Imagine the following scenario: a user starts a conversation with a virtual agent to report a problem. The agent, autonomously, opens a ticket in ITSM. Days later, the same user returns and asks: "What is the status of my ticket?".
The business rule is very clear: the virtual agent needs to provide the status; the client doesn't want email notifications or any automation showing the status.
It seems simple, doesn't it? But the question remains: can a virtual agent that follows only one interaction flow retrieve this information while remembering that it will need to re-enter the flow.
The Nature of the Problem
The virtual agent, by itself, is an interaction layer. It is designed to converse, but not necessarily to "remember." When the conversation ends, the context is often lost. ITSM, on the other hand, is a transactional, rigid system, full of states, statuses, and business rules.
When a user inquires about the status of an issue, the virtual agent needs to somehow bridge the gap between the user's natural language and the technical language of the database.
How would they solve this? Via API, using web requests within the virtual agent?
