Natural Raises $30M to Build Payment Infrastructure for AI Agents
AI agents are becoming capable of finding suppliers, comparing prices, negotiating tasks, and coordinating services. The payment step, however, remains largely built around human approval. Natural, a fintech startup founded in 2025, is trying to remove that bottleneck after raising $30 million in Series A funding to build infrastructure specifically for AI agent payments.
 |
| Credit: Natural |
The important point is not simply that another fintech startup has raised venture capital. Natural is betting that AI agents will eventually need a financial system designed around software acting at machine speed rather than people manually approving every transaction.
What happened: Natural raises $30 million for AI agent payments
Natural has raised a $30 million Series A round led by Forerunner founder and managing partner Kirsten Green. The financing brings the startup's total funding to $40 million.
The company was founded in 2025 by Kahlil Lalji, Eric Wang, and Walt Leung. Lalji and Wang previously worked together at Ivella, a financial product for couples that was acquired by Earnin in 2023. Leung previously served as an engineering manager at Nextdoor.
Natural is developing what it describes as an agent orchestration layer for financial transactions. Its infrastructure is intended to allow companies to give AI agents the ability to make payments, receive funds, store money, and transact with other agents.
The startup had been operating in beta. According to Lalji, Natural has now made enough architectural decisions to move toward a more ambitious effort to compete with established payment infrastructure providers.
The company has also recruited senior employees with backgrounds at companies including Stripe, Ramp, and Square.
Why AI agent payments are becoming a separate infrastructure problem
Traditional payment systems were primarily designed around people and businesses initiating transactions. A person enters payment details, approves a purchase, or authorizes a bank transfer. Businesses then use established systems to process the transaction.
Autonomous AI agents introduce a different pattern.
An agent might identify a supplier, compare offers, communicate with a vendor, arrange a delivery, and then need to pay for the transaction. If the agent must stop and wait for a person to approve every financial action, the automation chain becomes less autonomous.
That creates a problem that is broader than simply adding an AI interface to existing checkout systems. An agent may need clearly defined permissions, access to funds, ways to transact with other software, and systems for handling failed or disputed transactions.
Natural's argument is that these requirements may justify a new financial infrastructure layer rather than a collection of minor changes to existing payment tools.
That distinction matters. The question is not merely whether an AI can click a payment button. It is whether a software system can safely operate within financial boundaries that humans define in advance.
Natural is targeting more than automated checkout
Natural's ambitions extend beyond helping AI agents purchase products or services for consumers.
The startup is also working on infrastructure for agents to collect money and transact with one another. That could eventually support business workflows in which software systems negotiate services, purchase resources, or settle transactions without a human manually handling each financial step.
The company is also looking at the way disputed transactions are handled. That is an important detail because autonomous payments create questions that traditional systems were not necessarily designed to answer.
If an agent makes a purchase based on a set of instructions, who is responsible when the transaction is incorrect? What happens if the agent is manipulated? How should permissions be revoked? How can a company limit the amount an agent can spend?
Natural has not publicly solved every one of those problems through the information available here. But its focus on the underlying payment architecture suggests that the company sees financial control and transaction management as central parts of the problem, not secondary features.
Natural's technology bet includes bank payments and stablecoins
Natural plans to support traditional bank payments while also incorporating stablecoins into its architecture.
That approach sets it apart from startups pursuing AI agent payments through a more narrowly focused digital-currency model. The company is effectively betting that autonomous commerce may require multiple types of financial rails rather than one universal replacement for existing systems.
Traditional bank payments offer compatibility with the financial infrastructure businesses already use. Stablecoins, meanwhile, could potentially provide a programmable settlement option for transactions between software systems.
The available information does not establish which model will ultimately dominate. Natural's decision to support both reflects a practical reality: AI agents may need to operate across different payment environments, depending on the users, businesses, and systems involved.
The real challenge is permission, not payment speed
Natural's most interesting opportunity may not be making payments faster. It may be defining how much financial authority an AI agent should have in the first place.
The conventional description of agentic payments focuses on speed: an AI finds something, decides to buy it, and completes the transaction without waiting for a person. But speed is only useful if the system can reliably determine what the agent is allowed to do.
That makes payment infrastructure inseparable from permissions and accountability.
An AI agent that can spend money autonomously needs rules about limits, approved vendors, transaction types, timing, and exceptions. It also needs a way to stop or reverse activity when something goes wrong. In that sense, the company that builds the financial layer for agents may be building a control system as much as a payment system.
This is why Natural's attempt to address disputed transactions could become more important than the simple ability to connect an agent to a bank account. The company that helps businesses trust autonomous financial activity may have a stronger long-term position than one that merely makes checkout invisible.
That is an interpretation rather than a confirmed prediction about Natural's future. But the architecture of AI agent payments makes the distinction significant: autonomous transactions require both execution and control.
What the funding means for the wider industry
Natural's $30 million round shows that investors see payment infrastructure as a potentially important layer of the AI agent economy.
The company is not operating in isolation. Established payment infrastructure providers are also exploring how their systems could evolve for AI-driven transactions, while other startups are developing alternative approaches, including stablecoin-based systems.
That competition could benefit developers and businesses if it leads to more standardized tools for giving agents controlled access to financial services.
For developers, the practical appeal is clear. Instead of building separate systems for payment authorization, fund storage, collection, and transaction monitoring, they could eventually use specialized infrastructure designed for autonomous software.
Businesses could also gain more flexibility in automating workflows that currently require employees to approve or execute individual transactions.
But the risks are equally practical. A mistake made by a human can affect one transaction. An automated agent operating at scale could potentially repeat the same mistake many times before anyone notices. The infrastructure supporting agent payments will therefore need to make monitoring and limits as important as convenience.
The next phase of AI agents may depend on financial autonomy
Natural's funding arrives as AI agents move from answering questions toward performing multi-step tasks. The more useful these systems become, the more often they will encounter situations where completing the task requires money to change hands.
That creates a natural pressure to build payment systems around autonomous software.
The company's challenge is that financial transactions are less forgiving than many other agent tasks. An agent can usually retry a search or rewrite a message. A payment can create legal, financial, and operational consequences.
Natural's focus on a dedicated infrastructure layer is therefore a logical response to a problem that existing payment systems may not fully address on their own. Whether the startup can compete with much larger incumbents remains uncertain, particularly as established companies develop their own tools for AI-driven commerce.
For now, the most important takeaway is that AI agents cannot become truly autonomous simply by becoming better at reasoning. They also need controlled access to the systems that make the real world operate, including money.
Natural's $30 million raise is a bet that payments will be one of the most difficult—and potentially most valuable—of those infrastructure layers. The winners in this market may not be the companies that let agents spend money with the fewest clicks. They may be the ones that make autonomous spending controllable enough for businesses to trust.
