I lead digital transformation for a regional creative agency. Before building anything, I mapped how work actually moved through it. The pattern I found will be familiar to anyone who has run an agency, and the fixes are more about operating model than software.
Where the day goes
- Intake. Requests arrive by chat, email and phone. Somebody has to notice them, understand them, and turn them into a brief. That somebody is usually a senior person.
- Scheduling against capacity nobody can see. Work is committed to clients before anyone checks whether the designer is free. Overload shows up as a missed deadline.
- Approvals in the wrong place. Internal and client approvals happen in message threads. Finding the current version, and who approved what, takes real time.
- Status reporting by hand. Account managers assemble updates from memory and messages, because the information is not in one place.
The operating model that fixes it
One record per request. However it arrives, it becomes a single item with an owner, a brief and a status. AI can draft the brief from the request; a person completes it.
Resource-first scheduling. Check capacity before committing. If it does not fit, the system says so, and a person decides whether to proceed anyway. The decision is logged, which changes behaviour on its own.
Exceptions, not dashboards. Nobody reads a dashboard of everything. People act on a short list of the things that are about to go wrong: a deadline that cannot be met, an approval waiting too long, a designer at 140 percent.
Non-blocking rules. Hard stops get worked around. Warnings that let people proceed, with the override recorded, get respected.
What it took
About six weeks from mapping to a working pilot with the Client Services and Creative teams. The software mattered, but it was built to the mapped operating model, not the other way around. That order is the whole point.
