I’ve found a few Reddits posts people saying that tracking prompt to determine user interaction with the agent is helpful. Does anyone here think the same? If yes, how does tracking human interaction with an agent help in improving agent development?
One purpose that I can think of is: It will help a PM judge what is the next workflow or problem to solve.
Any more suggestions and ideas are welcome. Anybody tried tracking user interaction so far?