OpenAI Intent Detection

Discover how Flexito OmniAI's seamless integration with OpenAI enhances your chatbot's capabilities by detecting user intent and collecting relevant parameters effortlessly. This powerful feature streamlines interactions, ensuring your bot responds accurately and efficiently to user queries.

Flexito OmniAI now features an intent detection system, thanks to its integration with OpenAI. This system allows users to gather parameters based on the intent of a query, functioning similarly to Google's Dialogflow by creating intents and populating entities.

To use this feature, you must integrate your OpenAI account with Flexito OmniAI. Let's walk through the process of establishing this connection.

Connecting Your OpenAI Account

Here's how to link your OpenAI account with your Flexito OmniAI account.

Go to

Log in with your credentials.

Click the “Personal” tab in the top-right corner.

Here, you can generate an API key.

[Note: You can view your API key only once.]

Enter your API key into Flexito OmniAI and click “Save” to complete the connection.

Your account is now successfully connected to Flexito OmniAI.

Utilising Intents

To use intents, navigate to the Automations tab within your bot.

In automations, the bot prioritises recognising intents first, followed by keywords, and finally the default reply.

Creating Intents

Activate the “Auto Intent Detect” slider. If you encounter an error, ensure your OpenAI account is properly integrated.

Next, click “+ New Intent” to proceed.

In the creation window, you'll find the following settings.

Click “+Add Parameter” to create a new parameter.

The List of Values tab offers users options to choose from, useful for setting limits on data received from the intent's reply.

Add values by clicking on the “Values” tab and entering likely values for the intent.

Finally, select a subflow to trigger based on the intent's status.

Clearing User Fields After Completing an Intent

Once an intent cycle concludes, Flexito OmniAI saves the intent parameter values in custom user fields. If the user interacts with the bot again, these old values are retrieved.

It's crucial to clear user fields after each intent cycle. The intent system checks parameter values, and if old values exist, they are automatically marked as complete.

You can automate this by saving intents directly in JSON CUF instead of individual custom fields, then adding an action node at the end of your subflow to clear the values.

How Intents Work Within the Bot

Intents are triggered similarly to keyword or comment automations. When a user types a question or statement, the Intent feature evaluates the sentence's intent based on keywords and the confidence score you've set.

Once an intent matches successfully, it collects parameter values until all are filled before triggering the final attached flow.

For example, if configured, a statement can trigger an appointment-setting intent. This demonstrates how the intent system functions in a live bot.

Token Utilisation

The intent detection feature uses tokens based on the number of parameters and the confidence score required. More parameters and a higher confidence score result in higher token consumption.