Intents

Section to manage the intents that your App needs to be able to recognise

Only the Producer and the Developer roles are allowed to manage intents.

When automating human conversation, understanding of the user's intent is key. Giving the right answer to the wrong question is the biggest mistake made by bots today, so identifying the intention of the user message is the biggest challenge.

The DialoX platform offers a special section to manage all the intents that are relevant for your app.

Intents section

Scoping your intents

The system differentiates between global and local intents. When creating or editing an intent, you can set its scope by toggling the 'Only use locally' switch.

  • Local intents Local intents can be recognized by the circle icon in front of their name (see "done" and "talk_to_someone" in the example above). They are specific to a certain dialog or question (prompts or asks). For example, when someone says "Yes", "No", or "8:00 PM". These kinds of replies are always tied to a question or part of a dialog, and are not acted upon without any context. Local intents can be used in specific Ask nodes in your flow. This can be in labelled, or dialog trigger links:

Local intents are used in Ask or Prompt nodes as the intents that trigger the right links to other flow nodes
  • Global intents Global intents are used in intent triggers. They can be picked up from any point in a conversation, even if another topic is being discussed. For example, if a caller asks, "What was the name of your company again?", you can configure the Smart App to respond and then continue the previous conversation:

Global intents are used in intent triggers

Intent providers

The DialoX platform supports multiple AI intent providers for classifying user intents. The three most commonly used providers are:

  • LLM: using the OpenAI ChatGPT 4o mini model to classify intents using mainly a label and description (and optionally positive example utterances)

  • Q&A: using the opensource library called Bert to classify intents using positive and negative example utterances.

  • Google Dialogflow: advanced DialoX users can connect their own Dialogflow agent

The LLM provider is the most commonly used model that requires no training at all, just a label and a description. It outperforms the Q&A provider as well as Google Dialogflow and is therefore the default choice.

Switching between Q&A and LLM intents

As of right now you still have the opportunity to switch back and forth between the legacy Q&A intents and the new LLM intents.

When switching to LLM intents, we automatically generate a description for each intent in case there was none yet. As mentioned before, the description plays the lead role in classifying the intent.

Click Upgrade to LLM to make use of the new LLM intents.

If you want to revert to the legacy Q&A intents, click on the "revert" icon:

Click the rightmost button to switch back to Q&A intents

Last updated

Was this helpful?