Hi Danilo,
I'm not sure whether there is already an Idea for this behavior, but as a workaround I usually consider two approaches.
The first is to use inactivity handling and route the interaction to a flow, where you can apply the required logic when the customer returns.
For a more manual approach, the agent can transfer the interaction from the script to a queue with no agents assigned. That queue can trigger an In-Queue Flow, where the return logic is handled. When the customer replies, you can identify the last agent and route the interaction back to the main queue using preferred agent routing, increasing the chance that it returns to the original agent.
I'm not aware of a native Message Parking configuration that guarantees the interaction will always return to the same agent, so it would be interesting to see whether someone else in the community has found a cleaner solution or is using a different approach.
I hope these ideas help provide some direction
------------------------------
Raphael Poliesi
------------------------------