surface↓summarization↓condense
Summarizationscalar · returns varchar

CONDENSE

Condense individual text into brief summary (scalar, per-row)

Per-row — runs once for each row.

summarizationllmtext

Syntax

CONDENSE({{ text }})
TLDR({{ text }})
CONDENSE({{ text }}, '{{ focus }}')
TLDR({{ text }}, '{{ focus }}')

Arguments

nametypedescription
textVARCHARText to condense
focus(optional)VARCHAROptional focus hint (e.g., 'focus on pricing', 'key takeaways only')

About

Scalar text summarization - condenses individual texts into brief summaries. Unlike SUMMARIZE (aggregate that summarizes a collection), CONDENSE operates per-row on single texts. Perfect for: - Condensing long reviews/articles into short summaries - Creating executive summaries of documents - Generating TL;DR for verbose content - Extracting key points from lengthy text Supports optional focus hint to guide summarization (e.g., 'focus on pricing', 'highlight technical issues', 'extract action items'). Returns: Concise summary (typically 1-3 sentences unless text is very long)

Examples

Text condensed

SELECT
  semantic_condense (
    'This is a very long text that goes on and on with many words and sentences that could be summarized into something much shorter and more concise'
  )

Nearby rabbit holes

same domain
aggregate

CONSENSUS

Finds common ground among texts via centrality + LLM summary

summarizationllmscales-largejson
aggregate

CONSENSUS_LLM

LLM-only consensus (escape hatch for CONSENSUS)

summarizationllmllm-escape-hatchjson
aggregate

MERGE_TEXTS

Combine multiple text values into one coherent output

summarizationllmtext
aggregate

SUMMARIZE

Summarize a group of texts into one concise overview

summarizationllmtext
scalar

SUMMARIZE_URLS

Extracts URLs from text, fetches with browser, returns summary

summarizationllmurl
aggregate

THEMES

Extract N main topics from texts (embed centrality + LLM naming)

summarizationllmscales-largejson
Climb back to The Looking Glass