Engagement
Advisory and system-level work on evaluative AI architectures, execution-layer control, and deployment of next-generation systems in high-stakes environments.
Mandate
The practice is centred on building the next layer of AI: systems that operate on evaluation itself, not just optimise within fixed objectives.
This work targets a structural limitation in current machine learning systems. Today’s architectures cannot revise the standards that define success, failure, or admissible action. The focus here is on designing and developing systems in which evaluation becomes an object of system-level operation.
In parallel, I advise boards, senior counsel, and technical leadership on the implications of these systems as they move into deployment. This includes questions of authority, control, and responsibility where decision-making is embedded in infrastructure and cannot be meaningfully overridden.
Typical matters include:
- design and development of evaluative AI architectures beyond fixed-objective optimisation;
- structural treatment of out-of-distribution failure and objective misspecification;
- allocation and traceability of responsibility in AI-mediated systems;
- locus of evaluative control (who defines and can change system objectives);
- liability exposure and defensibility, including Section 230 and related doctrines;
- architectures for auditability, constraint, and execution-layer control;
- governance of systems operating under distributional shift or epistemic rupture.
The work is architectural first. Advisory follows from system design, not the reverse.
Forms of Engagement
Engagements span system design, strategic advisory, and high-level technical or institutional briefings. Work is typically at founder, board, or senior technical level.
- system architecture design and technical direction for next-generation AI;
- strategic advisory on AI systems where evaluation and control are central;
- independent expert opinion, including litigation and regulatory matters;
- confidential briefings for boards, executives, and technical leadership;
- keynote lectures and invited talks on AI, agency, and epistemic systems;
- research collaborations and institutional partnerships.
Engagement is selective and focused on problems with structural or long-horizon implications.
Relationship to Scholarship
The work is grounded in an ongoing research programme on epistemic generativity, evaluative systems, and the architectural limits of current AI.
This includes the formalisation of the threshold at which systems move from optimisation within fixed standards to the capacity to revise those standards under conditions of failure or rupture.
The result is a framework that is both technically actionable and institutionally legible—engaging directly with system design while remaining usable in legal, regulatory, and governance contexts.
Enquiries
For engagement enquiries, including system design, advisory, expert opinion, or speaking invitations, please use the contact page.