I realize the public beta for agents is ending, but I finally had a chance to test the Reinforcement agent from both the admin preview and as a test user. I wanted to share some feedback and see if others came across similar behaviour.
In both cases, the practice session didn't seem to end naturally. At the start of the conversation, it was clear there would 3 activities/questions. I completed all three. I answered the third question partially correct and received some feedback, but then the agent didn’t move the conversation forward.
A few hours later, I returned to the conversation and asked if I could practice with two more questions. The agent responded that it had one more follow-up question before we were to finish the session. After I answered that question correctly, it finally confirmed that the session was complete.
From a learner perspective, this could be confusing because it wasn't clear earlier whether the activity had finished and not everyone may think to continue interacting with the agent.