Oil painting of a cream envelope with a copper wax seal on a stack of papers, beside a brass desk bell and an open pocket watch, with larger envelopes behind.

Quick answer

To chase a late invoice with AI, check your records first, give the assistant the real invoice number, amount and due date, and ask for three versions: a friendly nudge, a firmer follow-up and a final reminder. Keep the tone fair and factual, read every draft against the invoice, and send it yourself. AI writes the words; you stay responsible for the facts and for what goes out under your name.

Key takeaways

  • Records first. Make sure the invoice is unpaid before anyone writes a word.
  • You supply the facts. Invoice number, amount, due date and payment details come from you, never from the AI.
  • Three versions, written once. A nudge for a day or two late, a firmer note for about a week late, and a final reminder when you have heard nothing.
  • No threats. Do not let AI invent legal action, fees or deadlines you have not decided on.
  • Seriously overdue? Get proper advice rather than escalating by chatbot.

Chasing late payments is the admin job most of us put off, which is exactly why it is worth setting up properly. This guide follows on from my main article, How to Use AI for Business Admin and Operations: 4 Jobs to Start, where invoice chasing is the fourth of four admin jobs I suggest handing to AI.

📘 Main article: New to using AI for admin? Start with How to Use AI for Business Admin and Operations: 4 Jobs to Start. It covers the one rule behind everything here (AI drafts, you decide) and the other three jobs: follow-up emails, inbox triage and appointment reminders.

This guide answers one question: how do you use AI to write late invoice reminders that are polite, firm and accurate? The method works with any general assistant, such as ChatGPT, Claude, Gemini or Copilot.

Why use AI to chase a late invoice?

Because the hard part is usually the wording, not the sending. Many owners know an invoice is late but delay the email because they worry about sounding rude or desperate. A good first draft in seconds removes that friction.

As I say in our video, nobody enjoys this one. Having a few ready-made versions means you are choosing the right tone rather than writing from scratch each time you are annoyed.

💡 Tip: The aim is not to be clever. A clear, short reminder with the right facts gets paid more often than a long one.

What do you need before you start?

  • Your invoice list or accounting software, so you can see what is genuinely unpaid.
  • The invoice itself: number, amount, date issued, due date and your payment details.
  • One reminder you have sent before and were happy with, if you have one. It helps the AI match your tone.
  • An AI assistant you already use. You do not need a specialist tool to start.

Step 1: Check the invoice really is late

Before you chase, confirm the money has not already arrived. Payments can land under a different name, be split across two transfers or sit unmatched in your bank feed. Chasing a client who has paid damages trust quickly.

If you do not already have a habit for this, my guide on How to Reconcile Client Payments Each Week shows a simple 15-minute routine.

Step 2: Write a short fact sheet

Write the facts in a few lines so the AI never has to guess them. Keep it to what the reminder needs:

FieldExample placeholderWhere it comes from
Client first name[Client name]Your records
Invoice number[INV-0000]The invoice
Amount due[£0.00]The invoice
Due date[date]The invoice or your terms
What it was for[service or goods, one line]The invoice
How to pay[bank transfer / payment link]Your accounting tool
Contact so far[e.g. one reminder sent on date]Your sent folder

⚠️ Warning: Do not paste your bank login, card numbers or a full client file into a chat tool you have not checked. You can use placeholders such as [sort code] in the prompt and add the real details yourself before sending.

Step 3: Ask AI for three reminder versions

Ask for all three versions at once, so the tone steps up evenly. Here is a prompt you can copy:

Prompt to copy
You are helping me, a small business owner, chase an unpaid invoice. Using only the facts below, write three short emails in British English:
1. A friendly nudge for an invoice one or two days late. Assume it may be an oversight.
2. A firmer follow-up for about a week late. Clear and polite, restating the amount, invoice number and due date.
3. A final reminder for when I have had no reply. Calm and factual, asking them to pay or tell me when they will.
Rules: do not add any facts I have not given you. No threats, no legal language, no fees or deadlines unless I list them. Keep each email under 120 words with one clear next step.
Facts: [paste your fact sheet]
Example of my tone: [paste a past email you liked]

That line about not adding facts matters. Assistants sometimes fill gaps with things that sound right, such as a made-up due date or a late fee you never agreed.

Step 4: Check every draft before it goes

Read each draft against the invoice, line by line. Use this quick check:

  1. Name: right person, spelled correctly.
  2. Invoice number and amount: match the invoice exactly.
  3. Dates: due date correct, and no new deadline you did not choose.
  4. Payment details: correct and added by you.
  5. Tone: fair and factual, nothing you would be embarrassed to see forwarded.

Step 5: Send it yourself and log it

Press send yourself, from your normal email or accounting tool, and make a note of it. A one-line log is enough: date, invoice, which version you sent and any reply.

Save the three approved versions as templates. Next time an invoice slips, you swap in the new facts and check them. If you bill the same clients every month, my guide on How to Set Up Recurring Invoices for Retainer Clients can cut down how often you need to chase at all.

What should each reminder say?

Each version has a different job. Here is a simple ladder to follow:

VersionWhenToneMust include
Friendly nudge1–2 days lateLight, assumes an oversightInvoice number, amount, how to pay
Firmer follow-upAbout a week lateClear and direct, still politeAmount, number, original due date, a request to confirm when it will be paid
Final reminderNo reply to the first twoCalm, factual, shortEverything above, plus what you will do next, such as phoning them

The timings are a starting point, not a rule. Adjust them to your payment terms and how well you know the client.

What should AI never write in a payment reminder?

Do not let AI write anything you have not decided on yourself. In particular, remove:

  • Threats of court, debt collectors or legal action.
  • Late fees, interest or charges you have not chosen to apply.
  • New deadlines or consequences you did not set.
  • Guilt-trips, sarcasm or anything personal.
  • Any detail about the client that the AI has guessed.

💡 Tip: In the UK, businesses can claim statutory interest and fixed debt recovery costs on late business-to-business payments. If you want to use that, read GOV.UK’s guide to late commercial payments first and add it to your reminder deliberately, not because a chatbot suggested it.

What if the invoice is seriously overdue?

Stop drafting and get proper advice. When an invoice is well overdue, or the client has stopped replying, the next steps involve your contract, your terms and the law, and that is not a job for a chat assistant.

Good places to start in the UK are GOV.UK’s late payment guidance and the Office of the Small Business Commissioner, which helps small businesses with payment disputes involving larger firms. Your accountant may also be able to help.

⚠️ Warning: This guide is general information, not legal or financial advice. Check your own contract and terms before taking any further action.

Watch: the source video

What this video covers: In about four and a half minutes I walk through four admin jobs to start with AI, including chasing late invoices with a polite first nudge and a firmer follow-up, and why you should never let AI make threats. This guide expands on that fourth job.

Video: our own guide on the Prodify Digital YouTube channel, presented by Garry Knight.

Frequently asked questions

Can AI chase invoices for me automatically?

Some accounting tools can send scheduled reminders using the invoice data they already hold, which is safer than a chatbot guessing details. Use AI to improve the wording of those templates, and keep a person checking anything unusual.

Is it rude to chase an invoice the day after it is due?

Not if the tone is friendly. A short, light nudge that assumes an oversight is normal practice for many small businesses.

Should I mention late payment interest in my reminder?

Only if you have decided to claim it and checked the rules. In the UK, GOV.UK explains when businesses can claim statutory interest and debt recovery costs.

What should I do if the client disputes the invoice?

Stop sending reminders and talk to them. Find out what the issue is, check your records and agreement, and get advice if you cannot resolve it.

Can I paste the client’s details into ChatGPT?

Keep it to the minimum the email needs, and check how the tool stores what you type first. Placeholders work well: write [client name] and add the real name yourself.

The bottom line

Chasing a late invoice with AI works best when you treat the assistant as a fast writer, not a debt collector. Check your records, give it the real facts, ask for three versions once, and send the right one yourself. Next step: run the prompt on your oldest unpaid invoice today, then read the main article, How to Use AI for Business Admin and Operations: 4 Jobs to Start, to pick your next admin job.

author avatar
Garry Knight
I'm Garry Knight, the person behind Prodify Digital. I write about email list building, email marketing, SEO, AI search and the tools that connect them. My aim is to make online marketing easier to understand, so creators and small business owners can make informed decisions about building an audience and keeping people engaged. Here you'll find straightforward guides and product reviews that explain what something does, where it fits and which limitations matter. The focus is on clear explanations and useful next steps—not hype, shortcuts or promises of easy earnings.

Leave a Reply