }
  }
) AS sentiment
FROM

SELECT AI_CLASSIFY(
  review_text,
  ['positive', 'negative', 'neutral'],
  {
    'examples': {
      'positive': ['Great product!', 'Love it!'],
      'negative': ['Terrible experience', 'Do not buy'],
      'neutral': ['It works', 'As expected']
    }
  }
) AS sentiment
FROM

SELECT AI_CLASSIFY(
  review_text,
  ['positive', 'negative', 'neutral'],
  {
    'examples': {
      'positive': ['Great product!', 'Love it!'],
      'negative': ['Terrible experience', 'Do not buy'],
      'neutral': ['It works', 'As expected']
    }
  }
) AS sentiment
FROM

Example 5: Image Classification

SELECT 
    file_name,
    AI_CLASSIFY(
        TO_FILE('@image_stage/' || file_name),
        ['indoor', 'outdoor', 'portrait', 'landscape']
    ) AS image_category
FROM DIRECTORY('@image_stage')
WHERE file_name LIKE '%.jpg'

SELECT 
    file_name,
    AI_CLASSIFY(
        TO_FILE('@image_stage/' || file_name),
        ['indoor', 'outdoor', 'portrait', 'landscape']
    ) AS image_category
FROM DIRECTORY('@image_stage')
WHERE file_name LIKE '%.jpg'

SELECT 
    file_name,
    AI_CLASSIFY(
        TO_FILE('@image_stage/' || file_name),
        ['indoor', 'outdoor', 'portrait', 'landscape']
    ) AS image_category
FROM DIRECTORY('@image_stage')
WHERE file_name LIKE '%.jpg'

Data Output Examples

Support Ticket Categorization

Input: "The app crashes every time I try to upload a file"
Categories: ['bug', 'feature_request', 'question', 'feedback']
Output: "bug"

Input: "The app crashes every time I try to upload a file"
Categories: ['bug', 'feature_request', 'question', 'feedback']
Output: "bug"

Input: "The app crashes every time I try to upload a file"
Categories: ['bug', 'feature_request', 'question', 'feedback']
Output: "bug"

E-commerce Product Classification

Input: "Wireless Bluetooth headphones with noise cancellation"
Categories: ['electronics', 'clothing', 'home', 'sports']
Output: "electronics"

Input: "Wireless Bluetooth headphones with noise cancellation"
Categories: ['electronics', 'clothing', 'home', 'sports']
Output: "electronics"

Input: "Wireless Bluetooth headphones with noise cancellation"
Categories: ['electronics', 'clothing', 'home', 'sports']
Output: "electronics"

Model Information

Model Used: Snowflake managed model
Context Window: 128,000 tokens
Supported Inputs: Text strings and images

Limitations & Considerations

Category Limits

Maximum 500 categories per classification
Categories, descriptions, and examples count as input tokens
Billed per record processed, not per batch

Cost

Input tokens include: text + all categories + descriptions + examples
Processed for EACH record, not once per query
Use concise category names and descriptions

Accuracy Tips

Provide clear, distinct category names
Use descriptions for ambiguous categories
Add examples for better accuracy (2-3 per category)
Avoid overlapping categories

Regional Availability

AWS US West/East: ✓
Azure East US: ✓
EU regions: ✓
Cross-region inference: ✓

Best Practices

1. Use Descriptions for Clarity

-- Better with descriptions
AI_CLASSIFY(
  text,
  ['urgent', 'normal', 'low'],
  {'descriptions': {
    'urgent': 'Requires immediate attention within 24 hours',
    'normal': 'Standard processing time 2-5 days',
    'low': 'Can be addressed when convenient'
  }}
)

-- Better with descriptions
AI_CLASSIFY(
  text,
  ['urgent', 'normal', 'low'],
  {'descriptions': {
    'urgent': 'Requires immediate attention within 24 hours',
    'normal': 'Standard processing time 2-5 days',
    'low': 'Can be addressed when convenient'
  }}
)

-- Better with descriptions
AI_CLASSIFY(
  text,
  ['urgent', 'normal', 'low'],
  {'descriptions': {
    'urgent': 'Requires immediate attention within 24 hours',
    'normal': 'Standard processing time 2-5 days',
    'low': 'Can be addressed when convenient'
  }}
)

2. Optimize for Cost

-- Use fewer, well-defined categories
-- Avoid: ['cat1', 'cat2', 'cat3', ... 'cat100']
-- Better: ['primary', 'secondary', 'other']

-- Use fewer, well-defined categories
-- Avoid: ['cat1', 'cat2', 'cat3', ... 'cat100']
-- Better: ['primary', 'secondary', 'other']

-- Use fewer, well-defined categories
-- Avoid: ['cat1', 'cat2', 'cat3', ... 'cat100']
-- Better: ['primary', 'secondary', 'other']

3. Multi-Label When Appropriate

-- For items that can belong to multiple categories
SELECT AI_CLASSIFY(
  'Fitness tracker watch with heart rate monitor',
  ['electronics', 'fitness', 'wearables', 'health'],
  {'multi_label': true}
);
-- Output: ["electronics", "fitness", "wearables", "health"]

-- For items that can belong to multiple categories
SELECT AI_CLASSIFY(
  'Fitness tracker watch with heart rate monitor',
  ['electronics', 'fitness', 'wearables', 'health'],
  {'multi_label': true}
);
-- Output: ["electronics", "fitness", "wearables", "health"]

-- For items that can belong to multiple categories
SELECT AI_CLASSIFY(
  'Fitness tracker watch with heart rate monitor',
  ['electronics', 'fitness', 'wearables', 'health'],
  {'multi_label': true}
);
-- Output: ["electronics", "fitness", "wearables", "health"]

Related Functions

AI_FILTER - For binary true/false classification
AI_SENTIMENT - Specialized for sentiment analysis
AI_COMPLETE - For more complex classification tasks

Interested to Learn More?
Try Out the Free 14-Days Trial

Start free trial

Learn

Feb 18, 2026

BigQuery Global Queries: How to Run Cross-Region SQL in 2026

Learn

Feb 18, 2026

BigQuery Global Queries: How to Run Cross-Region SQL in 2026

Learn

Feb 18, 2026

BigQuery Global Queries: How to Run Cross-Region SQL in 2026

Analytics

Feb 18, 2026

Context Engineering and AI Quality for Data Teams

Analytics

Feb 18, 2026

Context Engineering and AI Quality for Data Teams

Analytics

Feb 18, 2026

Context Engineering and AI Quality for Data Teams

Product

Feb 18, 2026

Accelerate Analytics Development with Paradime and Tableau

Product

Feb 18, 2026

Accelerate Analytics Development with Paradime and Tableau

Product

Feb 18, 2026

Accelerate Analytics Development with Paradime and Tableau

Experience Analytics for the AI-Era

Start your 14-day trial today - it's free and no credit card needed

Start for free

Experience Analytics for the AI-Era

Start your 14-day trial today - it's free and no credit card needed

Start for free

Experience Analytics for the AI-Era

Start your 14-day trial today - it's free and no credit card needed

Start for free

Platform

Resources

ADD-ONs

Industries

About

Legal

Made with ❤️ in San Francisco ・ London

*dbt® and dbt Core® are federally registered trademarks of dbt Labs, Inc. in the United States and various jurisdictions around the world. Paradime is not a partner of dbt Labs. All rights therein are reserved to dbt Labs. Paradime is not a product or service of or endorsed by dbt Labs, Inc.

Start for free

Platform

Resources