50 posts.
One honest account.

Most CFOs I know are waiting. Waiting for vendors to add AI to their ERP. I did that for a while. Then I stopped — and started experimenting instead.

Everyone's talking about AI in finance. Most people are confused about what it actually does. That confusion is expensive — because companies keep buying chatbots when what they need is a system that runs without them.

The most common finance setup in small business is also the most invisible. One person. Too much work. No budget for more. Here's why AI changes the math — without changing the headcount.

A company with €2M revenue can look completely healthy on Monday and be in crisis by Friday. Not because it's losing money — but because no one was watching the cash in real time.

This conversation happens every month in many companies under 200 people. The CEO needs forward-looking data. The accounting system produces backward-looking reports. It's not the accountant's fault — it's a systems problem.

Most descriptions of AI CFO agents are too abstract. Here's the actual plumbing — what systems connect to what, what the agent layer does, and why it's simpler than it sounds once you know where to start.

The first report the agent produced — what the CEO said when it landed, and the one number that was wrong and why.

What happened when the agent flagged a problem that turned out not to be one — and what that taught me about tuning alerts.

A currency move spread across thirty ordinary days can quietly cost more than most invoices are worth — and nobody notices until the books close.

Money that's already in the account can cost you exactly like money that's late — it just does it quietly, one basis point at a time.

The cash in the last post sat still because no one moved it. This time the cash moves — just not in the order the due dates suggest.

The last post was about money leaving the company, and in what order. This one is about money owed to it — and why the reminder that gets paid rarely looks like the one sent first.

Post #11 ranked the money going out and Post #12 the money coming in. This one asks what changes when both lists are read as one: two queues drawing on a single bank balance.

Post #13 ended on a payment the balance couldn't carry. This one follows the money after that: the short drawdown that just happens, and what the days it stays open actually cost.

Post #14 priced the days a gap stays open. This one asks how many weeks ahead the gap has to be seen, and what a 13-week rolling forecast needs to contain to show it.

Post #8 showed an alert that was wrong. This one asks what an alert needs before it earns the right to fire: a picture of what a normal week looks like.