SolvaPay Builds Payments for AI Agents
FinTech firm SolvaPay is developing payment infrastructure for AI agents, addressing challenges like machine-readable payments and usage-based billing as
A Stockholm-based FinTech company is building payment infrastructure for artificial intelligence agents. SolvaPay's work tackles the complications that arise when bots are authorized to make purchases, handling systems originally designed to verify human customers.
SolvaPay CEO and Co-Founder Viggo Stenseth explained that payments represent a key friction point. AI's growing autonomy in completing tasks collides with legacy systems built around human assumptions. Current infrastructure uses tools like 3DS and CVV codes to block unauthorized bots. Merchants now face the dual challenge of maintaining these fraud protections while enabling legitimate AI agents to spend. They must distinguish between authorized software and malicious programs, all while controlling the agent's spending limits and permissible purchases.
Stenseth noted that businesses have had little time to prepare for this shift. He compared it to the slow adoption of websites in the 1990s and the lag between smartphone invention and related payment experiences. In contrast, agentic AI is developing rapidly. "The timer already started a couple of years ago," Stenseth said. Agents can now act autonomously, but "the billing, the payment, the flow hasn't really been solved yet." SolvaPay aims to bridge the gap between fast-moving agent software and the deliberately paced financial infrastructure governed by controls and regulation.
How Agent Transactions Differ
Payments initiated by AI agents will not mirror human transactions. An agent given a single assignment can delegate tasks to other agents, generating a cascade of payments. One agent may pay another, which then pays several specialized agents for components of the work. Stenseth said these transactions can run several layers deep. This creates complex reconciliation issues absent when a payment is a single event. The industry has built systems for reconciling one transaction. Now, if a failure occurs deep in the chain, providers must decide whether to roll back the entire sequence or only address the failed portion.
The Rise of Usage-Based Billing
Agents can function as both buyers and sellers. Stenseth cited an example of a developer who built an agent for financial analysis. The software autonomously communicated with data providers and paid for the information it needed. The developer could then resell his enhanced analysis tool to others. This model shifts billing preferences. Developers building with agents are already more accustomed to usage-based charges for individual API calls or datasets, rather than flat monthly software subscriptions. The money, however, must still flow through the established financial system, complying with all existing anti-money laundering rules, licensing, and security protocols. You can't skip steps, Stenseth emphasized.
Implications for Merchant Systems
For merchants, accommodating AI agents requires more than just accepting a software-initiated payment. Fraud detection systems must learn to recognize legitimate automated activity. Billing platforms may need to handle micro-payments for individual units of data or computing services. Reconciliation must track the web of payments generated when agents subdivide tasks. Looking ahead, payment choice could also be delegated. A consumer might instruct an agent to optimize for airline miles, and the software would select the best card for each purchase to meet that goal. The integration of autonomous software is applying new pressure on security and operational controls across the payments landscape.
