Stripe's AI journey as both vendor and client

Stripe headquarters in San Francisco on Dec. 3, 2020.
Stripe is in the midst of a product diversification initiative that includes generative AI.  
Bloomberg

Artificial intelligence is one of the few sectors of the technology industry where investment is growing quickly. It's also becoming a focus for payment firms and other fintechs.  

Stripe has upgraded its checkout suite to add more transaction options and expanded payments orchestration as well as ways to test new payment methods using minimal coding. It's one of the largest updates in the payment company's history. And while the upgrade is aimed at all merchants, it comes at a time in which Stripe is diversifying its revenue streams and expanding its presence in AI, such as through a growing partnership with ChatGPT developer OpenAI. 

Stripe is experimenting with ChatGPT and is working with OpenAI to commercialize the AI company's products. Stripe is also supplying payment and international tax compliance technology to OpenAI, which opened its first international offices in London and Dublin in the past few weeks. 

"OpenAI is at the cutting edge of innovation and we hope to be there with them," said Abhinav Tiwari, head of product for checkout experiences at Stripe. "There's a lot of folks that are trying to use OpenAI and we want them to use our one-click checkout." 

Stripe's work with OpenAI

OpenAI is a user of Stripe's checkout suite, which accepts payments for ChatGPT+, OpenAI's premium product. OpenAI's payment options for clients that use its products include one-time purchase, monthly and usage based subscriptions. OpenAI has added Stripe's tax products to support compliance in international markets, and another Stripe program that manages disputes and refunds. OpenAI has also deployed Link, Stripe's one-click buy button. 

"They can quickly sign up for a ChatGPT+ product and have an easy checkout experience," Tiwari said. "My biggest pet peeve is when you go to a payment page on the internet and have to punch in your credit card number. We want to eliminate that altogether." 

Stripe updated its checkout suite at the end of September and added more than 100 payment methods, using algorithms to present users with the most relevant payment option. 

"Offering access to over 100 global payment methods makes Stripe a more appealing partner to companies that do business all over the world and to companies that operate in countries that have popular payments that aren't used elsewhere," said Daniel Keyes, senior analyst for merchant services at Javelin Strategy & Research. 

Stripe named River Island, a multichannel fashion retailer, as a client at launch for the optimized checkout suite, and did not comment on plans to migrate its merchants to the upgraded version. Stripe, which is competing with dozens of payment technology companies and merchant acquirers in a challenging economy, is attempting to diversify merchant services and expand relationships with clients that increasingly include generative AI companies. 

Stripe's generative AI clients include Digital Runway, Diagram and Moonbeam. Moonbeam  uses generative AI to help content producers write blog posts, essays and other articles. Diagram is an AI-powered design company and Runway generates video from text.    

Runway uses Stripe for payments and tax filing, and has used other Stripe products to digitize paperwork involved in establishing and running its business. Diagram uses Stripe to accept payments and update payments technology without using programming code. Moonbeam manages subscriptions, cancellations and one-time payments via Stripe.

"As these new AI companies proliferate, we're helping them with monetization," Tiwari said. "And these companies are helping us expand revenue sources." 

OpenAI has become one of the most recognizable companies in generative AI, largely due to its ChatGPT product that produces original content. Banks and payment companies are considering how to use generative AI, attracting a lot of investment over the past year. Amazon, for example, recently pledged to invest up to $5 billion in generative AI firm Anthropic, which has a language program similar to OpenAI's ChatGPT. Amazon could use its investment in AI to fuel its expansion into point-of-sale payments and merchant services, putting it in competition with Stripe. Among Stripe's traditional rivals, Square has developed more than 30 AI tools in the past year and PayPal has touted the technology's ability to save costs for "years to come." 

Stripe's work with OpenAI ties Stripe to the emergence of AI, potentially helping it add clients in the space, Keyes said. 

"Additionally, Stripe has the opportunity to leverage OpenAI's capabilities and see if it can find ways to use AI to improve its solutions and build new ones," Keyes said. 

Stripe's work beyond AI

In an effort to reach clients in and outside of the AI market, Stripe's updated algorithms expand the company's ability to support payments orchestration. 

Payments orchestration, which analyzes multiple data sources to automatically route a transaction to the easiest and most economical option, has become popular with investors in the payments technology market due to its ability to shave time and cost for payments, particularly those in different countries.  

For example, Stripe's orchestration engine would automatically recommend a French payment option for shoppers visiting Japan, but a local mobile wallet for Japanese shoppers making the same purchases. Stripe has additionally introduced an A/B testing tool that uses no-code programming. 

A/B testing refers to measuring different versions of the same content, such as a marketing or checkout message, against one another simultaneously. No-code programming uses visual elements in place of programming code, a method designed for non-technical users. Banks such as Western Union and NatWest are using no-code in their payment upgrade projects, and the method has gained steam with fintechs. 

Stripe is also attempting to shorten checkout time as a way to mitigate cart abandonment. 

Sixty-six percent of users begin to get "impatient" when checkout time takes four minutes or more, according to Capterra, which reports 28% get impatient at two minutes or more. 

"A lot of websites aren't optimized for checkout," Tiwari said, adding Stripe's goal is to reduce average checkout time to less than one minute. "When you get to the last page of checkout any customer you lose is an expensive proposition."

For reprint and licensing requests for this article, click here.
Payments Artificial intelligence
MORE FROM AMERICAN BANKER