Automate Email Follow-Ups with AI: A Copy-Paste Workflow

Why Follow-Ups Slip

Receivables chasing is the classic task that everyone agrees matters and nobody quite keeps on top of. It is repetitive, it is slightly awkward, and it competes with everything else on a finance team's desk at month-end. So invoices that a polite nudge would have collected sit a week or two longer, and the business carries cash it should already have.

A Dubai marketing agency we worked with, around 60 staff, had let this slip to the point that average days-to-pay had drifted past seventy days. The finance manager was not short of intent, just time: chasing always lost to month-end. They put the prompt below to work, with an account handler reviewing and sending the drafts each morning, and within about two months days-to-pay had come back into the mid-fifties. Nothing about the underlying client relationships changed. The chasing simply stopped falling through the cracks.

The work itself is not hard, which is exactly why it suits AI. The judgement, who to chase, how firmly, in what order, follows rules you can write down. Once the rules are explicit, drafting the actual emails is mechanical. The workflow below puts the rules in the prompt and lets the AI do the drafting and triage, while you keep the relationship-sensitive decision of what actually goes out.

The Copy-Paste Prompt

Paste everything in the box into Claude or ChatGPT, then paste your exported data directly underneath it. Adjust the thresholds and tone to suit your business.

You are helping me chase overdue invoices. Below is a list of open
items, each with the customer name, invoice number, amount, due date,
today's date, and a note of any previous contact.

Apply these rules:
1. Only draft a follow-up for items that are more than 3 business days
   past their due date.
2. Only consider items due within the last 2 weeks. Flag anything older
   than that separately for me to handle personally, do not draft it.
3. Skip any item where the previous contact note shows it was already
   chased in the last 3 business days.
4. Skip anything marked as in dispute, on a payment plan, or query
   raised. List these separately as "excluded" with the reason.

For each item that passes the rules, write a short, professional
follow-up email in a polite but clear tone. Reference the invoice
number, amount and due date. Make the first chase friendly and later
chases firmer based on how overdue the item is and how many times it
has been chased before.

Order the drafts from most important to least important, judging
importance by amount outstanding and how overdue the item is.

Do not send anything. Output the drafts as a numbered list for me to
review, copy and send myself.

Here is the data:

How the Prompt Works

Every line in that prompt earns its place. The three-business-day threshold stops the workflow nagging accounts that are barely late. The two-week window keeps the AI focused on the recent items where a nudge still works, and pushes genuinely aged debt to you, because that needs a human, not a templated email. The skip rule on recent contact prevents the double-chase that annoys good customers, and the exclusion rule keeps disputed and payment-plan accounts out of the firing line entirely.

Ordering by importance matters because attention is finite. If you only get through half the list today, the prompt has made sure the half you handle is the half that matters most to cash. And the final instruction, output a list for you to review and send, is the line that keeps the whole thing safe.

Why You Never Auto-Send

It is tempting to wire this straight into your email so the chases go out by themselves. Do not. The cost of an AI-sent chase landing in the inbox of a client who paid yesterday, or whose account is in genuine dispute, is a damaged relationship that took years to build and seconds to dent. No amount of saved time justifies that risk.

The human review step costs a few minutes and removes nearly all of the downside. You still capture the real saving, which is in the drafting and the triage, not the clicking of send. This is the same principle that runs through our AI in finance guide: AI drafts, a person decides, nothing leaves the building unreviewed.

Scaling It Up

Run it from a chat window and a spreadsheet to begin with. Once it earns its keep, move the prompt into a Claude Project loaded with your house tone of voice and chase templates, so every draft sounds like your business rather than a generic collections script. Our Claude Projects for finance workflows guide shows how. From there, an automation tool can pull the export and produce the drafts on a schedule, leaving you with just the review-and-send step each morning. The workflow also improves over time if you feed your best chases back into it, exactly the loop in our self-improving finance function guide.

The Monday Email Chase-Up

Invoice chasing is one use of this pattern. The same idea works for the asks you are personally still waiting on: the information you requested, the decision you need signed off, the review that has gone quiet. These slip for the same reason overdue invoices do, and a Monday-morning sweep catches them before they cost you a week.

This is where agent tools earn their place. A Cowork Agent inside Microsoft 365 Copilot, or in Claude, can run a prompt on a schedule, point itself at your sent mail, and email you a batch of drafts to review with your coffee. For Dynamics 365 users there is extra context worth knowing: Microsoft has put an ERP plugin into Copilot Cowork in preview, which lets it work inside your finance system on your own login and security roles. We cover that separately in our Copilot Cowork for Dynamics 365 Finance guide. You do not need an ERP to get value here, though. Point an agent at your inbox and you are most of the way there.

Paste the prompt below into a scheduled Cowork Agent, swap in your own email address, and set it to run on Monday morning. The agent finds the asks you are still waiting on, drafts a nudge for each in your own writing style, and sends the batch to you to approve.

Review my sent emails from the last two weeks and find any where I asked
someone for something: information, a decision, a review, a meeting, or an
action. Keep only the ones where it has now been at least 3 business days
and the person still has not replied. Ignore automated messages, bulk
sends, FYI-only notes, anything where no reply was really expected, and
anything I have already followed up on.

For each unanswered ask:
1. Note the recipient and a one-line summary of what I asked for.
2. Study my recent sent items to learn my writing style: tone, greeting,
   sign-off, sentence length and typical structure. Then draft a short,
   friendly follow-up nudge in that same voice. Keep it natural and
   tailored, not templated.
3. If you cannot confidently draft a good follow-up, flag it for manual
   review with a one-line reason.

Compile everything into a single email to me (your@email.com), marked
high importance and titled "Monday Follow-Up Nudges [date]" so it sits at
the top of my inbox. Order the list with the most important or
time-sensitive asks first. For each item show: recipient, what I asked,
the original send date, and the suggested draft, formatted so I can copy,
paste or forward with minimal edits. Group ready-to-send drafts separately
from needs-manual-review items. If nothing qualifies, send a brief note
confirming the check ran and nothing needed following up.

The line that makes it land is point two: telling the agent to learn your voice from your own sent mail, so the drafts read like you wrote them rather than a robot. Set it once and the Monday chase-up becomes a five-minute review instead of a job you keep meaning to do. The same rule still applies as with invoices: the agent prepares, you approve, nothing leaves unreviewed.

If you would rather have this built into your systems properly, our AI consulting team sets up receivables and follow-up automations with the right controls. An AI readiness assessment finds the other repetitive workflows worth automating, and our AI for finance training teaches your team to write prompts like the one above.

Frequently Asked Questions