Here's a fun thought experiment: imagine you deploy an AI agent to manage your company's freight logistics. It finds vendors, compares rates, negotiates terms, schedules delivery—and then grinds to a halt because it needs a human to approve a wire transfer. That's not an agentic workflow. That's a very expensive to-do list.

This is the exact crack in the foundation that a startup called Natural is betting $30 million it can fill. The company just closed a Series A led by Forerunner Ventures, bringing its total funding to $40M, with a mission to rebuild payment infrastructure from scratch for AI agents rather than bolting autonomy onto systems designed for humans clicking "confirm purchase" in 2003.

Why Today's Payment Rails Are an Agentic Dead End

Credit cards, ACH transfers, and most traditional payment systems share a foundational assumption: a human authorized this. That assumption is baked deep into the architecture—authentication flows, dispute resolution, fraud detection, all of it. When an AI agent tries to operate inside that system autonomously, you're not hitting a policy wall. You're hitting a structural one.

Think of it like trying to run modern containerized microservices on a mainframe from 1978. Technically possible, in theory. Practically, a nightmare that nobody with a deadline should attempt.

Natural's positioning as an "agent orchestration layer" means it's not just wrapping existing rails in an API. According to the company, it's rearchitecting how funds move and are stored when the transacting parties are software agents—including agent-to-agent transactions, autonomous vendor payments, and fund collection without human intervention at each step.

The Team That Couldn't Quit Finance

CEO Kahlil Lalji has a complicated relationship with fintech. His previous startup, Ivella—a YC-backed banking product for couples—was acquired by Earnin in 2023. After two years there as an engineer, he reportedly wanted out of the sector entirely. Then the agentic payments opportunity kept staring him in the face.

He eventually co-founded Natural in 2025 alongside Eric Wang (his Ivella co-founder) and Walt Leung, a former engineering manager at Nextdoor. The team has since recruited senior engineers from Stripe, Ramp, and Square—which is either a sign of genuine technical credibility or a very expensive LinkedIn flex. Probably both.

Forerunner's Kirsten Green led the round, which tracks—her firm focuses on the future of commerce, and autonomous agent-driven transactions are exactly where consumer commerce is heading whether anyone is ready or not.

So Who Else Is Racing for the Same Prize?

Stripe isn't sitting still. The payments giant has been openly building features for AI agent use cases, and with its distribution, brand, and existing developer relationships, it starts with a structural advantage that $40M doesn't simply overcome.

Then there's Skyfire Systems, backed by DCVC, which is approaching the same problem via USD-backed stablecoins—essentially treating crypto's programmability as the path to agent-native payments. Natural plans to support stablecoins too, but it's also building for traditional bank payments, suggesting a hedge-your-bets approach rather than a crypto-native bet.

• Natural: Traditional bank payments + stablecoin support, full infrastructure rearchitect
• Stripe: Extending existing rails with agent-focused APIs and authentication layers
• Skyfire Systems: Stablecoin-first, programmable money for autonomous agents

The strategic question isn't who has the best tech demo. It's who gets embedded in enough agent frameworks and enterprise workflows to become the default before the market consolidates.

The Orders-of-Magnitude Argument

Lalji made a claim that's either visionary or wildly optimistic, depending on your priors: if transactions happen at machine speed rather than human speed, the total volume of payments in the world could be two to four orders of magnitude greater than today. That's 100x to 10,000x current transaction volumes.

Is that number grounded in anything rigorous? Not that we can verify from the outside. But the directional logic isn't crazy. Every time compute gets dramatically cheaper, things that used to be done once get done thousands of times. If agents are negotiating and transacting autonomously at scale across supply chains, financial markets, and consumer services simultaneously, the payment volume math does get very interesting very fast.

The flip side: that same scale creates fraud surface area, settlement risk, and regulatory exposure that would make any compliance team's eyes water. The infrastructure that handles 100x transaction volume also needs to handle 100x of everything that goes wrong.

Hot Take

Natural has a real thesis and a real problem to solve—give them credit for that. But the brutal truth is that payment infrastructure is one of the hardest moats to build against incumbents who already have banking relationships, regulatory licenses, and millions of developers using their SDKs. Natural is essentially trying to do to Stripe what Stripe once did to the legacy processors—and Stripe had the advantage of a genuinely broken status quo and a decade-long runway to build trust.

Prediction: Natural won't beat Stripe head-on, but it doesn't need to. The agentic payments space will likely fragment, with specialized infrastructure players capturing vertical niches—freight, healthcare, autonomous trading—before any single platform wins the whole stack. Natural's best outcome isn't dethroning Stripe; it's becoming the default payment layer for one or two high-volume agentic verticals and getting acquired for a number that makes everyone happy.

What Should You Actually Watch For?

Natural has been in beta. The real test isn't the fundraise—it's whether enterprise customers start building production agentic workflows on top of Natural's infrastructure rather than waiting for Stripe's next product release. Watch for design partner announcements and whether any major agent framework (LangChain, AutoGen, or the emerging crop of vertical agent platforms) ships a native Natural integration.

If that happens in the next 12 months, the $30M looks prescient. If Natural is still "in beta" by late 2026, the window may have closed.

What do you think? Is purpose-built agentic payment infrastructure a genuine architectural necessity—or can Stripe just ship a few new APIs and make this a non-problem? Drop your take in the comments.