Skip to main content
When your users are interacting with your chatbot, phone, or email support, they have the option to vote (👍, 👎) on the quality of the AI response to their query. We store these votes in a database and use them to calculate the quality of the AI response. Additionally, in some cases, we give our dashboard users a way to put a human in the loop and provide feedback on the AI response. This practice is called RLHF and is a new way to improve the quality of AI responses.

​
Is it available?

At the moment, the RLHF feature is only available to a few users. We are working on making it available to everyone.

​
What can I do for now?

You can still use the votes endpoint to vote on the quality of the AI response, and you can use the history endpoints to see the history of votes.

​
From where I can get the message ID?

You will need the message id in order to vote on. You have two options:
  1. In the chat widget, after every AI response you will get a socket event with the name vote. The event will contain the message ID of the last AI response.
  2. Or you can use the history endpoint to get the history of votes and find the message ID of the last AI response.