An internal concept project using fictional examples. This case describes a workflow to evaluate before integrating tools. The linked calculator compares times entered by the person using it; it does not run an AI model or access email, CRM or client data.
The starting brief
A fictional team receives campaign, website and support inquiries in one inbox. It needs to identify the task, preserve the original message and recognize when a question is required before assignment.
The product decision
The proposed workflow would assign a category using explicit rules: project, support or billing. Ambiguous cases would enter review, with a person deciding the next step. This is a scoping choice in the concept case, not a classification feature available in the calculator.
To size the work, the automation time calculator compares monthly volume, current task time, the eligible share and human minutes remaining per eligible case. It shows capacity you could recover from those inputs, not demonstrated savings or classification accuracy.
Cases the workflow should cover
A clear campaign request should suggest preparing a brief. A message mixing an error and a new app would need review, and a context-free phrase would need clarification. A future implementation would test these cases against reviewed examples. The calculator explores another risk: remaining review taking longer than the current process.
What we would measure
Time from receipt to confirmation, agreement with a human-reviewed category, review rate and errors reaching confirmation. Preserve sample size and workflow version. Compare the whole process, including corrections.
Before integration
Connect an inbox or CRM, agree on permissions and provider data flows, build a reviewed evaluation sample, set limits and record failures. Add AI when an evaluation establishes which task needs it.
