I have a bot-processing scenario where there is a need to handle multiple cases with the same type and transaction type for the same customer. The application used in the automation has contention issues if multiple bots are working on cases for the same customer simultaneously.
Ideally, I would be able to allow one case for each unique customer to be worked by a bot or queued to be worked one at a time to avoid these contention issues and mitigate requeuing fallout. Volumes are too large to dedicate a single bot to work the entire case load and the transactions are spread out across many different customers.
Also of note, I am also looking to achieve this without the use of the parent/child relationship model due to the way cases are created by upstream processes.
The first solution that comes to my mind to control that only a single assignment is created for each client, is to create a pre-assignment work basket. All your assignments go to this work basket before going to the robotic assignment. In this workbasket you either by using an SLA or a dedicated job scheduler evaluate whether there is an assignment for that client in the robotic assignment work queue defined. If there is none, you move the process forward, if there is, you keep the assignment where it is until the robotic assignment for that client is resolved.
This way you will ensure that there is only one robotic assignment for that client.