The invoice is already approved. The work is done. The money is simply not in the bank. 83% of firms have yet to fully automate their accounts-receivable operations, which means invoices routinely go out late and collections remain reactive email-and-phone work. Collections is the most underrated output multiplier in a medium business, because it protects cash without adding headcount.

The reactive trap

The typical pattern is familiar: the invoice is created when someone remembers, chased when cash gets tight, and escalated when the relationship gets awkward. Each chase costs attention, and attention is the scarcest resource in a small finance function. Meanwhile days-sales-outstanding climbs, and working capital that should fund the next project sits in somebody else’s account.

The automated cadence

Automation detects past-due invoices, segments customers by payment behavior and risk, enforces reminder cadences, and routes exceptions. The rules are yours: first nudge at seven days, firm follow-up at fifteen, escalation at thirty, human judgment beyond that. Gen AI drafts each reminder in the tone the situation deserves, using the client’s history and the invoice status. The first nudge is a courtesy. The firm follow-up is direct. The escalation preserves the relationship while making the consequence clear.

The results are measurable. Intuit reports automated, personalized invoice reminders get businesses paid an average of 5 days faster, and overdue invoices are 10% more likely to be paid in full. Aston AI reports customers achieving +50% free cash flow by automating 80% of reminders, a 5-day reduction in DSO in the first few months, and a 50% reduction in overdue receivables. That is working capital released without a single additional collections hire.

The boundary discipline

The calculation, aging, and escalation logic belong to automation. The message drafting belongs to Gen AI. The decision to escalate, litigate, or write off belongs to a human. That boundary is not a limitation. It is what makes the system trustworthy enough to run unattended on the eighty percent that is routine.

A 30-day experiment

1. Export every open invoice with its age.
2. Segment by behavior: usually pays on time, historically slow, currently past due.
3. Configure one reminder cadence per segment with clear escalation rules.
4. Draft the first three reminders with AI, then have a human approve the tone before anything sends.
5. Measure DSO and overdue balance at day 30. Those two numbers are the only score that matters.

The lesson across every workflow in this series is the same. Automation owns the trigger, the rule, the calculation, the routing, and the audit trail. Gen AI owns the drafting, the personalization, the summarization, and the retrieval. Humans own approval, escalation, and anything that touches money, access, or compliance.

You are not short on revenue. You are short on collection. The cash is already yours on paper. The system is what brings it home.