Politeness collects more overdue invoices than threats do
Every founder who has run a business long enough has the same drawer of unpaid invoices.
The work was delivered. The customer was happy. Then the payment date came and went, and now there is an awkward silence where the money should be. You send a reminder. Nothing. You send another, a little firmer this time. Still nothing. Eventually you face a bad choice: keep chasing it yourself and poison a relationship you spent months building, or hand it to a collection agency and accept that both the money and the customer are probably gone.
That second option is the one most businesses eventually take, and it is worse than it looks. A traditional agency typically keeps 30% to 50% of whatever it recovers, works slowly, and does the one thing guaranteed to end the relationship: it treats your customer like a deadbeat. The industry was built on the assumption that pressure is the product. Threaten enough people and some fraction will pay.
Respaid was built on the opposite bet.
Respaid is a Y Combinator-backed fintech that collects overdue invoices without torching the relationship behind them. It pairs communications sent through a real law firm with AI that works out the right moment, the right message, and the right channel to reach each debtor. The company's own line for it is plain: collect overdue invoices without losing clients. And the numbers it reports suggest the polite approach is not a handicap. It recovers more.
Why the old model recovers so little
To see why Respaid can exist, you have to look at how badly the incumbent model performs on its own terms.
Debt collection has a reputation for being ruthless and effective. It is ruthless. It is not especially effective. Agencies recover a minority of what they are handed, take months to do it, and keep a large cut of the proceeds. The reason is baked into how they operate. A collector working hundreds of accounts has no context on any single one. They do not know whether the invoice went unpaid because the customer is broke, because it landed in a spam folder, because an approver left the company, or because of a genuine dispute about the work. So they reach for the only tool that scales without context: pressure, applied uniformly, at whatever time is convenient for the collector rather than the debtor.
That approach fails in two directions at once. It annoys the people who were always going to pay, the ones who simply missed an email, and it hardens the people who might have paid if someone had made it easy and given them a graceful way to do it. Along the way it burns the commercial relationship, which for a B2B seller is often worth far more than the single overdue invoice. You do not want to win back $4,000 and lose a customer who would have spent $40,000 next year.
The founder story behind Respaid comes straight out of this pain. John Banner started the company after an earlier business of his nearly collapsed because customers would not pay their invoices. When he went looking for a better way to collect, there wasn't one, and he watched how many small companies get pushed toward bankruptcy by the same cash-flow gap. Respaid is his answer to a problem he had already lived through once.
A lawyer for weight, AI for judgment
Respaid's method splits the job into two parts and gives each to whoever is better suited to it.
The first part is authority. Respaid sends its communications through a real law firm rather than a generic collections script. That matters for a reason that has nothing to do with intimidation. A message that carries genuine legal standing gets read and taken seriously, where the tenth automated "FINAL NOTICE" gets ignored. But the tone stays soft. The messages are written to leave the door open for the ordinary explanations behind a late payment: a mistake, a miscommunication, an invoice that never reached the right person. The debtor is treated as someone who probably intends to pay and simply needs a clean way to do it, not as an adversary.
The second part is timing and targeting, and that is where the AI does its work. Every debtor is different, and the difference decides whether you get paid. Respaid's models predict the moment a given person is most likely to respond, the wording most likely to land, and the channel most likely to reach them, whether that is email, a text, or something else. Getting those three right is the difference between a message that is opened and acted on and one that is deleted on sight. It is the part of collections that a human agent working from a call list simply cannot do at scale, and it is the part that determines the recovery rate.
Underneath both is a design choice that shapes everything: the debtor is offered a simple, direct way to pay the moment they decide to. No login, no dispute portal, no friction between the intention to pay and the act of paying. A surprising amount of uncollected money is stuck behind an inconvenient process rather than an actual refusal to pay. Respaid's job is to remove the inconvenience and give people a reason to act now.
What the respectful approach actually returns
Respaid publishes numbers that make the case better than the pitch does.
The company reports collecting about half of the receivables it takes on within twenty days, and a debtor Net Promoter Score of 93. That second figure is the surprising one. NPS measures how people feel about an experience, and a score in the 90s from people who were just asked to pay a debt they had been avoiding is not something the traditional industry can produce. It is direct evidence that being collected from by Respaid does not feel like being collected from. Respaid also claims it recovers substantially more than conventional agencies and does it far faster, and says its platform handles a high daily volume of invoices across its customer base.
That customer base is not a handful of small merchants. Respaid says it works with hundreds of companies across a range of industries, including names like DoorDash, Checkr, and Podium. Those are businesses with real collections operations and real alternatives. They are the kind of customer that validates a claim, because they can measure recovery rates precisely and would not keep using a softer method if it collected less.
The NPS figure and the recovery rate are two halves of the same argument. The whole premise of aggressive collection is that you trade goodwill for money: you accept that the customer will hate you, in exchange for getting paid. Respaid's data says that trade was never necessary. You can keep the goodwill and collect more of the money, because the goodwill is part of what makes people pay.
Pricing that removes the risk
The way Respaid charges is as pointed as the way it collects.
You pay only if it works. There is no retainer and no fee for invoices that go uncollected. A business can hand over its overdue accounts with nothing more than an uploaded spreadsheet, or a direct integration if it prefers, and owe nothing unless the money actually comes back. For a company sitting on a pile of receivables it has already written off in its head, that is close to a free option. The downside is a spreadsheet upload. The upside is cash it had stopped expecting.
Success-based pricing also keeps Respaid honest in a way the agency model does not. A collector on a flat fee gets paid whether or not you recover anything. Respaid earns only when its customer earns, which means every incentive points at actually collecting rather than at looking busy. It is the same alignment that made contingency work standard in other corners of the legal world, applied to a category that mostly still bills regardless of results.
There is a second product hiding in the same data. Because Respaid has learned what a recoverable debt looks like, it can score how likely a given account is to be collected before anyone lifts a finger. That prediction is useful on its own. It tells a finance team which invoices are worth pursuing and which are genuinely gone, so effort goes where it will actually return something instead of being sprayed evenly across a list. The collection engine and the prediction tool feed each other: every account Respaid works teaches the model what recoverability looks like next time.
The larger shift in getting paid
Step back and Respaid is an early example of a pattern that is going to spread across every unloved corner of business operations.
Collections, like a lot of back-office work, has been stuck for decades on a model that treats a nuanced problem with a blunt instrument because the blunt instrument was all that scaled. One collector cannot learn the context of a thousand accounts, so the industry stopped trying and leaned on volume and pressure instead. AI removes that constraint. It can carry context on every account at once, reason about each debtor individually, and pick an approach per person rather than per call list. When the constraint that forced the crude approach disappears, the crude approach stops being the efficient one. It just becomes the rude one.
That is why an incumbent agency cannot simply copy Respaid. Their economics run on keeping a large share of what they recover and on the labor of human collectors making calls. A method that recovers more while charging only for success and treating debtors gently would undercut the fee structure their whole business depends on. Respaid has no such conflict, because it was built from scratch around the new economics rather than defending the old ones.
The framing Banner chose captures the whole thesis. Collections was never really about being harsh. It was about getting paid, and harshness was a workaround for not knowing enough about the person on the other end to do anything smarter. Respaid knows more, so it can be kinder, and it collects more precisely because it is kinder. For a founder staring at that drawer of unpaid invoices, the offer is simple. The gentle method brings back more of the money than the aggressive one ever did, and it keeps the customer besides.