AI agents are now making decisions with people's money. Meanwhile banks, insurers, fintechs, and other regulated businesses are developing and deploying agents across processes that include approving transactions, moving funds, and handling sensitive customer data.
However, agents are not immune from making mistakes.
When an agent sends a payment to the wrong account or exposes customer records, someone has to absorb the loss. Most enterprises today have no reliable way to measure that exposure and almost no way to transfer it. That's the problem Harbor is building to solve.
We believe agentic insurance will be one of the critical pieces that makes the agent economy work.
Insurance has always helped businesses take risks they couldn't comfortably carry alone. As more work moves from people to agents, demand for coverage will follow. But nobody can price a risk they don't understand.
Measure, Mitigate, Insure
Agent risk depends on far more than which model a company uses. It depends on what the agent is permitted to do and which systems it can reach. Two companies running the same model can carry very different exposures. Taken in aggregate, this means that every company’s experience with agents, and their risks, is slightly different. As such, the underwriting process for that risk is also bespoke.
Harbor starts by building that understanding, building software that tests agents for vulnerabilities and monitors how they behave in production, to give enterprises a way to find and fix problems before they turn into losses. The same telemetry gives Harbor something other carriers don't have: real data on how agent behavior functions within enterprises and translates into financial loss. That data becomes the foundation for underwriting the risk that remains.
Financial services is a natural place to start: banks have clear reasons to care about the cost of a security or compliance failure, and clear upside from automation – if they can effectively manage and mitigate the downside risks.
The Right Team for an Uncharted Market
Building this company requires two things that rarely sit in the same room: frontier AI research and the relationship-driven work of insurance. Customers need tools that actually reduce their risk. Carriers and reinsurers need a credible basis for taking that risk on. Cristian Bartolomé, Daniel Elechiguerra, Luis Botín, and Jesus Santos are building both from day one. We heard great things about this team from one of our existing portfolio company founders, and are floored by their pace of execution since we started working with the Harbor team.
One thing that stands out to us is how they recruit. Harbor's research team includes international olympiad medalists in math, informatics, and physics, among them Spain's most decorated math olympiad medalist. They have paired that technical depth with real standing in the insurance world as well.
They are also disciplined about the order of operations. Harbor could have chased small, quick contracts. Instead, the founders have focused on the large enterprises where agent risk matters most, and on understanding how losses actually happen before rushing to write policies. As they see it, there is no value in being early to a market whose loss behavior you don't understand.
As agents become economic actors, someone will need to measure, price, and carry the risk they create. We think this team is the one to do it.
We're excited to partner with Cristian, Daniel, Luis, Jesus, and the Harbor team as they make agent risk measurable and insurable.
© 2026 Restive®, Inc.
