Marqeta, a card-issuing platform, has rolled out an AI tool that that leverages Open AI’s Language Learning Model.
Through Marqeta Docs AI, users can ask certain questions and receive bespoke responses to their particular use cases, Finextra reports. Marqeta is betting on the tool to help its clients gain a better understanding of how to incorporate embedded payment solutions—including BNPL, earned wage access, expense management, processing, as well as issuance of physical cards—within their sites.
“One of the points we’ve stressed with clients is that they are most likely to consume generative AI products through vendor-partners, as Marqeta is doing,” said Christopher Miller, Lead Analyst of Emerging Payments at Javelin Strategy & Research. “This offers opportunities for earlier availability and vertical specialization, while also making it impossible for companies to avoid, as their existing partners are likely to implement generative AI-based tools relatively quickly. The result is that companies have to develop an understanding of the strengths and weakness of this technology regardless of their own viewpoints on its fitness for their strategic goals. “
“At this stage of the game, it’s less important to focus on the efficiency number claims, and more relevant to focus on fitness to task. If the task is one that fits with generative AI’s approach to data, then the efficiency will come. If it isn’t, there’s no way to achieve value in the long run,” he said.
Generative AI is Gaining Traction
For many forward-thinking businesses, generative AI has the potential to enhance their current workflow and streamline more convoluted systems. For payments, generative AI can potentially improve the consumer experience as well as speed up payments.
As with any new technology, generative AI is not without risk—and security is certainly top-of-mind. Banks who currently use generative AI models do so with the use of APIs which are sent from their private data centers, which can lead to compliance risk. Several security breaches have already taken place as a result.
It’s recommended that generative AI not be used for any type of client-facing applications, at least for the time being—and to reserve this type of use case for internal purposes only.
