TheAlert

Executable authority over the actions of AI systems, bound to a real human.

Every action an agent takes goes through a gateway and only happens within a limit that an identified person granted and can withdraw. An intervention order blocks new actions across the integrated environments and produces a signed record of what stopped, what is still pending and what is unknown.

Three commitments: executable authority, measurable intervention, evidence with explicit limits.

Two questions, two artefacts, one boundary.

Hummand identity answers who took part and which act was proven, and issues the humanproof/1 receipt. TheAlert answers which action is allowed, how to restrict it and which result was observed, and uses that receipt to bind every intervention order to a real person.

TheAlert produces neither receipts nor biometric results. Identity does not modify the supervised infrastructure. Hummand is the root of the product's identity, never its surname.

It lives in the infrastructure, outside the model; it does not judge cognition.

It is not a command the model obeys, nor a prompt guardrail. It is not a cognition kill switch, it does not judge whether a decision was good, it does not reach copies outside the perimeter, and it does not replace IAM, gateway, policy and logs, which remain necessary.

Every guarantee names the perimeter where it holds and what lies outside it.

A start document, in discovery.

No capability in the catalogue yet. What exists is the manifesto and the design; what does not exist yet is written down, not hidden.

What discovery looks for: a concrete case, a buyer, a perimeter that can be closed and a pilot someone pays for. Only after that is there a product.

Alerts are signals from the register of acts.

Expired proof, act outside mandate and expected act missing are situations the register exposes on read. TheAlert is something else: supervision of the actions of AI systems. It reads those signals; it does not replace them.