referenceJune 20, 2026· 3 min read

How to use CSIM: score how close two queries are

CSIM returns the semantic similarity between two strings, the primitive behind every hit decision, exposed so you can calibrate thresholds.

CSIM answers 'how alike are these two questions?' as a number. It's the fastest way to understand why something hit or missed, and to pick a threshold that fits your traffic.

crowkis cli

> CSIM "how do refunds work?" "what's the refund process?"
"0.93"

> CSIM "how do refunds work?" "how do I cancel my plan?"
"0.41"

High scores mean safe reuse for factual intents; calibrate against your own pairs with CSIM before trusting a global threshold.