Latest / Fintech Conversations with Fexingo: Banking, Payments, and Financial Software Companies / How Fintechs Use AI for Automated Invoice Factoring
Transcript
- Lucas: So, invoice factoring has been around for centuries. Basically, a business sells its unpaid invoices to a third party at a discount, gets cash immediately, and the factor collects from the customer. But the process has always been slow, manual, and expensive. Luna: Right, and small businesses especially get stuck waiting 30, 60, even 90 days for payment. Factoring was one of the only options, but the fees could be brutal. Lucas: Exactly. And that's where fintech has stepped in. A company like Fundbox is using AI to automate the entire factoring process. Instead of a human underwriter poring over paper invoices, their model ingests data from the business's accounting software, analyzes hundreds of signals, and can approve funding in minutes. Luna: What kind of signals? I imagine it's more than just the invoice amount. Lucas: Much more. The model looks at the client's payment history, the creditworthiness of the invoiced customer, the industry, the time of year, even the wording of the invoice. Things like how many days past due the invoice is, whether there are disputes, or if the customer has a pattern of late payments. Luna: So it's underwriting the debtor, not just the seller. Lucas: Exactly. Traditional factors often rely heavily on the seller's credit, but Fundbox's AI evaluates both sides. That's a big shift. It means a small business with thin credit history can still get funded if its customers are solid. Luna: And the turnaround time? You said minutes. Lucas: Yeah. They claim under three minutes from connection to decision. The business links their QuickBooks or Xero account, selects the invoices they want to advance, and the AI scores them instantly. The funds hit the bank account as soon as the next business day. Luna: That's a far cry from the old days of faxing invoices and waiting a week for a decision. Lucas: Right. And the fee structure is simpler too. Fundbox charges a flat fee per invoice, typically one to three percent for a 30-day term. That's competitive with traditional factoring, which often runs three to five percent for the same period, plus hidden charges. Luna: How does the AI keep default rates low? If they're lending against invoices that might not get paid, the model has to be pretty accurate. Lucas: They train the model on millions of invoices and payment outcomes. The algorithm learns patterns that correlate with default: things like repeated late payments from a particular debtor, or invoices in certain industries with high dispute rates. They also use natural language processing to scan invoice line items for red flags—like vague descriptions or mismatched amounts. Luna: So it's basically a real-time risk score for each invoice. Lucas: Exactly. And the model gets smarter over time. If a debtor pays consistently, future invoices from that customer get a lower risk score and a better fee. Conversely, if a debtor starts paying late, the model flags it immediately. Luna: That's a virtuous cycle for good payers. Lucas: Absolutely. And for the fintech, it means they can offer factoring to businesses that traditional factors would reject. There's a whole segment of the market that's underserved because they don't have a decade of credit history or huge collateral. Fundbox has funded over three billion dollars since its founding, and a lot of that went to companies that couldn't get a bank loan. Luna: What about the risk of fraud? Someone could create fake invoices or inflate amounts. Lucas: That's a key area where AI helps. The model looks for anomalies: invoices that don't match the business's typical pattern, or that reference customers the business hasn't worked with before. It also cross-references data across the network—if multiple businesses submit invoices to the same fake customer, the pattern gets flagged. Luna: So it's a combination of underwriting and fraud detection in one engine. Lucas: Right. And it's not just Fundbox. Other players like BlueVine and Lendio are doing similar things. The whole space is moving toward automated, ai driven factoring. Even traditional banks are starting to build their own systems, though they're slower to adopt. Luna: I want to zoom in on one aspect: the data connection. The business has to give read-only access to its accounting software. That's a big trust hurdle, isn't it? Lucas: It is. But fintechs have been working on that for years. Fundbox uses OAuth-based connections, so they never store login credentials. They only pull the data they need—invoice details, payment history, customer info—and they have strict data privacy policies. Plus, the value proposition is clear: you get funding in minutes without paperwork. Luna: And businesses are used to that now, thanks to apps like Stripe and Square. Lucas: Exactly. Once you've seen how seamless payments can be, waiting a week for a factoring decision feels archaic. The AI is just the engine that makes speed possible. Luna: What about the cost? You mentioned one to three percent for 30 days. How does that compare to a traditional bank line of credit? Lucas: A bank line might have an annual percentage rate of 8 to 12 percent, but you have to qualify, and the approval process takes weeks. For a small business that needs cash today to meet payroll, factoring is more expensive on an annualized basis, but it's available. And with AI factoring, the rate is often lower than traditional factors because the model reduces risk. Luna: So it's a trade-off: speed and accessibility for a higher cost than prime borrowing, but lower than old-school factoring. Lucas: That's the sweet spot. And as the models improve, those fees should come down further. We're already seeing more competition, which helps. Luna: Are there any downsides? I mean, what happens if the AI misprices risk and approves a bad invoice? Lucas: That's the biggest risk for the fintech. If the model is wrong, they lose the advanced amount. That's why they continuously train it on new data. But no model is perfect. Fundbox has had to write off some losses, but they claim their default rates are lower than the industry average. Luna: So the model is constantly learning from its mistakes. Lucas: Exactly. And that's the advantage of AI over static rules. A human underwriter might take months to adjust their criteria; an AI can retrain overnight. Luna: Let's talk about the broader impact. If AI factoring becomes mainstream, does it change how small businesses manage cash flow? Lucas: I think it does. Today, many small businesses rely on credit cards or personal loans to bridge gaps. If they can factor invoices quickly and cheaply, they can avoid high-interest debt. It also lets them offer net-30 or net-60 terms to their customers without hurting their own cash flow. Luna: So it's almost like a liquidity on demand. Lucas: Yeah, exactly. And that can be a game-changer for growth. Instead of waiting for payments to come in, a business can reinvest in inventory or hiring right away. Luna: Is there any regulatory risk? Factoring is still lending in many jurisdictions, and AI underwriting is under scrutiny. Lucas: That's a real concern. Regulators are looking at whether AI models introduce bias or lack transparency. Fundbox and others need to be able to explain why a particular invoice was rejected. If the model is a black box, that could be a problem. Luna: So explainability is key. Lucas: Exactly. The industry is moving toward more interpretable models, or at least providing human-readable reasons for decisions. It's still early, but it's something every fintech in this space has to address. Luna: Before we wrap, let's talk about the human side. For business owners, this technology is invisible—they just see fast funding. But behind the scenes, it's a complex AI system. Lucas: Right. And that's the beauty of good fintech. The complexity is hidden. The business owner connects their accounting software, selects invoices, gets money. They don't need to understand the machine learning model. Luna: It reminds me of how Stripe made payments invisible. The best fintech is the kind you don't think about. Lucas: Exactly. And that's what these conversations are about. If you're building or running a business, and these kinds of tools are useful, we hope the show helps you think about what's possible. We deliberately keep these episodes ad-free because we believe the content should stand on its own. If you find value in that, and you want to support the choice, you can find us at buy me a coffee dot com slash fexingo. No pressure, just a way to keep it independent. Luna: Yeah, it's a small way to keep the show exactly what it is. Lucas: Alright, back to the topic. So where do we see AI factoring going in the next few years? Luna: I think we'll see more integration with payment platforms. Imagine a business uses Stripe for its invoices, and at the same time, it can factor them instantly through the same dashboard. Lucas: That's already happening to some extent. Stripe has its own capital products, but I think deeper integration is inevitable. Also, we'll see the models incorporate more external data—like macroeconomic indicators or industry trends—to improve risk assessment. Luna: So factoring becomes a real-time financial management tool, not just a last-resort funding option. Lucas: Exactly. And that shift is huge. For centuries, factoring was seen as a sign of desperation. AI is turning it into a proactive cash flow strategy. Luna: It's a great example of how AI doesn't just automate an old process—it transforms the economics and the psychology of a financial product. Lucas: Well said. That's really the core of what we cover on this show: not just the technology, but the way it changes behavior.