AI_SUMMARIZE

Overview

AI_SUMMARIZE is an AI text summarization function provided by Singdata Lakehouse. It generates concise summaries of input text. Supports Chinese, English, Japanese, and other languages. Use the max_words parameter to control summary length — no prompt writing required. One line of SQL handles text summarization.

Singdata pushes AI computation down to the storage and execution engine layer. AI_SUMMARIZE can be called directly on any text column in a SQL query, freely combined with filters, aggregations, JOINs, and other operations — no need to export data to external systems.

Syntax

AI_SUMMARIZE(model, content [, max_words])

Parameters

Required Parameters

Parameter	Type	Description
`model`	STRING	Model identifier specifying the AI model to use for summarization
`content`	STRING	Text to summarize; supports CHAR, VARCHAR, STRING types

Optional Parameters

Parameter	Type	Default	Description
`max_words`	INT	50	Target word count for model output. Set to `0` to return the original text (no summarization); negative values cause an error

model Parameter Details

The model parameter supports two sources:

Source 1: API Gateway Endpoint (Recommended)

A platform administrator pre-configures model services in the API Gateway. Regular users reference them with the endpoint: prefix.

'endpoint:qwen3-max-preview'

Source 2: API Connection Object

CREATE API CONNECTION conn_bailian TYPE ai_function PROVIDER = 'bailian' BASE_URL = 'https://dashscope.aliyuncs.com/api/v1' API_KEY = 'sk-xxxxxxxxxxxxxxxxxxxxxxxx'; -- Reference format: <connection_name>:<model_name> SELECT AI_SUMMARIZE('conn_bailian:qwen3.5-plus', content, 30) AS summary FROM articles;

Return Value

Returns STRING containing a summary of the input text.

Returns NULL when input is NULL
Returns empty string '' when input is empty string ''
max_words is a target, not a hard limit; actual output may vary by ±20%

Error Behavior

Error scenario	Error message	Resolution
`max_words` is negative	`max_words must be non-negative, got: -1`	`max_words` must be ≥ 0
Endpoint does not exist	`CZLH-67000 No available endpoints found`	Check that the endpoint name is correct
Invalid model format (missing prefix)	`CZLH-65000 Invalid model coordinates`	Use `'endpoint:name'` or `'connection:model'` format
Missing required parameter	`CZLH-65000 AI function must have at least two arguments`	Ensure both `model` and `content` are provided

Usage Notes

model is required; omitting it causes the first string argument to be misinterpreted as the model name.
max_words is a target, not a hard limit. Actual output may vary by ±20% — do not rely on exact word counts in downstream processing.
When max_words=0, the function returns the original text without calling the model.
Output language automatically follows the input language; no additional specification needed.
For large tables, use WHERE to filter down to the rows that need processing first, avoiding unnecessary model calls.
LLM output is non-deterministic; the same input may produce slightly different results across executions.

Examples

Basic Usage

-- Specify max_words=15 SELECT AI_SUMMARIZE( 'endpoint:qwen3-max-preview', 'Singdata Lakehouse is a fully managed lakehouse architecture platform built from the ground up on cloud-native design principles, supporting real-time analysis of petabyte-scale data.', 15 ) AS summary;

Using Default Word Count

-- No max_words, defaults to approximately 50 words SELECT AI_SUMMARIZE( 'endpoint:qwen3-max-preview', 'Singdata Lakehouse is a fully managed lakehouse architecture platform built from the ground up on cloud-native design principles, supporting real-time analysis of petabyte-scale data.' ) AS summary;

English Text Example

-- English input returns an English summary SELECT AI_SUMMARIZE( 'endpoint:qwen3-max-preview', 'Singdata Lakehouse is an enterprise data platform supporting batch, streaming, and interactive analytics across multiple cloud environments.', 15 ) AS summary;

max_words=0 Returns Original Text

SELECT AI_SUMMARIZE( 'endpoint:qwen3-max-preview', 'Singdata Lakehouse is a fully managed lakehouse architecture platform.', 0 ) AS summary; -- Returns: Singdata Lakehouse is a fully managed lakehouse architecture platform.

Batch Processing Table Data

SELECT id, AI_SUMMARIZE('endpoint:qwen3-max-preview', review_content, 30) AS summary FROM customer_reviews WHERE review_content IS NOT NULL LIMIT 100;

Using an API Connection

SELECT AI_SUMMARIZE('conn_bailian:qwen3.5-plus', article_body, 50) AS abstract FROM news_articles;

Limitations

Item	Description
model parameter	Must use `'endpoint:name'` or `'connection:model'` format; cannot be omitted
Input length	Limited by the underlying model's context window (qwen3-max-preview: approximately 32K tokens)
Output length	`max_words` is a target; actual output may vary by ±20%
Aggregate summarization	No aggregate function version (e.g. summarizing multiple rows grouped by GROUP BY)
Model dependency	Requires a configured Endpoint in the AI Gateway, or a pre-created API Connection
Result determinism	LLM output is non-deterministic; the same input may produce slightly different results across executions