Sapiom Secures $35M Series A to Power the Next Trillion AI Agents

Sapiom announced on August 18, 2026, that it had completed a $35 million Series A funding round led by Dragonfly, with participation from Accel, Gradient, Coinbase Ventures, and others, and continued support from existing investors Anthropic and Menlo Ventures. The round comes just 11 months after the company's founding and six months after its $15 million seed round, bringing total funding to $50 million.

Sapiom was founded by Ilan Zerbib in 2025 in San Francisco, positioning itself as a platform to help engineering teams 'deliver, run, and scale AI agents.' The company claims to have processed over 270 million transactions, run over 100,000 agents daily, and had a customer reduce inference costs by 75% after migration. These figures come from the company's press release and have not been independently verified.

Sapiom also launched three products: Sapiom Router, which selects the most appropriate model for each call rather than defaulting to the most expensive; Agent Studio, which provides a local environment for engineering teams to build, test, and deploy agents in the context of their existing codebase; and Runtime, which offers managed production infrastructure with access control, routing, recovery, and step-by-step visibility.

Sapiom's core approach is to turn fixed infrastructure choices into real-time decisions at the execution point, selecting the best path for each action based on task requirements, cost, quality, latency, reliability, and company policy. Budgets and permissions are enforced before execution, and every action is metered and recorded in a complete audit trail.

Sapiom's funding announcement cites Gartner's prediction that over 40% of agentic AI projects will be canceled by the end of 2027, with rising costs as the primary reason. This reflects the significant gap between agents in demo and production.

Dragonfly's Haseeb Qureshi stated in the announcement: 'Agents are becoming employees without managers or budgets, and CTOs are effectively playing CFO, allocating real money without seeing where it goes.' He believes this is not a governance problem but an infrastructure problem that needs to be solved at the point where agents act and money flows.

Sapiom's solutions directly address cost and governance: real-time routing selects the optimal model to avoid defaulting to the most expensive option; budget and permission controls set boundaries before execution; and a complete audit trail provides visibility. These features aim to make agents run 'economically, reliably, and controllably' in production.

However, these claims come primarily from the company itself and lack third-party verification. For example, details of the specific customer case with a 75% reduction in inference costs have not been disclosed, and Gartner's prediction is cited only as background. Whether Sapiom can truly solve the problem of agent project cancellations remains to be seen.