Chat AI

Owned by Olesya Tantsyura
Aug 07, 2024
2 min read

1. Description

The component is used to work with a Machine Learning model.

This component is present in flow types like:

  • Chat.

The block (Fig.1(1)) is used in space to build the Flow. The settings panel for this component opens (Fig.1(2)) when you click on the block.

Chat AI.png
Fig.1. Chat AI component

1.1. Chat AI block

The block consists of the following elements:

  1. The name of the component;

  2. in branch — receiving (connecting with the previous block) branch;

  3. out branch — outgoing (connecting with the following block) branch.

1.2. Chat AI settings panel

It consists of the following elements:

  1. Set variables;

  2. Connection;

  3. Model name;

  4. Categories;

  5. Response;

  6. Defined category;

  7. History length;

  8. Timeout;

  9. Break;

  10. Limit.

1.2.1. Set variables

It allows assigning an additional variable that can be used later.

It consists of the following:

  1. Key;

  2. Value;

  3. Remove tool;

  4. Add.

1.2.1.1. Key

The name of the variable is entered here.

1.2.1.2. Value

The value of the variable is entered here.

1.2.1.3. Remove tool

It is intended to remove a variable in a row with which it stands — clicking the button removes it.

1.2.1.4. Add

It is designed to add a new variable.

1.2.2. Connection

The connection address of the machine learning model server is entered here. If the address is entered, the connection will be made to it. If not, the value is taken from the system variable.

1.2.3. Model name

The name of the model to be used is entered here.

1.2.4. Categories

The ability to enter a list of data categories based on which the model will analyse Subscriber messages and provide a response.

1.2.5. Response

The name of the variable in which the text of the response received from the model will be written is entered here.

1.2.6 Defined category

The variable name is entered here. This variable will contain the name of the category into which the machine learning model classified the message. The category can be the one with the highest percentage of matches.

1.2.7 History length

The number of recent chat messages that will be sent to the model for analysis is entered here.

1.2.8. Timeout

The timeout for receiving a response from the model is entered here. In seconds.

1.2.9. Break

1.2.10. Limit

Fig. 2. Setting a limit