AI Chatbot – Plivo CX

Navigate to Channels → Chats → AI Chatbots.
Click on “Create” in the top right corner.
Provide the below details:
- AI Chatbot Name - Assign a friendly name for your AI Bot
- Tone - You can select a tone that will decide the manner or style in which a chatbot communicates with users. It can be empathetic, formal, or friendly.
- Introduction message - A greeting message for the users
- Company details - Provide the name, website, and description.
  
  Then click on "Save and Proceed".
The next step is to provide a data source for the AI chatbot to be trained on:
- You can provide a web link - The system will verify the URL, and then a :white_check_mark: will show against it.
- You can also choose to use articles published through the Knowledge Management System as a data source.
- You can select both as well
  
  Once done, click "Create Chatbot" in the bottom right and your chatbot will start training.
- Once the training is completed, the status of your chatbot will change to "Ready"

After you have created an AI chatbot by following the steps mentioned here, you can configure it in your flow by using the "AI Chatbot" component from the functions menu.
Drag and drop the AI chatbot component in the canvas. Drag a line from the "Incoming chat" box of the ‘Start’ node and connect it to the ‘AI Chatbot’ component.
Now, configure the AI Chatbot component by providing the below details:
- Select Chatbot - From the dropdown, select the chatbot you built.
- Reply Timeout - Configure the time for which the bot should wait for a response from the user.
The AI chatbot component has 4 output states - Resolved, Agent Assist, Timeout, and Error.
- If the customer responds with a reply that determines their query is resolved, like a “Thank you” message, the state “Resolved” is triggered which can then be configured as per your use case - for example, you can send a closure message using “Send Chat” component.
- Alternatively, if the customer responds with a message that indicates they would like to speak to an agent, the state “Agent assist” will be triggered which can then be configured to route the conversation to an agent. In the above example, the “Agent assist” state is connected to “Queue and Route” component which will connect the user to the agent.
- If the customer does not respond within the time set up for the node to expire, the “Timeout” state will be triggered.
- Finally, we have the state “Error” which is triggered in case there are any technical glitches in the flow execution.
Publish the flow and associate it with the chat app that you create by following the steps here.