Category: General-Purpose AI Function (Preview)

Description

Generates a boolean (true/false) value for each row based on a natural language prompt and data analysis. Provides full control over prompts and models while returning a simple BOOL output. Useful for row-by-row validation, filtering logic, or conditional checks.

Use Cases

Data Validation: Check if records meet specific criteria

Content Moderation: Determine if content is appropriate

Business Rules: Apply complex conditional logic

Quality Checks: Validate data quality or completeness

Compliance: Check if content meets regulatory requirements

Syntax

AI.GENERATE_BOOL(
  prompt => 'PROMPT',
  [, data => INPUT_DATA ]
  [, model => 'MODEL_ENDPOINT' ]
  [, parameters => STRUCT(...) ]
  [, connection_id => 'CONNECTION' ]
)

AI.GENERATE_BOOL(
  prompt => 'PROMPT',
  [, data => INPUT_DATA ]
  [, model => 'MODEL_ENDPOINT' ]
  [, parameters => STRUCT(...) ]
  [, connection_id => 'CONNECTION' ]
)

AI.GENERATE_BOOL(
  prompt => 'PROMPT',
  [, data => INPUT_DATA ]
  [, model => 'MODEL_ENDPOINT' ]
  [, parameters => STRUCT(...) ]
  [, connection_id => 'CONNECTION' ]
)

Parameters

prompt: Natural language instruction/question (required)

data: Text or ObjectRefRuntime values to analyze

model: Gemini model endpoint (e.g., 'gemini-2.0-flash-exp')

parameters: Generation config (temperature, max_tokens, etc.)

connection_id: Vertex AI connection

Code Examples

Example 1: Validate Product Descriptions

SELECT 
  product_id,
  product_name,
  description,
  AI.GENERATE_BOOL(
    prompt => 'Does this product description mention the materials used?',
    data => description,
    connection_id => 'us.my_vertex_connection'
  ) AS mentions_materials
FROM products
WHERE category = 'furniture'

SELECT 
  product_id,
  product_name,
  description,
  AI.GENERATE_BOOL(
    prompt => 'Does this product description mention the materials used?',
    data => description,
    connection_id => 'us.my_vertex_connection'
  ) AS mentions_materials
FROM products
WHERE category = 'furniture'

SELECT 
  product_id,
  product_name,
  description,
  AI.GENERATE_BOOL(
    prompt => 'Does this product description mention the materials used?',
    data => description,
    connection_id => 'us.my_vertex_connection'
  ) AS mentions_materials
FROM products
WHERE category = 'furniture'

Example 2: Content Appropriateness Check

SELECT 
  post_id,
  content,
  AI.GENERATE_BOOL(
    prompt => 'Is this social media post appropriate for a professional business audience?',
    data => content,
    connection_id => 'us.my_vertex_connection'
  ) AS is_appropriate
FROM social_media_posts
WHERE reviewed = FALSE

SELECT 
  post_id,
  content,
  AI.GENERATE_BOOL(
    prompt => 'Is this social media post appropriate for a professional business audience?',
    data => content,
    connection_id => 'us.my_vertex_connection'
  ) AS is_appropriate
FROM social_media_posts
WHERE reviewed = FALSE

SELECT 
  post_id,
  content,
  AI.GENERATE_BOOL(
    prompt => 'Is this social media post appropriate for a professional business audience?',
    data => content,
    connection_id => 'us.my_vertex_connection'
  ) AS is_appropriate
FROM social_media_posts
WHERE reviewed = FALSE

Example 3: Multi-Field Validation

SELECT 
  application_id,
  AI.GENERATE_BOOL(
    prompt => 'Based on this application, does the candidate meet the minimum requirements: 5+ years experience, Bachelor degree, and expertise in Python?',
    data => STRUCT(
      years_experience,
      education_level,
      skills,
      work_history
    ),
    connection_id => 'us.my_vertex_connection'
  ) AS meets_requirements
FROM

SELECT 
  application_id,
  AI.GENERATE_BOOL(
    prompt => 'Based on this application, does the candidate meet the minimum requirements: 5+ years experience, Bachelor degree, and expertise in Python?',
    data => STRUCT(
      years_experience,
      education_level,
      skills,
      work_history
    ),
    connection_id => 'us.my_vertex_connection'
  ) AS meets_requirements
FROM

SELECT 
  application_id,
  AI.GENERATE_BOOL(
    prompt => 'Based on this application, does the candidate meet the minimum requirements: 5+ years experience, Bachelor degree, and expertise in Python?',
    data => STRUCT(
      years_experience,
      education_level,
      skills,
      work_history
    ),
    connection_id => 'us.my_vertex_connection'
  ) AS meets_requirements
FROM

Example 4: Image Validation

SELECT 
  image_id,
  AI.GENERATE_BOOL(
    prompt => 'Does this image show a product with clear visibility and professional lighting?',
    data => OBJ.GET_ACCESS_URL(image_ref, 'r'),
    connection_id => 'us.my_vertex_connection'
  ) AS is_high_quality
FROM product_photos
WHERE status = 'pending_review'

SELECT 
  image_id,
  AI.GENERATE_BOOL(
    prompt => 'Does this image show a product with clear visibility and professional lighting?',
    data => OBJ.GET_ACCESS_URL(image_ref, 'r'),
    connection_id => 'us.my_vertex_connection'
  ) AS is_high_quality
FROM product_photos
WHERE status = 'pending_review'

SELECT 
  image_id,
  AI.GENERATE_BOOL(
    prompt => 'Does this image show a product with clear visibility and professional lighting?',
    data => OBJ.GET_ACCESS_URL(image_ref, 'r'),
    connection_id => 'us.my_vertex_connection'
  ) AS is_high_quality
FROM product_photos
WHERE status = 'pending_review'

Data Output Examples

Product Description Validation

Content Appropriateness

Best Practices

Ask clear yes/no questions: Frame prompts for binary outcomes

Provide context: Include relevant data fields for better accuracy

Test edge cases: Validate behavior with ambiguous inputs

Consider AI.IF for filtering: Use AI.IF in WHERE clauses for better optimization

Monitor false positives/negatives: Validate results periodically

When to Use

✅ Use when you need custom boolean logic beyond AI.IF

✅ Use for row-by-row validation or checks

✅ Use when you need full control over the prompt

✅ Use for complex multi-field validation

Alternatives

AI.IF: For filtering with managed optimization

Traditional CASE WHEN: For rule-based logic (cheaper)

AI.GENERATE: For structured output with multiple fields

Platform Support

Regions: All Gemini-supported regions + US/EU multi-regions

Models: Gemini 2.0 Flash, Gemini 1.5 Pro, Gemini 1.5 Flash

Preview Status: Currently in Preview (Pre-GA)

Cost: Charged per Vertex AI API call

Returns

BOOL value (true/false). Returns NULL if the Vertex AI call fails.