Skip to content

The Studio is an application and practice space for testing concepts through multi-perspectival human–AI iteration. Its methodology supports human-AI reasoning by generating simulations of different critical perspectives, while embedding quality controls that distinguish coherent, usable analysis from otherwise fluent but unsupported claims. It is designed to aid human thinking, and deliberately does not give ‘yes/no’ answers.

The protocol-based scaffolding sits as an infrastructure stack ‘on top’ of LLMs, enabling AI-powered inquiry in a structured environment while retaining overall human interpretation and judgement. The Studio is not a general-purpose AI assistant.

The Studio Design

  • The Studio loads fixed protocols and code before any human input. This ensures methodological consistency and enables audit — the analytical approach is determined by the framework, not by the AI’s default behaviour.
  • There are no pre-determined answers or libraries, and the methodology is not ‘prompt strategy’. The Studio is a sophisticated protocol stack layered over the LLM substrate. The AI model provides generative capacity; The Studio provides rigorous intellectual scaffolding and built-in quality controls.
  • Almost 200 protocol-based agent simulations are used to generate structured multi-perspective analysis. Users can add custom agents. When The Studio calls on a named expertise or tradition, it means an AI system is applying frameworks from published works in that tradition — not that any real person is involved or represented. No attribution to any real person, living or dead, should be inferred. 
  • All outputs carry explicit epistemic status markers — distinguishing what is empirically grounded, what is inference-based, and what requires further validation before use in decisions.
  • Human authority is retained throughout. The Studio generates structured analytical input; final judgement and decisions remain with the user.

Quality Foundations

The Studio’s quality control architecture draws on several frameworks , modified for use with AI. This includes the well-established NUSAP (Numerical Unit Spread Assessment Pedigree), developed by Ravetz and Funtowicz for managing scientific uncertainty in policy-relevant research, re-imagined for use with AI. The Studio also uses proprietary vacancy diagnostics (V-Symbols) for detecting conceptual vacancy in AI-generated outputs. These are not bolt-on features. They are architectural — built into the protocol stack rather than applied after the fact.

Intellectual Property

The underlying analytical processes, protocols, agent orchestration, quality controls, and verification procedures constitute proprietary trade secrets developed by Mick Yates / LeaderValues Ltd, protected under UK Trade Secrets Regulations 2018, the US Defend Trade Secrets Act (18 U.S.C. § 1836), and the EU Trade Secrets Directive (EU 2016/943). Recipients may use findings and recommendations with appropriate attribution but may not reverse-engineer, replicate, or disclose the methodology.

Studio outputs are generated using commercial AI services provided by third-party vendors under subscription arrangements. The Studio’s value lies in the protocol layer, not in the underlying model.

As a matter of good practice, should any individual object to the use of their published work within Studio agent simulations, the objection may be communicated to Mick Yates (mick@mickyates.com) and will be actioned promptly.