Sean Chihwendu

Enterprise governance, delegated authority, and approval frameworks are mature, well-understood disciplines inside large organizations. None of them were built with a mechanism for enforcing themselves against a system that can act on its own. As autonomous AI moves from producing recommendations to directly executing business actions, that absence stops being a theoretical gap and becomes an operating risk.

Sean's premise is that this is not a missing feature inside existing AI governance or AI security tooling. It's a distinct problem: keeping delegated authority intact and enforceable as work moves from a person to an AI agent to the systems that agent touches. He sees a loose parallel to how identity and access management emerged as enterprises needed to authenticate machines as well as people, though he treats that as a hypothesis worth testing, not a settled conclusion. Authority continuity is his name for that problem.

Sean's view is that this isn't fundamentally an AI problem, it's an organizational design problem. As AI moves from assisting people, to executing tasks on its own, to functioning as part of the workforce itself, organizations need a way to preserve the authority, decision rights, and accountability they already built for people, now that some of that workforce is autonomous. Runtime Authority doesn't decide who is authorized. It evaluates the authority an organization's operating model already grants, action by action.

Sean's path to this problem started in procurement and tender management, not software. He spent years watching a specific, recurring failure: an organization approves a payment or a purchase under a clearly written policy, and still has no reliable way to confirm that what actually happened matched what was authorized. The policy existed. Nothing evaluated it at the moment that mattered, and nothing proved afterward whether it had actually been followed.

In 2024, he began working on the problem directly, validating it with procurement and audit professionals while teaching himself to build the technical solution. He met his future co-founder, Nathan Obiekwe, after posting about an early fraud-detection tool; what began as a remote collaboration became AI Securewatch.

The company's first working system targeted payment execution specifically: the environment where an unauthorized action has the most immediate, measurable cost, and therefore the fastest environment in which to prove the model. Payments were the proving ground, not the destination.