Nandan Sheth, CEO of Splitit – Interview Series

Nandan Sheth, CEO of Splitit, is a seasoned fintech executive and entrepreneur with deep expertise in payments, digital commerce, and financial infrastructure, having led the company since 2022 while also serving on its board. Prior to Splitit, he spent five years at Fiserv as Head of Carat & Digital Commerce, where he helped shape modern payment ecosystems, and earlier co-founded Acculynk, pioneering secure online payment authentication technologies. His career also includes leadership roles at American Express following the acquisition of Harbor Payments, a company he co-founded and scaled into a major electronic billing platform. Across more than two decades, Sheth has consistently focused on building payment innovations that reduce friction, enhance security, and improve merchant economics, positioning him to lead Splitit’s evolution into a next-generation buy now, pay later infrastructure provider.

Splitit is a fintech company focused on transforming buy now, pay later (BNPL) into a merchant-first infrastructure layer rather than a consumer-facing lending product. Through its platform, the company enables shoppers to split purchases into installments using their existing credit cards, eliminating the need for new loans, credit checks, or lengthy applications. Its core innovation lies in a white-label “Installments-as-a-Service” model that allows merchants to fully embed BNPL into their own checkout experience, maintaining ownership of customer relationships and data while improving conversion rates and average order value. By leveraging existing credit card networks and integrating directly into merchant systems via a single API, Splitit positions itself as a lower-risk, more seamless alternative to traditional BNPL providers, aligning with increasing regulatory scrutiny while offering a frictionless, brand-controlled payment experience.

You have built and exited multiple payments companies, including Harbor Payments and Acculynk. How has that experience shaped your perspective on why payment certainty may become the defining factor in the rise of agentic commerce?

My career has focused on the payment layer, the part of commerce that few notice unless it fails. I’ve learned that discovery gets attention, but authorization ultimately enables revenue.

In agentic commerce, where AI agents influence shopping, agents prefer payment methods that are predictable, fast, and likely to clear. Payment uncertainty creates friction and might lead agents to avoid certain merchants or products. Payment certainty shifts from a back-end concern to a key factor in recommendations.

Splitit recently launched its Agentic Commerce Partner Program. How does embedding card-linked installments into autonomous AI shopping agents address the payment certainty bottleneck that you believe is limiting conversion today?

By embedding card-linked pay-later capabilities into autonomous shopping agents, Splitit’s Agentic Commerce Partner Program moves installment payments purchases upstream in the AI journey, not just at checkout. This lets the agent evaluate affordability and payment fit while it narrows options and decides what to buy.

Another benefit is that our model uses existing cards and rails, making payments more reliable and easier for AI agents. This solves a big challenge: many conversions fail due to payment uncertainty rather than product fit. If consumers must apply for new credit or wait for approval, the process breaks. Drawing on existing credit speeds up the process.

In practical terms, how does your program allow AI agents to factor affordability into recommendations using existing cards and payment rails, without requiring new credit applications or account creation?

Splitit allows the agent to convert a total purchase price into a monthly payment using a card the shopper already has. That is vastly different from pushing someone into a separate lending flow.

The shopper uses their existing card without filing a new application, opening a new account, or leaving for a third-party site. The installment plan stays within the shopper’s current bank relationship, bringing affordability into the decision-making process earlier and helping AI agents assess not just a product’s features and price but also whether the shopper can realistically complete the purchase.

You argue that AI-driven discovery is already ahead of conversion. Where specifically do payments create friction in the agentic commerce flow?

Friction occurs in three areas: eligibility, authorization, and workflow. A shopper may find the right product via AI, but the process can fail if the payment option requires a credit decision, results in unpredictable authorization, or needs a separate application or approval.

This is the gap between discovery and conversion. AI already drives high-intent retail traffic, but payment infrastructure lags. The opportunity is there. The challenge is making purchase completion as seamless as discovery.

Many merchants rely on Buy Now Pay Later marketplaces today. How does a card-based installment model differ from traditional Buy Now Pay Later platforms when integrated into AI-driven purchasing journeys?

Our card-based installment model uses the consumer’s existing credit, whereas traditional BNPL often asks the shopper to apply for new credit at the time of sale. That difference matters in AI-driven purchasing journeys because every new credit decision introduces the risk of a decline. When that happens too often, the AI agent starts to deprioritize merchants.

Traditional BNPL requires more steps and new brands. Our model keeps merchants in control and ensures that shoppers use trusted cards, reducing uncertainty for AI agents.

From a technical standpoint, is checkout optimization becoming less important than authorization predictability in an agentic environment?

Checkout optimization matters, but authorization predictability matters more. In other words, a clear, simple buying flow still helps, but payment approval ultimately drives the outcome. In traditional commerce, companies focused on front-end efficiency because human shoppers handled each step themselves. In agentic commerce, AI agents handle much of that navigation.

The harder problem is whether the payment will clear in a stable, low-friction way. If the authorization path—the process by which banks or payment networks approve a transaction—is unreliable, a beautifully designed checkout does not solve the real problem. In this environment, authorization predictability becomes part of commerce performance, not just payment operations.

As autonomous agents begin to make purchase decisions on behalf of consumers, what new compliance or regulatory considerations should financial technology companies prepare for?

Consent is key. Companies must define agent authority and clarify approval requirements.

Accountability follows. There must be clear audits for agent purchases and limit exceedances.

Control is essential. Companies need robust permissions, limits, and exception logic.

In my view, the payments layer must enable agent purchases and ensure accountability. That requires robust security, clear authorization, and well-defined consent. As humans step further from transactions, strong governance at the payment layer becomes fundamental to building trust in agentic commerce.

How do installments influence AI recommendation engines differently than traditional checkout options? Does changing affordability meaningfully alter how agents rank or prioritize products?

Traditional checkout appears after product selection. Installments are influenced earlier by changing affordability. Products out of reach at full price become viable with predictable card-based installments. This changes how AI agents rank options: they consider not just product fit, but also the realistic ability to purchase.

What signals or metrics are you watching to determine whether agentic commerce is moving from experimentation to scaled adoption?

Five signals can indicate when agentic commerce moves from novelty to a scalable channel that reshapes transactions.

First, monitor the share of commerce traffic driven by AI-powered shopping journeys. That shows whether consumers are adopting the technology, not just testing it.

Second, watch the conversion quality. It matters that AI-driven sessions convert at meaningful rates, not just generate clicks.

Third, track whether payment methods with greater authorization confidence gain a share of recommendations. That would show payment certainty shaping agent behavior.

Fourth, look for deeper integration. When merchants, platforms, and payment providers embed payments directly into agentic workflows, they turn experimentation into infrastructure.

Fifth, watch for higher approval rates, conversion rates, and average order value when affordability is built into recommendations.

Looking ahead, do you see agent-driven transactions extending beyond e-commerce into areas such as B2B procurement or subscription management?

E-commerce is the first step, not the last. Agents add value to any purchase process with set rules and budgets. B2B procurement and subscription management are obvious examples.

All this relies on a payment layer that companies trust, connect to, and embed. That’s why Splitit supports open standards like Google’s Universal Commerce Protocol to enable real agentic transactions across categories.

Thank you for the great interview, anyone wishing to learn more should visit Splitit.