Imagine building a custom app without writing a single line of code. That’s the promise of “vibe coding” platforms like Lovable, which turn plain-language descriptions into working prototypes. But what happens when you need to connect that app to real-world services like SMS messaging, email, or payment processing? Suddenly, non-technical creators hit a wall of backend infrastructure headaches. Now, a new wave of financial infrastructure is emerging to solve this exact problem, and it’s poised to transform how AI agents operate in the business world.
The $15 Million Bet on AI’s Financial Layer
Last summer, Ilan Zerbib – former director of engineering for payments at Shopify – launched Sapiom with a bold vision: create the financial layer that allows AI agents to securely purchase and access software, APIs, data, and compute resources. This week, his startup raised $15 million in seed funding led by Accel, with participation from Okta Ventures, Array Ventures, Menlo Ventures, Anthropic, and Coinbase Ventures. The investment signals a growing recognition that for AI agents to become truly autonomous, they need their own payment systems.
“Every API call is a payment. Every time you send a text message, it’s a payment. Every time you spin up a server for AWS, it’s a payment,” said Amit Kumar, a partner at Accel who has met with dozens of startups in the AI payments space. “Right now, there’s no easy way for agents to actually access all of that.”
Beyond Vibe Coding: The Enterprise AI Revolution
While Sapiom initially targets vibe-coding platforms, its implications stretch far beyond helping non-coders build apps. The real opportunity lies in enterprise AI adoption, which is accelerating despite significant challenges. According to the 2026 Connectivity Benchmark Report from Salesforce, MuleSoft, and Deloitte Digital, organizations will spend 19% of their IT budgets on AI agents this year. The average enterprise currently manages 12 AI agents per organization, with projections showing a 67% increase to 20 agents within two years.
But here’s the catch: 82% of IT leaders cite data integration as a major AI challenge, and only 27% of applications are currently connected. This integration gap creates what experts call “AI agent silos” – isolated systems that can’t communicate or collaborate effectively. Financial infrastructure like Sapiom’s could be the missing link that enables seamless multi-agent systems.
The Security Paradox: More Agents, More Risk
As AI agent adoption grows, so do security concerns. A recent breach at Moltbook – a social network for AI agents with over 1.5 million active agents – exposed 1.5 million authentication tokens and 35,000 email addresses. What’s particularly striking is the ratio: only 17,000 human owners control those 1.5 million agents, creating an 88:1 agent-to-human ratio that raises questions about oversight and accountability.
“There is a real risk of AI systems being taken over by malicious actors. This will happen!” warns Mike Wooldridge, Computer Science Professor at the University of Oxford. His concern highlights why financial infrastructure must include robust security protocols. Sapiom’s approach of handling authentication and micro-payments in the background could actually reduce security risks by eliminating the need for developers to manually manage API keys and credentials.
The Competitive Landscape Heats Up
Sapiom isn’t operating in a vacuum. OpenAI recently launched Frontier, an end-to-end platform for enterprises to build and manage AI agents. The platform allows users to manage agents built outside of OpenAI, program them to connect to external data and applications, and limit their access. Enterprises like HP, Oracle, State Farm, and Uber are already customers.
Meanwhile, Salesforce has its own agent-management platform called Agentforce, and companies like LangChain and CrewAI are competing in the same space. Gartner’s December report called agent management platforms “the most valuable real estate in AI,” suggesting that whoever controls the infrastructure layer could dominate the enterprise AI market.
The Integration Imperative
What makes Sapiom’s approach particularly interesting is its focus on the financial layer rather than trying to build another agent-management platform. By specializing in payments and authentication, Sapiom could become the plumbing that connects various AI systems. This aligns with broader industry trends: 94% of IT leaders say AI agents increase development team speed and efficiency, but 96% struggle to use data from across the business for AI initiatives.
The integration challenge isn’t just technical – it’s financial. Traditional payment systems weren’t designed for AI agents making micro-transactions in real-time. Sapiom’s solution handles these transactions seamlessly, charging users through their vibe-coding platforms as pass-through fees. This could dramatically lower the barrier to entry for businesses wanting to deploy AI agents.
Looking Ahead: From B2B to Consumer AI
While Sapiom is currently focused on B2B solutions, Zerbib acknowledges that the technology could eventually empower personal AI agents to handle consumer transactions. The expectation is that individuals will one day trust agents to make independent financial decisions, such as ordering an Uber or shopping on Amazon. However, Zerbib believes that “AI won’t magically make people buy more things,” which is why he’s focusing on creating financial layers for businesses instead.
This pragmatic approach may be what sets Sapiom apart in a market filled with hype. As AI agents become more sophisticated – OpenAI’s new GPT-5.3-Codex runs 25% faster and can handle processes lasting over a day – the need for reliable financial infrastructure becomes more urgent. The companies that solve this problem won’t just be facilitating payments; they’ll be enabling the next generation of autonomous business operations.
The $15 million investment in Sapiom represents more than just confidence in one startup. It signals a broader shift in how investors and businesses are thinking about AI infrastructure. The race isn’t just about building smarter agents – it’s about creating the financial and security frameworks that allow those agents to operate safely and effectively in the real world. As enterprises prepare to nearly double their AI agent deployments in the next two years, the winners will be those who understand that autonomy requires more than intelligence; it requires infrastructure.