NextFin News - In a move that signals the maturation of the autonomous agent economy, San Francisco-based startup Sapiom announced on February 5, 2026, that it has raised $15 million in seed funding to enable AI agents to independently purchase and manage their own technology tools. The investment round was led by Accel, with significant participation from a coalition of strategic heavyweights including Okta Ventures, Array Ventures, Menlo Ventures, Anthropic, and Coinbase Ventures. According to TechCrunch, the capital will be used to build the foundational financial infrastructure required for AI systems to navigate the complex world of B2B procurement without human intervention.
Founded by Ilan Zerbib, the former Director of Engineering for Payments at Shopify, Sapiom addresses a fundamental friction point in the current AI landscape: the "manual glue" required to connect intelligent agents to the services they need to function. While modern AI can write code and orchestrate workflows, it remains tethered to human-managed credit cards, API keys, and manual account setups. Sapiom’s platform acts as a financial and identity intermediary, providing AI agents with secure "wallets" and ephemeral credentials to execute micro-transactions for services like Twilio SMS, AWS compute, or specialized data feeds. This allows developers, particularly those using "vibe coding" platforms like Lovable, to move from prototype to production without the administrative burden of traditional software procurement.
The emergence of Sapiom reflects a broader shift in the unit economics of software. As Amit Kumar, a partner at Accel, noted, every API call is essentially a payment transaction. In a world where agents may perform thousands of tasks per hour, the traditional model of monthly SaaS subscriptions and human-in-the-loop approvals becomes a bottleneck. Sapiom’s infrastructure enables a transition toward metered, machine-to-machine commerce. By integrating with identity providers like Okta and payment rails like Coinbase, the platform ensures that these autonomous purchases are not only seamless but also compliant with enterprise security standards, featuring granular spend limits and immutable audit trails.
From an analytical perspective, Sapiom is tackling the "last mile" problem of AI autonomy. The current enterprise software stack was designed for humans; it assumes a user with a browser, a password, and a corporate credit card. AI agents, however, operate at a scale and speed that breaks these legacy systems. By creating a dedicated financial layer, Sapiom is effectively building the "Stripe for Agents." This is particularly critical for the rise of agentic workflows, where a single task—such as processing a customer refund—might require an agent to interact with four or five different paid services. Without an automated payment layer, the operational overhead of managing these micro-costs would outweigh the efficiency gains of the AI itself.
The strategic involvement of Anthropic and Coinbase Ventures suggests that the industry views this as more than just a fintech play; it is a core infrastructure requirement for the next generation of Large Language Models (LLMs). For AI labs like Anthropic, the ability for their models to "reach out" and buy the tools they need increases the utility of the model. For Coinbase, it represents a massive new use case for digital payment rails that can handle high-frequency, low-value transactions more efficiently than traditional banking systems. Data from Okta’s "Businesses at Work" report indicates that large enterprises now manage over 200 apps on average; Sapiom’s role is to turn this software sprawl into a programmable resource for autonomous systems.
Looking forward, the success of Sapiom will likely trigger a wave of "Agent-Native" service providers. Just as the cloud era forced software companies to move from perpetual licenses to subscriptions, the Agentic era will force a move toward hyper-granular, pay-per-use models. We can expect to see a new category of "Agent-Gated" APIs that are optimized for machine discovery and instant procurement. While Zerbib remains focused on the B2B sector, the long-term implications extend to consumer AI, where personal assistants could eventually manage household budgets and service subscriptions. However, the immediate impact will be felt in the enterprise, where Sapiom’s rails will allow companies to scale their AI workforces without scaling their procurement departments.
Explore more exclusive insights at nextfin.ai.
