How AI Agents Mark Conversations as Solved in ZazzyAgent
A busy support inbox needs a clear distinction between active issues and completed conversations.
ZazzyAgent AI Agents can mark a conversation as solved when the customer's issue has actually been resolved.
The action
Use:
##mark_as_solved##
Unlike resource-based actions, this action does not require an ID.
When should an AI mark a conversation as solved?
Use it when the customer's issue is complete.
For example:
Customer asked how to download an invoice.
AI provides the correct instructions.
Customer:
Got it, thanks.
The issue is complete.
The AI can mark the conversation solved.
Write a clear rule
Example:
When the customer's question has been fully answered
and no additional action is required,
mark the conversation as solved.
##mark_as_solved##
Don't solve conversations too early
Bad:
When you send an answer, mark the conversation as solved.
The customer may still have a problem.
Better:
Mark the conversation solved only when the requested issue has been resolved and no additional customer or staff action is required.
Customer says "thanks"
A simple:
Thanks!
often means the conversation is finished, but don't rely on that alone if an unresolved issue remains.
The agent should consider the conversation context.
Customer still needs something
Customer:
Thanks, but my order is still delayed.
This should not be marked solved.
The AI should continue handling the issue or transfer it to a human.
Combine with human handoff
Example:
If the issue requires manual investigation,
assign the conversation to Support.
##assign_human_role## : 3
Do not mark the conversation as solved.
That prevents a conversation from being simultaneously handed off and incorrectly closed.
Combine with internal notes
If a case has been resolved after a manual step, you can record useful context.
##add_note## : Customer's delivery issue was resolved and tracking information was provided.
##mark_as_solved##
See How AI Agents Add Internal Notes in ZazzyAgent.
Solved vs archived
These concepts can serve different purposes depending on your inbox configuration.
A conversation being solved means the issue is considered complete.
Follow your current inbox workflow for archiving and retention.
Example: FAQ support
Customer:
What are your opening hours?
AI:
We're open Monday to Saturday from 9 AM to 7 PM.
Customer:
Thank you.
If there is nothing else to handle:
##mark_as_solved##
Example: Order support
Customer:
Where is my order?
AI checks the order.
Customer:
Perfect, thanks.
If the order question is fully resolved:
##mark_as_solved##
If the customer says:
The tracking information says delivered but I haven't received it.
Don't solve the conversation.
The issue needs further investigation.
Example System Prompt
You are the Customer Support Agent.
Answer customer questions using the connected knowledge and available actions.
When the customer's issue has been completely resolved
and no further action is required,
mark the conversation as solved.
##mark_as_solved##
Do not mark a conversation as solved when:
- The customer still has an unresolved issue.
- A human needs to investigate something.
- The customer requested a callback.
- An external action is still pending.
- The customer explicitly asks for further assistance.
The goal
A solved conversation should mean:
This issue is finished.
Not:
The AI sent a message.
That distinction keeps your support inbox much more useful.