A global software company's support ticket system receives messages in dozens of languages and must automatically identify which language each incoming ticket is written in before deciding which regional support queue and translation pipeline to route it to. Which AI capability should run first in this pipeline?

The correct choice is language detection, which identifies the natural language a piece of text is written in. Before the system can route a ticket or hand it to a translation pipeline, it first needs to know which of the dozens of possible languages the ticket is written in, and identifying that language is exactly what language detection does. The text translation option is incorrect because it converts a passage from one identified language into another, so it needs to already know the source language before it can convert the passage into another one; it is a downstream step that depends on language detection having run first, not a substitute for it. The key phrase extraction option is incorrect because it pulls the main topics out of a passage of text, such as 'refund' or 'login error', but it does not tell the system what language those topics are expressed in, so it cannot drive routing decisions on its own. The sentiment analysis option is incorrect because it scores a passage of text on a positive-to-negative scale, which might help prioritize an angry customer's message but says nothing about which language the message was written in. Language detection is the necessary first step because every downstream routing and translation decision depends on knowing the ticket's language.