Programmes
The founding curriculum, the five professional levels and the capstone standard, as proposed. No programme is open for enrolment, no dates or fees are set, and no credential is currently awarded.
Founding curriculum · in development
Three founding programmes
Ten programme territories were identified as credible. Three were selected as the founding set because they are the most defensibly unoccupied: employers are hiring for the roles, the competency set is definable, and no existing degree covers them. One has been specified at version 1.0; two are proposed. None is open for enrolment.
Mandate Engineering — Foundation
Writing the machine-readable mandate: what an autonomous system may do, on whose behalf, with what budget, inside which limits, emitting which evidence, and under what conditions its authority ends. Includes delegation topology across humans, agents and sub-agents, and the competence a legally mandated human overseer must hold.
Verification & Evidence Architecture
Designing the evidence trail that an auditor, insurer or regulator will read, and the verification regime for output the verifier cannot fully inspect: rubrics, thresholds, sampling and audit design, escalation tiers, safety-case argument, and the residual human function at the apex of an oversight tree.
Anchors: Evidence Architect, Agent Evaluations Engineer, AI Auditor, Safety-Case Author, model-risk validator. Home school: Verification & Evidence.
Proposed · In developmentSynthetic Enterprise Architecture
Designing organisations composed of human roles, persistent agent roles and policies: delegation portfolios, intervention budgets, fleet economics, correlated-failure planning and the governance structure such an organisation needs before its first agent runs.
Anchors: Synthetic Enterprise Architect, Agent Fleet Manager, Human–AI Team Designer. Home school: Synthetic Enterprise & Organisation.
Proposed · In developmentFurther programme areas
Identified as credible, sequenced after the founding three: Agent Identity & Authority · Agent Governance & Oversight · Model Risk for Agentic Systems · AI Liability & Underwriting · Autonomous Workflow Engineering · Synthetic Labour Economics · Machine-to-Machine Commerce & Trust.
Five professional levels
The proposed formation ladder. Each level is defined by what its holder can build, operate or govern, not by hours attended. Durations are conceptual. These are proposed architectures, not enrolled programmes.
Foundation
4–8 weeks, conceptually
Operate a single agent under a stated mandate and produce logged supervision evidence. Intended to map onto the competence a human overseer must hold under EU AI Act Article 14.
Operator
About 3 months, conceptually
Operate multi-agent workflows under budget, escalation and revocation controls, and produce a risk-management report in the measure-and-manage form used by the NIST AI Risk Management Framework.
Architect
About 6–9 months part-time, conceptually
Design and deploy an agentic system with signed policy, identity and a verifiable evidence pipeline, for a named third-party principal.
Governor
About 9–12 months, conceptually
Govern a portfolio of autonomous systems belonging to others: audit, adjudicate, revoke; sign safety cases; testify.
Fellow
Research and standards contribution
Contribute to the mandate, verification or delegation discipline itself: standards drafting, novel methods, published evidence, teaching.
Build · Operate · Govern
Not know, explain, recall.
The Institute's pedagogical principle is that a graduate is someone who has operated a mandated agentic system under budget, risk and verification constraints and produced auditable evidence of it. The eventual credential is meant to attest demonstrated authority and performance, not classroom attendance. As learners rise through the levels, the authority they carry — budget, revocation power, signing capacity — is meant to rise with them.
What a graduate would leave with
- A signed capstone dossier: mandate, evidence trail, independent scoring, revocation log.
- Machine-readable attestations of skills demonstrated, verifiable by prospective employers.
- Placement in an alumni verifier network from which graduates may be called to serve as verifiers or independent assessors.
- Publication-quality evidence of a real autonomous system they mandated, operated or governed.
This describes the intended design. No graduate exists yet.
Real mandate. Real money. Real verification.
The proposed capstone standard for every level above Foundation. Not an essay: an autonomous system running under enforceable conditions.
Presented here as the proposed educational standard. No capstone has yet been run.
- A written mandate
- Scope, principal, prohibited actions, reporting frequency, revocation triggers.
- A defined principal
- A named party on whose behalf the system acts, with the authority to have delegated it.
- A budget
- Financial, tokens, compute, time — actual, not notional.
- A risk boundary
- Maximum loss, maximum harm, blast-radius controls.
- Verification requirements
- Evidence to be emitted; independent scoring by someone other than the operator.
- A revocation authority
- Someone who can stop it, on stated criteria, and a log showing that they could.
The ARENOMICS Arena (arenomics.com) already runs autonomous agents under identical conditions with an integrity ledger and verified simulation state. It is the natural technical substrate for capstones of this kind. Adapting it for that purpose is future work.