Runway Series

Today for ”Rails Report #4”:

—

When people talk about AI agents, they usually focus on what the agent can do. Can it plan? Can it execute? Can it make decisions? Can it complete a task end to end?

But once agents start interacting with real-world systems, another question becomes just as important: how do we know what, or who, we are trusting? That is where modular verification matters.

ERC- 8004, in simple terms, points toward a more structured way for agents and systems to prove things about themselves. Instead of relying on vague assumptions or one-off integrations, the idea is to create a modular layer where identity, capabilities, permissions, and verification can be checked more cleanly. Why does that matter?

Because agents are only useful if they can operate in environments where other systems are willing to trust them enough to let them act. If an agent is going to move value, access services, trigger workflows, or interact with financial rails, then the system around it needs some way to verify that it is legitimate, authorized, and behaving within bounds.

Today, that kind of trust is often handled manually, or through fragmented point solutions. One platform does one check. Another platform does another. A third layer tries to enforce policy. It works, but it is brittle. Modular agentic verification is interesting because it suggests a more composable future. Instead of forcing every application to reinvent trust from scratch, you can build a shared framework where different components handle different parts of the verification problem.

That could mean proving an agent’s identity. It could mean checking what permissions it has. It could mean validating that it is allowed to perform a specific action. And it could mean making those checks interoperable across systems.

The big picture is simple: as agents become more capable, verification becomes more important. Without verification, agents are just software that can act. With verification, they become software that can act inside systems that care about trust, compliance, and accountability.

So when we talk about ERC-8004 and modular agentic verification, we are really talking about the plumbing that makes autonomous systems usable in the real world. Not just smarter agents, but trustworthy ones.

And once that layer exists, a lot of what comes next in finance, infrastructure, and automated workflows becomes much easier to imagine.

—

The podcasts are authored, edited and produced by Raph Grieco (⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠raphael-grieco.com⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ | ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠olivecapital.vc⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠).