Key Takeaways
- SolvaPay is building payment infrastructure that lets artificial‑intelligence agents make machine‑readable, usage‑based payments while staying inside existing financial rails.
- Current fraud‑prevention tools (3DS, CVV, etc.) were designed to stop bots, not to differentiate between malicious bots and authorized AI agents.
- Agentic commerce can generate multi‑layered payment chains—one high‑level task may spawn dozens of micro‑transactions between agents, complicating reconciliation.
- Usage‑based billing aligns naturally with how agents consume APIs, data sets, or compute services, shifting away from flat‑rate subscriptions.
- For agents to operate safely, payments must still comply with AML, licensing, ledger, and security rules; there is no “separate” payments environment for autonomous software.
- In the future, agents may choose payment instruments on behalf of users (e.g., picking the card that earns the most miles) based on programmed objectives.
Payment Infrastructure for AI Agents
SolvaPay, a Stockholm‑based FinTech, is creating the rails that allow artificial‑intelligence agents to transact autonomously. According to CEO and co‑founder Viggo Stenseth, the company focuses on “machine‑readable payments, usage‑based billing and transactions between agents, along with controls over what an agent can spend.” He explained that the goal is to give software the same ability to move money that humans enjoy, but without bypassing the safeguards that underlie today’s payment networks. “We built really robust systems to verify that there’s a human doing the transaction, 3DS, CVV codes, everything to prevent a bad bot from doing things,” Stenseth said, noting that those controls are “really ingrained in the infrastructure.” The challenge, therefore, is not to dismantle those protections but to adapt them so they can recognize an authorized agent as a legitimate payer.
The Challenge of Existing Fraud Controls
Today’s fraud‑prevention mechanisms were erected to stop malicious bots from masquerading as customers. Merchants cannot simply discard those safeguards when an automated buyer has permission to spend; they must still tell the difference between a benign agent and harmful software while determining whose funds the agent may use and what it may buy. Stenseth observed that businesses have had little time to grapple with this question, comparing it to the early days of the web: “I recalled selling companies on websites in the 1990s, when many weren’t convinced that they needed an internet presence. It took years for widespread eCommerce to follow.” Agentic AI, however, is arriving on a much shorter timetable, and the payment ecosystem has not yet caught up. “The timer already started a couple of years ago,” he said. “Agents can now accomplish tasks autonomously, but the billing, the payment, the flow hasn’t really been solved yet.”
Timing and Adoption Parallels
Stenseth draws a clear parallel between the diffusion of smartphones and the rise of agentic commerce. Just as mobile devices appeared before many businesses had built payment experiences around them, AI agents are gaining capabilities faster than the financial infrastructure can adapt. “Financial infrastructure moves deliberately because payments require controls, checks and regulatory oversight. Agentic software can change more quickly,” he noted. SolvaPay’s work aims to bridge that speed gap without creating a siloed payments environment for bots. Instead, the company seeks to connect agent transactions to the existing financial system, ensuring that compliance, anti‑money‑laundering (AML) checks, licensing, and ledger requirements remain intact.
Multi‑Layered Agent Transactions
One of the most novel aspects of agentic commerce is the way a single assignment can spawn a cascade of payments. An agent tasked with a job may delegate sub‑tasks to other agents, each of which might pay yet more specialized agents for discrete pieces of work. “What begins as one assignment can produce a series of payments underneath it,” Stenseth said, adding that those transactions can run “several layers deep.” This creates reconciliation challenges that do not arise when a payment is treated as a single, atomic event. He illustrated the dilemma: “If something fails farther down the chain, the question becomes whether ‘we roll the whole thing back’ or deal only with the failed portion.” Consequently, merchants and payment processors will need new tools to trace, verify, and, if necessary, unwind complex, multi‑agent payment flows.
Usage‑Based Billing and API Economics
Because agents frequently consume granular services—individual API calls, specific data sets, or compute cycles—usage‑based billing fits the model far better than traditional monthly subscriptions. Stenseth pointed out that developers already familiar with agent‑oriented architectures are accustomed to paying for what they actually use. “His software may also need to pay for each component it consumes. That can favor usage‑based charges for individual API calls, datasets or services rather than a conventional monthly software subscription.” This shift could reshape how SaaS providers price their offerings, moving toward metered, pay‑as‑you‑go structures that align cost with actual consumption by software agents.
Connecting Agents to Traditional Financial Rails
Despite the novelty of agent‑initiated payments, SolvaPay insists that the money must still enter and leave the established financial system. “You can’t skip steps,” Stenseth warned. Agents cannot operate in a lawless payments vacuum; they must satisfy AML regulations, maintain proper ledgers, obtain any required licenses, and adhere to security standards. By anchoring agent transactions to existing rails, the company aims to preserve trust and regulatory compliance while enabling new kinds of commerce. Merchants, therefore, will need to upgrade their fraud and billing systems to recognize legitimate automated activity without blocking it outright.
Future Outlook: Agent‑Driven Payment Choice
Looking ahead, Stenseth envisions a scenario where the agent itself decides which payment instrument to use on behalf of a user. He gave the example of a consumer instructing an agent to “optimize spending for a goal such as earning miles.” The software could then evaluate each transaction and select the credit card that best meets that mileage‑earning objective. “Payment choice could eventually move to the agent as well,” he said. This would add another layer of sophistication to consumer finance, turning the agent into a personalized financial advisor that continuously optimizes payment methods based on programmed preferences. Realizing this vision will depend on continued collaboration between FinTechs like SolvaPay, payment networks, regulators, and merchants to build the necessary controls, standards, and user experiences.
By adapting legacy payment safeguards to accommodate authorized AI agents, SolvaPay is laying the groundwork for a new era of commerce where software not only performs tasks but also handles the money that powers them—while still respecting the rules that keep the financial system safe and trustworthy.
