Sapiom Raises $15M to Give AI Agents the Ability to Pay, Authenticate, and Operate on Their Own

As AI agents grow more capable of planning and executing complex tasks, a new bottleneck has emerged. While models can reason and decide, they still depend on humans to handle payments, API keys, and billing approvals. Sapiom wants to remove that friction.

The startup has raised a $15 to $15.75 million seed round led by Accel to build what it describes as a financial operating system for AI agents. The platform is designed to let autonomous agents securely authenticate, pay for, and audit their own use of software, APIs, data, and cloud infrastructure without manual human setup.

The Problem Agents Cannot Solve Today

Modern AI agents can write code, plan workflows, and orchestrate tasks across tools. What they cannot do is spend money.

If an agent needs to send an SMS, provision a cloud server, or call a paid API, a human still has to step in to create accounts, add credit cards, generate API keys, and manage invoices. That requirement breaks the promise of fully autonomous workflows and turns advanced agents into systems that still rely on constant manual intervention.

Sapiom’s premise is simple. If agents are expected to act, they need controlled economic agency.

How Sapiom Works Behind the Scenes

Sapiom sits between AI agents and the broader API and SaaS economy. Each agent is assigned a managed identity and wallet. When the agent decides it needs a capability, such as messaging or compute, it requests access through Sapiom.

The platform handles vendor authentication, issues short-lived credentials, processes micro-payments, and enforces spending limits and policies. Every transaction is logged with a clear audit trail that ties together the agent, the task, the vendor, and the cost.

Instead of developers wiring billing and authentication separately for every service, Sapiom turns that complexity into shared infrastructure.

From API Calls to Financial Events

In Sapiom’s system, every action becomes a traceable financial event. API requests, compute usage, data pulls, and messaging are all metered in real time.

For finance and security teams, this looks similar to issuing a virtual spend card per agent, with built-in controls and reporting. Usage data can be exported to existing tools like NetSuite, Ramp, or Brex, keeping autonomous agents visible inside traditional finance workflows rather than creating shadow IT.

The roadmap includes deeper integrations with identity providers and procurement systems so agent-driven purchases can follow the same approval and compliance rules as human spending.

Who Is Backing the Vision

The seed round was led by Accel, which sees Sapiom as foundational infrastructure for agent-driven software. Additional investors include Okta Ventures, Gradient Ventures, Menlo Ventures, Anthropic, and Coinbase Ventures, along with a group of strategic angels from companies like Shopify, OpenAI, GitHub, and Stripe-adjacent fintech firms.

The company was founded by Ilan Zerbib, who previously spent five years at Shopify leading payments engineering. His background in high-scale commerce systems directly informs Sapiom’s focus on controlled, auditable transactions at scale.

Why This Matters Now

Agentic AI tools are spreading quickly across software development, operations, and internal enterprise workflows. Teams can already deploy agents that write code, manage deployments, or coordinate campaigns. The missing layer has been financial and identity infrastructure built specifically for non-human actors.

Investors see Sapiom as a response to that gap. By giving agents limited, policy-driven purchasing power, companies can unlock workflows that adapt in real time, such as choosing cheaper APIs mid-task or provisioning resources only when needed.

The broader bet is that agents will soon act as economic participants inside organizations, and that infrastructure must exist before that shift becomes mainstream.

Looking Ahead

Sapiom is currently focused on enterprise and developer platforms where AI agents need to interact with multiple paid services. Over time, the company sees potential expansion into personal agents that manage consumer transactions, once trust models and safeguards mature.

For now, the goal is narrower but critical. Make it possible for AI agents to move from planning to execution without humans acting as financial gatekeepers.

As AI systems shift from answering questions to performing work, Sapiom is positioning itself as the layer that makes autonomous action possible while keeping businesses firmly in control.