Analytics and monitoring

One of the key features of cognition is the ability to analyze the quality of your GenAI agents. It must be reliable and trustworthy in order to be helpful enough to be of actual use. When annotators use the agent and provide feedback via the feedback system, you can view the results in your dashboard. This is e.g. helpful to understand if and how the agent got better over time and how feedback changed.

Also, you can use it to understand how users typically interact with the system and how long it takes for a message to be generated, so you can understand the experience of your endusers better.

There are different metrics that help the monitoring in different ways:

• Feedback overview: visual representation of positive, negative or neutral feedback in the last X days
• Feedback details: filter feedback from and to date and gives the option to jump directly to the conversation where the feedback was given. Apart from that the user can see the feedback message that was given and the header of the conversation
• Response time: visual representation of the response times, grouped messages and counted number of messages on every 0.5s
• Number of messages: visual representation of the number of messages in bar or donut, display of conversation distribution by message count (all time)

‍