Overdue invoices from day 7: AI drafts, the CEO approves

In most companies, an overdue invoice gets chased only when someone remembers it. Usually that is the day cash gets tight. By then the invoice is three or four weeks late, the client has forgotten about it, and the reminder goes out in a sharper tone than it should.

In my companies, reminders are drafted from day 7. I no longer write them: I read them and approve them. Here is why that timing matters, how the setup works, and how to put it in place in your business without damaging relationships you have spent years building.

Why reminders go out late

Chasing a payment is not hard. It is a short task, but it always lands at the wrong moment. Someone has to know which invoices are past due, check that the payment has not arrived in the meantime, find the right contact on the client side and write a message that does not offend. Four small steps, each one just tedious enough to be pushed to tomorrow.

The outcome is familiar. Reminders go out in waves, whenever the CEO or the CFO looks at the cash position. Between waves, delays settle in. A client who has not heard from you a month after the due date concludes, often in good faith, that the date was not that important.

So the problem is not the writing. It is the trigger. A reminder that waits for someone to think of it rarely goes out on time.

What AI does, and what it does not

My finance tracker is connected to my three banks. Every week, an agent pulls in the new statements and updates the P&L, the forecast and the 60-day cash view. Reminders are prepared on top of that.

The setup does three things.

1. It spots. It matches issued invoices against payments received. Any invoice 7 days past due with no matching payment goes on the list.
2. It drafts. For each invoice it writes a reminder: invoice number, date, amount, due date and how to pay. The tone follows the stage of the reminder.
3. It brings them to me. I receive the day's reminders, ready to go. I read them, adjust one if needed, and approve.

What it does not do is decide alone. No reminder leaves without my approval. The reason is simple: the tool sees the numbers, not the relationship. It does not know that one client has just signed an extension, that another is going through a hard quarter, or that a dispute is open on a delivery. Those I take off the list, or I pick up the phone.

Why day 7

Seven days is not a universal rule. It is mine, and it rests on a simple idea: the earlier a reminder goes out, the easier it is to write and to receive.

At day 7, a late payment is almost always an oversight or a paperwork issue: an invoice sent to the wrong department, a missing purchase order, a transfer prepared but not yet released. The message can stay light. "Unless we have missed it, we have not yet received payment for the invoice below" is enough, and clients often thank you for flagging it.

At day 30, the tone shifts whether you want it to or not. You have waited, you are irritated, and it shows. Meanwhile the client has had time to pay other suppliers first.

This matters even more in markets where business runs on long-standing relationships and where payment terms tend to stretch. An early, courteous reminder protects the relationship. A late, firm one tests it. That is the point most business owners underestimate.

Reliable invoices before reliable reminders

An automated reminder is only useful if the list of overdue invoices is right. Chasing a client who has already paid does more harm than chasing late. Before drafting anything, you need two things.

Invoices issued without gaps. At Oliverlist, we have automated the whole chain since 2020, from importing contacts to issuing the invoice. Every meeting booked generates its invoice automatically. No forgotten invoice means no forgotten reminder.

Payments matched every week. The tracker has to know a transfer has arrived even when the bank reference does not quote the invoice number. This is the most delicate part, and it needs written rules: this client always pays from that account, this reference means that kind of payment. When matching fails, the question should come to you, and your answer should become one more rule.

If those two conditions are not in place in your company, start there. Reminders will follow within days.

How to set it up in your company

1. Start from your real data. Issued invoices, bank statements, your invoicing system. One table where every invoice carries its due date and its status, across all entities and currencies.
2. Set your schedule. A first reminder at day 7, a firmer follow-up later, then a call. The timings are yours. What matters is that they are written down.
3. Write your templates once. One tone per stage, with everything the client needs to pay without calling you back: invoice number, amount, due date, bank details.
4. Keep the approval step. At first, every reminder goes through you or your CFO. You remove the sensitive cases and adjust the tone. Your corrections become rules.
5. Watch the effect on cash. With a 60-day cash view updated weekly, you can see whether money comes in sooner. Measure before and after, on your own figures.

This often makes a good first sprint: one tool in production in 14 days, used on your real workflows. If invoices and payments are spread across several systems and banks, a two-week audit measures the work first. Either way, accounts and subscriptions stay in your name, and every engagement starts with a quantified gain target, measured on your data: if it is not reached by the agreed date, I keep working without billing until it is.

Start with 30 minutes

If your reminders go out when someone remembers, or if nobody can say without digging which invoices are more than 7 days late, this is probably a good place to start.

Book 30 minutes: you walk me through how invoices and collections are tracked today, and I tell you what can run without you. If you would rather start on your own, take the online test: 10 questions, your level among four, and the 3 tasks to hand to AI first. The result appears on screen, and nothing is sent or stored.