Gartner forecasts more than 40% of agentic AI projects will be cancelled by 2027. The stated reasons: unclear ROI, escalating costs, inadequate risk controls. The first two get the attention. The third is the one that decides the other two. A project with adequate controls can prove its ROI, because its outcomes are attributable. A project without them cannot, so ROI stays unclear, so it gets cut. "Inadequate risk controls" is not a failure of project discipline. It is the absence of a layer that makes agent behavior bounded and provable. Origins is that layer. Protocol-level authority, verifiable execution, portable identity. The projects that survive 2027 will not be the ones with the best models. They will be the ones whose controls made the ROI legible.