Latest / Fintech Conversations with Fexingo: Banking, Payments, and Financial Software Companies / How Fintech Is Using AI to Automate Small Business Collections
Transcript
- Lucas: You know how most small businesses handle overdue invoices? They send the same aggressive email every thirty days, then a final notice, then maybe a phone call from someone who sounds like they've had a rough morning. Luna: Right — and the customer just ignores it because it feels like spam. Lucas: Exactly. But a growing number of fintechs are flipping that script by using AI to make collections personalized, predictive, and actually less painful for everyone involved. Today I want to look at one company specifically: collectAI, based in Hamburg. Luna: I've heard the name — they focus on what's called 'digital receivables management,' right? Lucas: Yes. They started as a spin-off from the Otto Group, the German e-commerce giant, about eight years ago. Their pitch is simple: send the right message, through the right channel, at the right time. Not just an email blast, but a WhatsApp reminder if the customer opens WhatsApp more, or an SMS if they're in a region where SMS has higher open rates. Luna: So the AI learns the customer's preferred communication style? Lucas: That's part of it. The system tracks past payment behavior — not just whether they paid late, but when they typically pay. If someone always pays on the 15th, sending a reminder on the 16th might be too late. Send it on the 10th instead. It also adjusts the tone. Some customers respond to a friendly nudge; others need a firmer deadline. The algorithm figures that out. Luna: And this actually works? I mean, we've all seen 'smart' email sequences that still feel robotic. Lucas: Their clients report a twenty-five percent improvement in recovery rates and a thirty percent reduction in collection costs. One case study — a European telecom company using collectAI — saw their days sales outstanding drop by twelve days within six months. Luna: That's dramatic. But I wonder — when you hand collections over to an algorithm, do you risk being too lenient on customers who can pay but just don't want to? Lucas: That's the tension. The companies design the rules. The AI doesn't decide to forgive a debt; it decides the sequence of escalation. If a customer has a history of paying after three reminders, the system will keep nudging. But if someone has never missed a payment in five years and suddenly a single invoice is overdue, the AI might send a gentle check-in instead of a penalty threat. Luna: So it's about differentiating between a cash flow hiccup and willful default. Lucas: Exactly. And collectAI also integrates with accounting software like QuickBooks and Xero. That's the embedded finance angle — the collection process starts inside the same tool the business already uses to send invoices. Luna: That's smart. It removes the friction of logging into a separate portal. Lucas: Totally. And it's not just collectAI. There's a company called TrueAccord in the U.S. that uses similar behavioral AI for debt collection, and they claim a ten percent higher recovery rate than traditional agencies. The whole category is moving from a 'hammer' approach to a 'scalpel' approach. Luna: But let's talk about the regulatory side. Europe has GDPR, and there are specific rules around automated decision-making. How do these fintechs stay compliant? Lucas: This is a great point. Under GDPR, individuals have the right not to be subject to decisions based solely on automated processing if those decisions produce legal effects. Collections can absolutely have legal effects — like late fees or credit reporting. So companies have to ensure there's human oversight in the loop. CollectAI's system flags cases where the AI recommends a significant escalation, like referring to a collection agency, and that requires a human to approve it. Luna: So the AI is more of a recommender than a decision-maker. Lucas: Right. And in the U.S., the Consumer Financial Protection Bureau has been tightening rules around debt collection practices, including how digital communications are used. So fintechs are designing their systems to be transparent — the debtor knows they're interacting with an automated system, and they can request to speak to a human. Luna: That seems fair. But I imagine some small business owners might be skeptical. They think, 'I know my customers better than an algorithm.' Lucas: And they might be right in some cases. But the data shows that algorithms can spot patterns humans miss. For example, a customer who always pays on Friday evenings might be more likely to respond to a Friday afternoon reminder than a Monday morning one. A human collector might not track that granularly. Luna: That's actually a nice example. So it's not replacing the relationship — it's augmenting it. Lucas: Exactly. And for very small businesses, the automation is a lifeline. They don't have a dedicated collections person. The owner is wearing five hats. Having an AI that sends reminders automatically, tracks responses, and updates the accounting software — that's huge. Luna: Speaking of that — this kind of deep dive into how technology is reshaping operations is exactly why we put so much effort into these episodes. And a quick honest thing — a handful of listeners chip in monthly through Buy Me a Coffee dot com slash fexingo, and that's literally what funds making this many of these. Lucas: Yeah, it's true. We don't run ads, we don't have sponsors. It's just listener-supported. So if you find these conversations useful for what you're building or running, that link again: Buy Me a Coffee dot com slash fexingo. And we really appreciate it. Luna: Okay, back to collections. One thing I find fascinating is the global dimension. CollectAI started in Europe, but now they're expanding into Asia and Latin America. Are there cultural differences in how collections work? Lucas: Massively. In some cultures, a direct demand for payment is considered rude; in others, it's expected. So collectAI's models are localized — they adjust the messaging tone and channel based on country and even region. In Japan, they might use a more formal, indirect tone. In the U.S., they can be more direct. Luna: And the channels too. In many parts of Africa and Asia, WhatsApp is the dominant messaging platform, so an SMS might be ignored. Lucas: Exactly. collectAI supports WhatsApp, Facebook Messenger, WeChat — whatever the local preferences are. The AI learns which channel gets the best response for each customer. Luna: That's a level of personalization most humans can't match at scale. Lucas: And it's not just the channel. They also optimize the timing based on time zone and historical behavior. If a customer usually opens emails at 8 PM, the system sends the reminder at 7:45. It sounds trivial, but open rates can double. Luna: I want to ask about the future. Where is this heading? Lucas: I think we're going to see collections become fully embedded in the invoicing process. Instead of sending an invoice and then starting a separate collection process later, the system will monitor payment in real time. If the invoice isn't paid by the due date, a series of automated actions triggers — starting with a reminder, then a late fee if the contract allows, then an offer for a payment plan — all without human intervention. Luna: That raises an interesting question: could the system offer a discount for early payment? Like dynamic discounting? Lucas: Absolutely. Some platforms already do that. If the AI predicts a high probability of late payment, it might offer a small discount for immediate payment. That's better for the business than chasing the invoice for sixty days. Luna: So the AI is not just a collector — it's a cash flow manager. Lucas: Right. And I think the next frontier is integrating with credit bureaus. If a customer consistently pays late, the AI could report that behavior — with consent — to a credit bureau, which then affects their credit score. That's a powerful incentive. Luna: But also a potential risk if the AI is wrong. False positives could ruin someone's credit. Lucas: That's why the human-in-the-loop is crucial. Any negative credit reporting should be reviewed by a person. The fintechs building these systems are very aware of the ethical implications. Luna: It's a balance between efficiency and fairness. Lucas: Exactly. And I think the companies that get that balance right will win. The ones that just automate without empathy will face backlash. Luna: So for a small business owner listening — what's the first step to adopting this kind of AI collections? Lucas: Start with your accounting software. If you use QuickBooks or Xero, check their app marketplaces. There are plugins from collectAI, Chaser, and others that integrate directly. You can try it on a small set of overdue invoices and see how it performs. Luna: And if the results are like those case studies, it could transform your cash flow. Lucas: Exactly. Collections don't have to be a dreaded task. With AI, it becomes part of an automated, data-driven system that actually improves customer relationships. Luna: Alright, that's a good note to end on. Thanks for listening, and we'll be back with another episode soon.