Sunday evening, end of the month. 120 rows waiting in the hours spreadsheet. Eight clients. Three billing tiers. Two number ranges. A process that lives only in my head. Here we go again — my monthly highlight. I advise companies on digitalization and automation. By day. In the evening I sort time entries by hand — row by row: which project? which tier? billable or not? For years, my fix wasn't automation. It was perfection. One more column, one more naming convention, one more sticky note for the next invoice number. 

The spreadsheet grew ever more elaborate; the end-of-month evening never got shorter. I hadn't solved the pain — I'd furnished it. By now it had curtains. The turning point was my own line in a client meeting: "Automate the process that hurts most first — not the one that sounds most impressive."

 On the drive home it hit me: I'd been selling that sentence for years and never followed it myself. I had visions of AI-enabled consulting — and a stack of invoices. So: theory back in the drawer, invoices on the desk. The next thing to automate wouldn't be the exciting one. It'd be the annoying one: the monthly billing. T

hat unspectacular decision set off everything I tell in this series: the invoice automation became a company portal, the portal became a firm with a new colleague — one who never tires but forgets everything every night. 

Told honestly, with real numbers and every failed attempt. I learned almost only from those.