Intent Classify
The Intent Classify Node is like a guide that listens to what users say and decides where the conversation should go next. It uses few-shot prompting with OpenAI to acheive the results.
The user desires are called “intents”.
In BotStacks, each intent has a name and examples of how a user might express it, like “I want to order pizza” or “I want to order a burger.”
We call these examples I want to order pizza utterances. This helps the assistant recognize when users are talking about different things using NLU (Natural Language Understanding).
The Intent Classify Node works like a traffic signal, directing the conversation flow based on the recognized intent. It has different paths, each one leading to a different action or response based on what the user wants.
When a user sends a message, the assistant tries to match their words to one of the intents. Once it finds the right one, it follows the path connected to that intent, like following a road on a map.
This helps keep conversations organized and ensures that users get the right response to their requests. So, if someone asks to order pizza, the assistant knows to guide them through the pizza ordering process.