The next big customer for a payments company may not be a person at all. It may be an AI agent, and a startup called Natural has raised $30 million to serve it. The Series A, led by Kirsten Green at Forerunner Ventures, closed when Natural was just 193 days old, taking total funding past $40 million. The company sets out to take on Stripe, the dominant payment processor for human-run businesses, but with a twist: its target is the emerging world of autonomous AI agents.
Stripe, but the customer is a bot
The pitch is deceptively simple. Stripe built payment rails for businesses run by people. Natural wants to build the same thing for AI agents. Those agents, the company argues, are becoming financial actors. They will hold wallets, accept payments, pay invoices, buy things, and move money across banks and currencies, all without a human signing off each step. Natural says it owns the plumbing rather than renting it. That means running its own ledgering, multi-bank settlement, fraud checks, compliance, and even agent identity. It is one of a wave of startups building this layer.
The concept of an AI agent as a financial entity is not new, but the infrastructure to support it is still nascent. For example, an AI assistant that books a hotel room or orders supplies for a smart factory needs a way to pay without manual intervention. Natural provides that by creating what it calls “agent wallets” — FDIC-insured accounts held through partner banks — that can receive and send funds programmatically. It also offers “vaults” that take money in but never let an agent move it out alone, a safety feature to prevent runaway spending. The company has already launched six products, including tools to send, request, and transfer funds, plus a layer for building marketplaces where AI agents can trade.
What it is shipping
Six products are already live. They include FDIC-insured wallets for agents, held through a partner bank, and “vaults” that take money in but never let an agent move it out alone. There are also tools to send, request, and transfer funds, plus a layer for building marketplaces. More is coming fast. Natural plans agent-issued debit and charge cards, card details collected by voice, per-API-call billing, and even lines of credit for agents. It aims to ship 13 products in its first year, with a team of just 17 people. This rapid development cycle suggests a startup operating with extreme focus and urgency, aware that the window of opportunity may be narrow.
The technical challenges are significant. Unlike traditional payments, where humans can verify and authorize transactions, AI agents must be given pre-approved controls, spending limits, and fallback protocols. Natural’s ledgering system is built to handle high-frequency, low-value transactions typical of machine-to-machine payments. Multi-bank settlement ensures that funds can move across different financial institutions, while fraud detection algorithms are trained on patterns of agent behavior rather than human behavior. Identity verification for agents is handled through cryptographic signatures and API keys, creating a trust layer that can be revoked if an agent goes rogue.
On the compliance side, Natural must navigate a complex web of regulations. Know Your Customer (KYC) rules were designed for human entities, not software bots. The startup works with partner banks to ensure that agent wallets are ultimately traceable to a responsible human corporation or individual. Anti-Money Laundering (AML) checks are applied to the flow of funds, not the agent itself. This requires a rethinking of how financial services are licensed and regulated. Natural’s approach is to embed compliance into the code, using rules engines that flag suspicious transactions based on volume, velocity, and counterparties.
A bet ahead of the market
Here is the catch. Almost no one uses agentic payments yet. One estimate cited by Tech Funding News put real transaction volumes at about $28,000 a day worldwide earlier this year. That is a tiny fraction of the trillions processed by traditional payment systems. Yet the forecasts are grander. The same report pegs the agentic payments market at $7 billion now, growing to $93 billion by 2032. Natural is building at that future scale before the demand has arrived. Chief executive Kahlil Lalji is unbothered. “We made a bet on agentic payments a year ago, when no one was talking about it,” he said. His argument is that the question is not whether agents will move money, but who builds the rails.
The market drivers are clear: the proliferation of AI assistants, autonomous vehicles, robotic process automation (RPA), and IoT devices that need to transact. An autonomous drone that delivers a package may need to pay for landing fees or recharging. A virtual assistant that negotiates contract terms may need to handle deposits. Natural sees itself as the underlying financial layer for this new economy, much like Stripe became the layer for e-commerce. The difference is that Stripe grew alongside the internet economy; Natural is trying to pre-build infrastructure for an economy that is still emerging.
The risk is that the market may not materialize as quickly as expected, or that regulatory hurdles will slow adoption. However, the trend toward machine-to-machine payments is inevitable. Major tech companies are already experimenting with agent-to-agent transactions. For instance, Amazon’s Alexa can order goods, but the payment authorization still requires a human to say “Alexa, order…” and confirm. Natural aims to remove that human step, creating a fully autonomous payment flow. This is both an opportunity and a challenge, as removing human oversight introduces new risks of errors or misuse.
A crowded, nervous field
Natural is not alone, and its biggest threat is the obvious one. Stripe, now valued above $150 billion, is adding agentic features to its own platform. In 2024, Stripe launched a developer preview that allows AI agents to make payments on behalf of users, with the human ultimately approving high-value transactions. This is a cautious step, but it signals that Stripe sees the opportunity. Smaller rivals such as Ralio and Paygentic chase the same idea from seed stage, each with slightly different approaches: Ralio focuses on invoice management for agents, while Paygentic offers a checkout API for bot-friendly payments.
Natural’s backers are betting on speed and instinct. Forerunner made its name early on Chime, Glossier, and Warby Parker, and Green says Natural has “consumer DNA” in an infrastructure company. The risk, plainly, is letting agents loose with money before the guardrails are proven. Lalji acknowledges this challenge. “We are building guardrails that are stronger than anything in human payments,” he says. “Agents are not impulsive or emotional, but they can be programmed incorrectly. Our systems detect anomalies in real time and can freeze a wallet with a single API call.”
The shift from human-run to agent-run payments, Lalji insists, is already under way. If he is right, this looks like Stripe’s early years. If he is wrong, $40 million buys a lot of time to wait for the market to catch up. The company’s runway, combined with its lean team and aggressive product roadmap, suggests that it is ready to pivot if necessary. But the momentum is on its side. As AI agents become more capable and trusted, the need for dedicated payment infrastructure will only grow. Natural is positioning itself at the intersection of two mega-trends: the rise of AI and the digitization of finance.
In the coming months, industry watchers will be looking for signs of adoption. Are Fortune 500 companies building AI agents that need to pay suppliers? Are marketplaces emerging where agents trade digital services? Natural’s success will depend on finding early use cases that generate real transaction volume, even if small. The startup’s biggest win could be to become the default provider for a new category, just as PayPal became default for online payments in the late 1990s. The history of fintech shows that first movers with strong technology and regulatory compliance often dominate, even when bigger players enter later. Natural may have the same chance, provided the agentic economy arrives as forecasted.
For now, the company is focused on product development and integration with popular AI frameworks. It offers SDKs for Python, JavaScript, and Go, allowing developers to add payment capabilities to their agents with a few lines of code. This ease of integration is crucial, because the target audience — AI developers and machine learning engineers — want to quickly experiment without diving into compliance details. Natural abstracts that complexity, offering a single API that handles wallets, transfers, and fraud checks. The goal is to make agent payments as frictionless as adding a payment button to a website.
The future of work may involve armies of AI agents negotiating and transacting on behalf of humans. Natural wants to be the bank that holds the keys. It is a bold bet, but one that reflects the accelerating pace of innovation in the age of generative AI. Whether it succeeds or not, the company has already forced the industry to think about a question that was science fiction a decade ago: how should money move when the mover is a machine?