Illustration: Maura Losch/Axios
Fintech's adoption of generative AI was on full display at last week's Money20/20 conference.
Why it matters: Fintechs see gen AI as a tool for use cases ranging from fraud prevention to customer service that can lower operational costs, increase internal efficiencies and drive increased revenues.
Context: Money20/20 featured keynote speakers from AI vendors like OpenAI, Anthropic and Nvidia, as well as early adopters like Stripe and Klarna.
- Even companies not there to explicitly discuss it had some AI story to tell.
The big picture: Fintech's adoption of AI underlies the booming industry that is raising huge amounts of venture funding and, more important, generating real revenue.
- "The AI economy is very real," Stripe head of information Emily Glassberg Sands said during a panel discussion. "We're seeing AI startups growing at unprecedented peace, not just in terms of their compute or spend, but in terms of their revenues and the breadth of customers they're reaching."
- In a recent LinkedIn post, she found that the 100 highest-grossing AI companies on Stripe were growing at 2-3x the pace of the highest-grossing SaaS startups on Stripe during a comparable stage of that tech's adoption in 2018.
Between the lines: Most companies we talked to were careful to say they were agnostic about the LLMs they used or AI companies they partnered with.
- "We deeply believe there will not be one model to rule them all. The model you want to use will depend on the use case and the trade-offs," Sands said in an interview with Axios.
- Credit Karma CEO Joe Kauffman echoed that sentiment, saying his company has "commercial relationships with all the usual suspects" but chooses the best technology based on its use case.
- One notable exception is Klarna, which has been very open about its partnership with OpenAI and its adoption of ChatGPT, in particular.
Yes, but: "We'll use whatever technology is out there that will accelerate our vision and our products," Klarna CTO Yaron Shaer said in an interview with Axios.
State of play: Money20/20 attendees talked up AI chatbots to help employees find answers faster, tools to better inform sales reps about prospects' business needs, and code co-pilots to make software engineers more efficient.
- Customer-facing uses included customer service agents who more efficiently solved user problems, product recommendation engines to drive engagement and boost sales, and AI-enabled checkout experiences to reduce cart abandonment.
The big picture: Stripe has over 100 gen AI models in use companywide, Sands says, for both internal and customer-facing use cases.
- Externally, the company's fraud prevention tool, Radar, has helped reduce fraud across the Stripe network by 10% year over year at a time when online fraud has risen by over 10% industrywide.
- Its AI-powered checkout flow can now show prices in a buyer's local currency, which has helped increase international sales by 17% for merchants with global businesses.
- Its Sigma assistant is a natural language interface that enables customers to ask questions using Stripe data to understand their business better.
Internally, Stripe has deployed a shared infrastructure for LLMs that Sands says is now used by more than half of the company's employees to find answers to common questions and share prompts with one another.
- Sales team members also use LLMs to "deeply understand how each product will meet the unique needs of each prospect," Sands says.
Klarna has been vocal about the benefits of adopting AI to power its customer communications and product recommendation engine.
- Earlier this year, it said its AI agent could handle two-thirds of all customer service chats, doing the work of 700 full-time agents and resulting in a 25% drop in repeat inquiries.
- Recently, it rolled out an AI-assistant product recommendation engine that offered users a conversational and personalized shopping experience.
- Internally, Klarna made its enterprise search product available to all employees and says gen AI tools are used by 90% of its workforce daily.
- "This is more than just a cool technology. This is something that can transform our company and the industry to accelerate our growth and innovation," Klarna's Shaer says.
Other firms outlined gen AI uses:
- Credit Karma has adopted its parent company's conversational chatbot, Intuit Assist, to answer users' money questions.
- Plaid uses AI to understand end-user cashflows and to power income verification for underwriting, among other use cases.
- PayPal leverages the technology for fraud prevention, to optimize payment options at checkout, and, soon, to provide more personalized product recommendations.
- Morgan Stanley uses LLMs to help wealth managers provide better, more personalized advice to their clients.
- Visa says it has built over 500 generative AI applications.
What's next: Fintechs see an opportunity for AI not just to help users do more faster but to complete tasks on behalf of those users in the future.
- Klarna's CTO shared a vision for a financial assistant that scans consumer spending habits and suggests saving opportunities to them.
- "We believe that retail banking in the financial industry will be transformed as a result of companies like Klarna, and tech like AI can help us accelerate that," Shaer says.
- Credit Karma's Kauffman took that a step further, saying his company is working to roll out so-called "done for you" experiences, moving from recommendations to actions.
- "Historically, we've made a recommendation … but the idea is you click a button, and automatically, in the background, we're paying down credit card debt for you," he says.
The bottom line: The number of gen AI applications for fintechs is vast, but they are just getting started.
