Short answer
An utterance is a single thing a customer says or types, the raw input a conversational AI system reads and interprets to determine what the customer wants."Where is my order?" is an utterance. So is a three-sentence email describing a damaged product. An utterance is the surface form, the actual words. The intent is the meaning underneath it. Many different utterances map to the same intent: "where's my package," "tracking isn't updating," and "did this even ship" all express order status.
This many-to-one relationship is the whole reason intent classification exists. People phrase the same need in endless ways. A system that keys off the literal utterance, matching words and phrases, is fragile. A system that maps utterances to intent is durable.
Aide, the AI platform for customer service, builds its Customer Intent Map from real utterances mined from a brand's own conversation history, not from a guessed list of phrases. It learns the many ways customers actually say each thing, so coverage reflects real demand rather than what someone imagined customers might type. An unfamiliar or low-confidence utterance goes to a human instead of being forced into the nearest intent, and the team keeps seeing the real phrasings, so its read on how customers talk stays current.