AI agents could control 5% of cross‑border payments

In a recent interview with PYMNTS, Mural co‑founder Sinclair Toffa said that AI agents could account for about 5% of cross‑border payment transactions within the next year. The claim, made in a conversation with PYMNTS CEO Karen Webster, highlights a shift from traditional banking interfaces to autonomous software that can initiate, approve and audit payments across currencies and jurisdictions.

What’s Happening

Toffa explained that the convergence of stablecoins, real‑time payment networks and artificial intelligence is forcing financial infrastructure providers to rethink how they build trust and governance into software. He compared the rise of AI agents to the transition from desktop computing to mobile apps, noting that the new interface will need dedicated financial accounts, granular permissions and transaction‑level audit trails. “Agents are the new interface,” Toffa said, adding that the 5% forecast is aggressive given the unresolved questions around authorization, liability and oversight.

According to Toffa, Mural is working to standardize the software objects that connect existing networks, allowing businesses to interact with fragmented financial systems through a common application programming interface that embeds compliance records, permissions and audit trails into the underlying financial software. He cited Starbucks as an example of a company that has integrated financial functionality into its customer experience, using embedded wallets, interest‑bearing accounts, rewards and foreign‑exchange capabilities to deepen relationships and capture more economic value.

Why It Matters

For the payments industry, the shift to autonomous agents means that banks, payment networks, stable‑coin issuers and infrastructure providers are converging on the same opportunity from different positions in the stack. Mural’s bet is that a standardized compliance and permissioning layer can create value without replacing the institutions underneath.

Consumers and businesses that operate across multiple currencies stand to benefit from reduced foreign‑exchange friction. In Latin America, for instance, businesses often convert dollars to local currency and back again to pay overseas suppliers, incurring repeated FX costs. Stable‑coin‑enabled products could lower some of that friction, though conversion, redemption and liquidity risks remain.

For investors, the potential market for AI‑enabled payment agents is sizable. If the 5% figure is reached, it would represent a significant portion of the global cross‑border payments market, which is worth trillions of dollars annually. The technology also opens new revenue streams for fintech firms that can offer compliance‑ready, agent‑native solutions to enterprises.

See how Amazon is boosting Alexa for shopping to counter AI agents as companies look to protect their own transaction ecosystems.

What It Means for Investors and Businesses

Investors should watch for:

• Companies that build agent‑native platforms or compliance layers, such as Mural, and their ability to secure partnerships with banks and payment networks.
• Regulatory developments that clarify liability and oversight for autonomous payment agents.
• Adoption rates among enterprises that operate across multiple currencies, especially in regions with high FX friction.

For businesses, the key takeaway is that trust and control will be the deciding factors in whether autonomous payments can scale. Without clear authorization protocols and audit trails, the risk of unauthorized transfers or errors could outweigh the speed benefits. Companies that can embed these safeguards into their own platforms will be better positioned to capture the economic value of autonomous payments.

What to Watch Next

Toffa’s comments raise questions about who will ultimately control the risk. The next few months will likely see:

1. Regulatory bodies issuing guidance on AI‑driven payment authorization and liability.
2. Fintech firms announcing pilot programs that test agent‑native payment flows in real‑world scenarios.
3. Research that explores vulnerabilities, such as the recent study that showed how AI agents can be tricked into transferring user funds.

These developments will determine whether the 5% forecast becomes a reality and how the payments ecosystem adapts to the new autonomous interface.

Sources

• AI Agents Could Control 5% of Cross-Border Payments. Who Controls Them? — pymnts.com