Agents file reports, change records and call systems that perform further actions. Their runtime needs to connect those operations to the policy and authority that permit them, with a record of the observed outcome.
Connect policy to execution
Policy documents express what an organization permits. Model and platform controls enforce parts of that policy. The runtime must connect a requested operation to the authority and rules that apply at the point of action, including when the workflow crosses a tool or system boundary.
Some deployments require agents and their oversight to run on infrastructure the organization controls. A portable governance contract should support those constraints and identify how its controls are enforced in the chosen environment.
The supervision has to travel with the agent
The design goal is supervision that follows the agent across model and deployment choices. A hosted model and a model running locally should be governed through the same contract, with the controls implemented and tested in each runtime.
XSI-AIMS specifies governance around the agent’s instructions, memory and permissions. It describes how intent, planning and actions remain connected to authority, including reversal where the underlying system supports it. Implementations map those requirements to their runtime controls; the specification does not retrain the model.
Policy meets the point of action
Select diagram to enlargeWhy we're giving it away
We could have kept it. A governance standard the rest of the industry licenses from you is, on paper, a fine business. We're not doing that, for two reasons. First, a standard one company owns is a product wearing the word. The whole value of a governance spec is that many parties can read it, implement it, argue with it, and trust it without having to trust us. Make it proprietary and it loses the one property that made it worth having.
I believe safety infrastructure should be open to inspection. Publishing the governance contract lets organizations examine its assumptions, implement its requirements and challenge its design without relying on a private description from one vendor.
Labs, universities, standards bodies and product teams are working on different parts of agent oversight. XSI-AIMS contributes a published supervision contract while leaving model training and deployment choices to the implementer. A runtime binding is needed to put that contract into operation.
A word on who's saying this
Extended Systems Intelligence is a small, self-funded company. We are publishing XSI-AIMS openly so other teams can inspect the governance contract, implement it and challenge the decisions behind it.
The ask
So here's the ask. The specification is open. Read it and try to break it. If you run agents somewhere no cloud reaches, see whether it holds. If we got the model wrong, say so in public; that's what an open process is for. We'll be wrong about some of it, and the fastest way to be less wrong is to have the people who do this for a living looking over our shoulder. The agentic era doesn't need our permission to arrive. The layer that governs it still has to be built, and we'd rather build it in the open.
XSI-AIMS is published openly at github.com/Extended-Systems-Intelligence/aims. Read it, fork it, and send a pull request when something breaks.
