I remember the buzz in the room. It was a cramped office in SoMa, probably 2012. We were huddled around a whiteboard, sketching out what we thought was the future of finance. The startup was a peer-to-peer lending platform, one of my earliest angel investments. We truly believed we were the good guys, the rebels fighting for the little guy against the big, faceless banks. And for a while, it felt like we were winning. We were connecting borrowers and lenders, cutting out the middlemen, and making capital more accessible.
But then came the hangover. We hit a wall, and we hit it hard. A wall of bad data, soaring default rates, and a flood of angry emails from investors who were losing money. It turns out, just connecting people isn’t enough. Lending money is a tough, brutal business. You need to be good at figuring out who will pay you back. That’s when I had a painful but important realization: the revolution wasn’t just about peer-to-peer. It was about peer-to-peer made intelligent. It was about AI.
Peer-to-peer (P2P) lending was a genuine disruption. The idea was simple and powerful: create a marketplace where people who need to borrow money can connect directly with people who want to lend it. For a moment, it felt like we were on the verge of completely democratizing finance. But the initial excitement quickly ran into some very old, very familiar problems. How do you really know who to trust with your money? How do you spot the sophisticated fraudsters? How do you grow your loan book without your risk getting completely out of control?
For all its shiny new technology, early P2P was still playing by the old rules of credit. It was a new bottle for some very old wine. That’s where AI comes in. It’s not just a feature or an add-on; it’s the core engine that’s finally allowing P2P lending to deliver on its original, radical promise. It’s making lending faster, fairer, and more efficient for everyone involved.
The Old Way vs. The New Way: AI in Action
The difference between a P2P platform running on old rules and one powered by AI is stark. It’s the difference between driving with a paper map and driving with a real-time GPS that sees traffic, construction, and the best shortcuts. It’s about moving from a world of static, incomplete data to a dynamic, holistic understanding of risk and opportunity.
Reinventing Credit Scoring
The old way was to worship at the altar of the FICO score. The problem is, these scores are often lagging indicators. They tell you a story about a person's past, but not necessarily their present financial health or their future potential. They’re a single snapshot, not the full movie. I’ve seen countless brilliant entrepreneurs with thin credit files get rejected for loans they could easily repay, simply because the old system couldn’t see beyond a three-digit number. It’s a blunt instrument in a world that requires surgical precision.
The new way is so much smarter. With AI, we can build sophisticated machine learning models—things like gradient boosting machines—that analyze thousands of data points in real-time to create a much richer, more accurate picture of a borrower. We’re talking about going way beyond just payment history. Modern platforms can plug directly into a business’s accounting software, analyze their real-time cash flow, and look at alternative data that FICO ignores. Think about it: a history of successfully completed projects on Upwork or a steady stream of 5-star reviews on Yelp can be incredibly predictive of a small business owner's reliability. In one of my portfolio companies, we saw a 30% reduction in defaults in the first year just by implementing our own AI-powered credit scoring model. We were able to approve more loans for good people who were being ignored by the banks, and we did it with less risk.
The War on Fraud
Fraud is an existential threat in any lending business. In the early days of P2P, it was a constant, exhausting battle. We had teams of analysts manually reviewing applications, looking for red flags. It was slow, expensive, and frankly, a lot of sophisticated scams got through. A fraud ring could set up dozens of synthetic identities, apply for small loans across all of them, and be gone with the money before anyone even noticed a pattern.
AI has been our most powerful weapon in this fight. Instead of just looking for obvious lies, AI algorithms can detect incredibly subtle, complex patterns of fraudulent behavior that a human analyst would never catch. Using network analysis, an AI can see that 20 different applications, all with unique names and details, are secretly connected through a shared digital fingerprint—a common device ID, a recycled IP address, or a single doctored image used across multiple profiles. It’s an ongoing arms race, for sure, but AI gives us a fighting chance. It’s about stopping fraud before it happens, not just writing off the losses after the fact.
Investing on Autopilot
It’s not just about the borrower. P2P lending is a two-sided marketplace. On the other side, you have investors looking for a decent, stable return on their capital. The old way was a terrible experience. Investors had to spend hours scrolling through hundreds or thousands of individual loan listings, trying to pick the winners. It was time-consuming, and unless you were a professional credit analyst, it was mostly a shot in the dark. You’d end up either concentrating your risk in a few loans you thought were safe, or just giving up.
Now, AI can act as a personal, automated portfolio manager for every single investor. It’s the robo-advisor concept, but for private credit. Based on your stated risk tolerance—whether you’re conservative and want to stick to the safest loans, or you’re aggressive and want to chase higher yields—the platform can automatically build a diversified portfolio of hundreds of tiny loan pieces for you. It can analyze thousands of loans in a split second, balancing risk and return in a way no human ever could. As a recent article from Warwick Business School highlighted, AI can optimize a pool of loans to construct a portfolio that perfectly matches a lender's goals. It’s like having a sophisticated hedge fund strategy working for you, even if you’re only investing a few hundred dollars.
The Hard Questions: Ethics and Bias
Now, I’m not a blind evangelist. Using AI in lending comes with its own set of serious responsibilities. The biggest one is the risk of bias. If you train an AI model on historical lending data, which is full of historical human biases, the AI will learn and even amplify those biases. You could end up with a system that systematically discriminates against certain neighborhoods, ethnic groups, or types of people, all while hiding behind the excuse of a “neutral” algorithm. That’s not just wrong; it’s illegal.
This is where I get opinionated: you can't just throw data at a model and hope for the best. You have to be incredibly intentional. It means actively auditing your models for bias, testing for fairness across different demographic groups, and building in explainability so you can understand why the AI is making the decisions it is. The goal of AI shouldn't be to just replicate the old system, but to create a fairer one. It’s about finding those creditworthy individuals that the old, biased system overlooked.
My Experience: Seeing it Work in the Trenches
I’ve made over 200 angel investments, and a good number of them have been in the fintech space. I’ve seen this transformation from the front row. I’m an investor in a P2P platform that focuses on providing inventory financing for small e-commerce businesses. When they started, they were using a pretty standard underwriting process, heavily reliant on the owner’s personal credit score. They were growing, but their default rates were creeping up, and their approval process was taking days.
We brought in a data science team to build out their AI capabilities from the ground up. It was a huge investment of time and capital, but it paid off almost immediately. They plugged into e-commerce platforms like Shopify and Amazon, analyzing real-time sales data, customer reviews, and inventory turnover rates. The AI could predict with incredible accuracy which products would sell and which businesses were well-managed. The results were stunning. They’ve now reduced their default rate by over 50% while simultaneously increasing their loan origination volume by 200%. Why? Because the AI can make better, faster decisions. It can identify a great business with a hot-selling product that a traditional bank would turn down flat because the owner has a limited credit history. It’s not just about risk reduction; it’s about finding the hidden gems.
The Future of Lending is Here
So where does this all go from here? I think we’re still in the early innings of this transformation.
First, AI will make P2P lending even more mainstream and essential. As the models get better and the platforms become more efficient, the cost of capital will continue to drop, making P2P a primary option for a much broader range of borrowers and investors, not just an alternative.
Second, we’re going to see a rise in hyper-niche P2P platforms. Think platforms specifically for dentists funding new equipment, for freelance graphic designers smoothing out their income, or for sustainable farmers investing in new technology. AI makes it possible to build highly accurate credit models for these specific communities, something that was never economically feasible before.
Finally, the line between P2P lending and traditional banking will continue to blur into irrelevance. Banks are already scrambling to adopt AI in their own lending, and many are partnering with or acquiring the very fintech startups they once dismissed. The future isn’t about P2P vs. banks; it’s about who can use technology to best serve their customers.
The days of the big banks having an absolute monopoly on who gets access to capital are over. AI is the great equalizer, putting more power into the hands of individuals and small businesses. And as an investor and an entrepreneur, that’s a future I’m excited to keep betting on.
Frequently Asked Questions
How has this view evolved over time?
My thinking on most topics has changed significantly over the years. Early in my career, I held many conventional views that experience proved wrong. I try to update my beliefs when the evidence changes.
What experience informs this perspective?
This perspective comes from over a decade of building companies in Silicon Valley, two successful exits (RemoteTeam to Gusto, MovieLaLa to Gfycat), and investing in 200+ startups including Anthropic, OpenAI, and Scale AI. I write about what I've lived.
What's the most common pushback you get on this?
People often push back by citing exceptions or edge cases. And they're usually right that exceptions exist. But building a strategy around exceptions rather than patterns is a losing game for most founders.
How can I apply this thinking to my own situation?
Start by identifying the core principle behind the opinion, not the specific example. Then ask yourself: does this principle apply to my context? If yes, test it in a small, low-risk way before going all in.