Money 20/20 USA: Plaid CTO Will Robinson on Teaching AI the Grammar of Money

We like to say that our models have a master's degree in how money moves. So they're designed to understand what a transaction means deeply beyond just the line that might appear on a bank statement.

Will Robinson joins us here on the ground at Money 2020 in Las Vegas, Nevada. He is the CTO of Plaid. It's a brand undoubtedly will so many of our viewers very well familiar with. Really grateful for your time. Thanks for joining us.

Thanks so much. Great to be here.

How has that business evolved as artificial intelligence continues to transform the entire ecosystem of financial services.

The way we like to think about it is, just connecting everyone's financial data, that was the first challenge. And now making sense of that data that's flowing through that connected network, that's the next horizon for Plaid and and really for the industry as a whole. I mean, we connected 12,000 plus banks to 9,000 plus apps. We're really proud of the work we've done on connectivity. Um and then on the intelligence side, we're using a bunch of AI models. Uh we've announced a bunch of them recently in fact, uh foundational models that we think of as learning the vocabulary and the grammar of money.

You've built your own AI models as well specifically for financial data. Some people may see that headline, hear that bit of news and say, why not just partner with a more generalized large language model? We got plenty to choose from. Talk me through the thinking on that.

We like to say that our models have a master's degree in how money moves. So they're designed to understand what a transaction means deeply beyond just the line that might appear on a bank statement. Um so I love general purpose models. They're fantastic but general purposes right there in the name. And our models we are trained with specific representations, specific priors, features around repeatability and around explainability that are required to be used in a compliant way, in a trusted way in these financial use cases.

And the proof's ultimately in the pudding. So when we apply our models to, for example, transaction classification, we saw a 48% jump up in classifying which of these transactions is income. 14% jump up in classifying which transactions are a loan repayment. So the proof's in the pudding and we're really happy with where we are.

Where would you say artificial intelligence is already changing the way that we bank, maybe in ways that we don't even know?

Yeah, and you put it exactly right. So I think people are wondering, when's this going to happen? When's this going to happen from the point of view of the everyday consumer? And the reality is, it's already happening just in a way that's behind the scenes for them. So the way they'll feel it or or sometimes even not notice is just safer logins, right, less fraud, less account takeover, uh better rates on their loan application, a higher loan limit, faster approvals. Uh another great example we like to talk about is, um when you're onboarding to a new product, companies that are using Plaid's, uh, IDV products, uh signal products, we will, um, make it so that you don't have to re-enter your bank data. You don't have to re-enter your ID information. You just sort of seamlessly on board and that's the power of all this data coming together in the network, um whether through artificial intelligence, machine learning, or these other means.

Detecting fraud. What a big challenge for lenders and I know you're working with. You've recently launched some new tools to try and help with that exact particular challenge. How do they work? And what do you think most makes them different in this ecosystem?

Yeah, it's a great question. So, we've just announced a bunch of new things at our recent fall release. One of them is our, uh, most recent anti-fraud model. And the interesting thing about fraud is, in isolation, an instance of fraud usually looks innocuous. And the thing that we found is really, really powerful is when you look across the network, across not just what's happening on this app or with this session, but over time and across different apps, across different banks.

And Plaid's models are able to synthesize data and intelligence from across everything we see, all of the applications, all the bank accounts, going back over time, and deliver scores that are really powerful for anti-fraud. We see, for example, about a 40% improvement in uh rates of fraud detection, holding false positives constant when you use our latest models.

You're also now allowing bank accounts to be linked directly to AI assistants. Walk me through if I'm the consumer, what does that mean for me? What does that allow me to do that maybe is, maybe previously, Will, I was not able to do?

Absolutely. So, we've announced four of the six biggest AI labs as our customers. For example, ChatGPT from Open AI, Muse, the list goes on and on. And for most of these uh experiences, you go as a consumer, um, you'd say, I want to talk about my finances, or I want to save for college or whatever it is. And there would be an option that pops up right there in that flow as you're talking to that agent, saying, okay, well connect your financial information via Plaid, and we're going to make sense of it together and give you some personalized, customized advice.

So, then you go through a Plaid flow with one click, you can permission whichever accounts you want to make uh readable to that agent. And then you can basically talk to your money. And if you haven't tried this, I really encourage you to check it out. It's a pretty magical experience. It's not just what's good financial advice, but what's good financial advice for me right now. I think that's really, really cool to see. The thing for Plaid, though, is, although the experience is very different, the trust model, the core connectivity, like the work that we've done to build up this network and our trust with consumers for the last 13 years, that's still all basically the same.

And now, if I am a banking consumer, Will, you know this better than anyone in the industry. There's a big difference between allowing my AI assistants, my AI agents, whatever it is to have pure play visibility into the ins and outs of my banking reality versus allowing it to be empowered to act on my behalf. How do you draw that distinction? How do you build trust to make sure me as the consumer, I still have control over what all these cool AI tools are actually doing?

Yeah, it's a great question. It's something that I think we're going to see the industry sort of ladder up on at some gradual pace over time. But one thing that's really clear is the demand is certainly there. If you look just at ChatGPT for example, over 200 million people ask ChatGPT financial related questions every month. So, you know, people want this. Another really interesting stat there is two-thirds of people that are connecting to these, um, AI applications and using Plaid to link their financial data, they have not linked to any previous personal financial management application before. So this is a dramatic expansion of that market. It's not just, oh, I used to use these other tools, now I'm trying AI. It's people who weren't using anything before in most cases. And now they're saying, let me get serious about my finances. So that's really cool. In order to be trusted, in order to make changes on your behalf, though, a lot does need to change.

I think Plaid's place as a core consumer permission trusted, um, sort of clearing house for this data is going to be more central than ever.

And about out of time. What most excites you finally about the intersection of fintech with AI and where all of this cool technology is going?

Absolutely. Well, I would say there's just never been a better time to build. And I really, really do feel that way. Uh, AI is transforming everything in this space, not just for the labs that are the sort of natural first movers here, but I think every fintech, every bank, every startup founder or potential founder sitting in their garage right now has an opportunity to have to let people have a better, more empowered, more personal relationship with their money. And I'm incredibly excited about that.

Will Robinson, CTO at Plaid. I'm incredibly excited that you sat down with us here in Vegas. Thanks a lot for taking the time. Good to see you.