six pillars. public. derivable. yours.

A standard is what an institution will let itself be held to. Most "standards" in operator-side AI right now are vibes — they are not adoptable in the technical sense, not derivable in the philosophical sense, and not defendable in the institutional sense. This document is the agentic-u alternative: six named pillars, each derivable from a named failure mode of contemporary operator life, each cross-walked to existing regulatory frames, each measurable against signed receipts. It is what we will let ourselves be held to.

why this standard exists.

Because five pressures are reshaping every operator at once — cognitive abundance, complexity scaling, agency proliferation, boundary dissolution, and accelerating change — and the institutions designed to credential operators were built for a slower world. The MBA was calibrated to a regime of stable complexity; the bootcamp to a regime of stable tooling; the coaching practice to a regime of stable identity. None of those regimes still hold. The standard names what does still hold: the six capacities an operator must demonstrate to remain the locus of decision in the new conditions.

the six pillars, in full.

Each pillar names a capacity. Each capacity is paired to a failure mode — the recognisable way operators fall apart when the capacity is absent. The failure mode is the diagnostic signal; the capacity is the practice. The standard is the union.

1. pillar I

   Sovereignty.

   The operator remains the locus of decision under load.

   When the inbox doubles, when the council disagrees, when the agent stack misbehaves, the operator does not delegate the call to chat completions, to consensus, or to the loudest voice in the room. The operator can still produce, in a paragraph, the reasoning behind the call. The voice is intact under noise.

   failure mode → the Drowning — losing one's own voice in the cognitive flood.

2. pillar II

   Systems.

   Visibility into wholes, not locals.

   The operator can name the couplings — what is upstream, what is downstream, what is holding the system stable, and what is silently drifting. Local fixes are made in service of whole-system coherence, never against it. The map of the system is held in a paragraph, not a diagram.

   failure mode → the Thrashing — fixing locals while the whole drifts.

3. pillar III

   Coordination.

   Multiple agencies acting in concert.

   Humans, agents, copilots, contractors, customers, regulators — multiple agencies acting at once. The operator can name who decided what, what they decided it against, and where the receipt lives. Authorship is attributable, decisions are reviewable, and the institution is held to its own log.

   failure mode → the Babel Collapse — no one can hear who decided what.

4. pillar IV

   Navigation.

   Re-orientation when the map changes.

   When the territory shifts beneath the map, the operator re-orients in less time than the shift consumed. The 90-minute roadmap rewrite. The pivot held in a single paragraph. The plan that is small enough to be amended weekly without ceremony. The territory is checked against the map; the map is amended; the standing is held.

   failure mode → the Disorientation — clinging to a roadmap that no longer matches the ground.

5. pillar V

   Translation.

   Bridging across formerly separate disciplines.

   The operator can translate between the legal regime and the engineering stack, between the compliance memo and the runbook, between the governance frame and the agent prompt. Work compounds across silos because the operator does the bridging work that the institutions on either side cannot.

   failure mode → the Silo Pile-up — work that does not compound because translation is unowned.

6. pillar VI

   Derivation.

   Claims provable from first principles.

   The meta-pillar. Every claim the operator publishes is traceable to something the operator can defend in writing — a receipt, a model, a memo, a citation. The standard earns its standing by being derivable; agentic-u's own claims earn their standing the same way. This is the pillar that holds the others honest.

   failure mode → the Unstanding — claims without a chain back to anything provable.

A pillar is not an aspiration. A pillar is a capacity that, in its absence, produces a recognisable institutional failure. We named six because we found six. We will name a seventh only when an operator's signed receipt proves the seventh is missing.

cross-walks to existing regimes.

The standard is adoptable. It is not a replacement for the regulatory frames the operator is already obliged to. It is the operator-side discipline that pairs with them — the human-side of every AI-governance regime now coming online. Below are the explicit cross-walks the founding council holds itself to.

• NIST AI RMF

  The U.S. AI Risk Management Framework's four operator-side functions — Govern, Map, Measure, Manage — correspond to Agentic-U's Sovereignty (Govern), Systems (Map), Derivation (Measure), and Coordination (Manage). Navigation and Translation cover the gaps the RMF leaves to organisational discretion.

• EU AI Act

  Risk-classification, operator-of-record obligations, and post-market monitoring map cleanly onto Agentic-U's Navigation (continuous re-orientation under regulatory drift) and Translation (bridging the regime to internal engineering and governance frames).

• ISO/IEC 42001

  The AI management system standard is the operator-side companion across all six pillars. Where 42001 defines an AIMS, Agentic-U defines the operator who runs one. The two are paired, not redundant.

• NEXUS

  The standards-for-standards lineage. Agentic-U adopts the meta-standard discipline of NEXUS: every claim made by the institution is itself derivable from the institution's own published reasoning. The Derivation pillar is the inheritance.

• Governable AI

  The paired institution at https://governable-ai.com/. Governable AI is the standard for the system that acts; Agentic-U is the standard for the operator who delegates to it. The two are explicitly siblings: shared body fonts, shared easing curve, shared receipts-not-testimonials commitment.

what derivation means here.

The sixth pillar deserves a paragraph of its own. Derivation is the discipline of being able to trace every claim back to something a thoughtful peer could defend. In the standard, this operates in three layers: (a) every pillar is derivable from a named failure mode in operator life (above); (b) every receipt produced in the initiation is derivable from the standard's pillar against which it is marked; (c) the standard's own amendments are derivable from signed operator receipts published in cohort archives. Derivation is not abstract. It is the audit trail.

what an organisation gets by adopting this standard.

how the standard is amended.

The standard is a working document. It is amended in response to derivations — signed receipts produced by operators in the initiation of initiation, in which an operator names a refinement, exception, or extension that the standard does not yet account for. The standing council reviews proposed derivations quarterly; ratified amendments are published with the signed operator name and the date. The standard does not change because of marketing pressure or institutional fashion. It changes because an operator's signed work has shown it must.

The standard's authority is exactly the sum of the receipts it has been amended by. If the receipts are weak, the standard is weak. If the receipts are strong, the standard earns its standing. The institution does not assert; it lets the receipts assert for it.

what the standard is not.