The Kennedy Idea Propulsion Laboratory investigates human-AI interaction challenges when standard playbooks fail. Research focuses on developing frameworks for complex problem diagnosis, investigative methodology, and collaborative intelligence systems.
FEATURED INVESTIGATION
Human-in-the-Loop (HITL) Interaction Design
Investigating how humans and AI systems can work together to amplify insight rather than automate judgment. The research challenges the prevailing “AI as automation” paradigm by developing frameworks for AI as forensic instrument: tools that probe systems, surface contradictions, and accelerate the Interpretation through Questioning loop.
Core premise: AI augments human bandwidth for complex problem diagnosis. It doesn’t replace expert judgment; it frees experts to focus on work only humans can do.
Applications: Diagnostic investigations in legal systems, organizational transformation, policy design, and any domain where problems exhibit wickedness (competing stakeholder realities, no clear stopping rule, high cost of being wrong).
Recent publication: The Holmesian Leap: Solving the “Wicked Problem” of Access to Justice with AI – National Association of Women Judges (October 2025)
ACTIVE RESEARCH PROJECTS
The Operational Protocol Method: Systematic LLM Specialization
Developing systematic methods for creating specialized AI collaborators through persona engineering and agent coordination. The Operational Protocol Method treats AI as a sculptural medium: you shape the interaction model to match specific investigative needs rather than accepting generic chatbot defaults.
Research question: How do we move from ad-hoc prompting to repeatable, documented protocols for human-AI collaboration?
White Paper: The Operational Protocol Method: Systematic LLM Specialization Through Collaborative Persona Engineering and Agent Coordination (SSRN, 2025)
Status: Operational protocols in active use; methodology documentation published.
The Guided Advisory Layer (G-A-L Method)
Investigating architectural approaches for structuring AI interactions to preserve human decision-making authority while leveraging AI’s pattern-matching capabilities. The G-A-L Method creates explicit separation between AI-generated options (guidance) and human-selected actions (advisory).
Research question: How do we design AI systems that expand human agency rather than constrain it?
White Paper: The Guided Advisory Layer for Large Language Models (SSRN, 2023)
Status: Framework published; applications under development across multiple domains.
The Holmes Canon Project
Examining Sherlock Holmes stories as a foundational text for investigative methodology in complex, ambiguous situations. The project treats the Canon as a practical guide to remaining in uncertainty long enough to see systems clearly, distinguishing between what appears to be true and what is actually true.
Research question: What investigative principles from 19th-century detective fiction remain essential for 21st-century problem diagnosis?
Status: Literary analysis in progress; applications to contemporary Wicked Problems under development.
Connection to current work: The NeoHolmesian Method draws directly from this research, applying Holmes’s investigative approach to AI-era challenges.
Adjacent Transformation
Examining how organizations can transform by leveraging capabilities from adjacent industries rather than incremental improvements within traditional models. Focus on identifying structural patterns that enable cross-domain innovation.
Research question: When should organizations look outside their industry for transformation models rather than optimizing existing practices?
Status: Pattern identification and case study analysis.
Megatrends and System-Level Change
Investigating how demographic shifts, technological convergence, and economic restructuring create new categories of Wicked Problems that resist conventional analysis and intervention.
Research question: How do we diagnose problems that emerge from the intersection of multiple megatrends rather than single-cause failures?
Status: Framework development for multi-trend analysis.
RESEARCH OUTPUTS
White Papers (SSRN)
- The Operational Protocol Method: Systematic LLM Specialization Through Collaborative Persona Engineering and Agent Coordination (2025)
- The Guided Advisory Layer for Large Language Models (2023)
- View all publications on SSRN
LegalTech Hub Column
Monthly column – Applications of research to legal technology and innovation
The Kennedy-Mighell Report Podcast
400+ Episodes – Real-time dialogue exploring technology, innovation, and investigative methodology
Books
Successful Innovation Outcomes in Law – Free PDF under Creative Commons
The Lawyer’s Guide to Collaboration Tools and Technologies
OPEN SOURCE PROTOCOLS
Research protocols and frameworks developed at KIPL are documented and released under Creative Commons Attribution License (CC BY), reflecting a commitment to accessible methodology over proprietary knowledge.
ENGAGE WITH THE RESEARCH
For speaking inquiries on Human-in-the-Loop methodology, the NeoHolmesian Method, or to propose a diagnostic investigation related to current research areas, visit Engagements or submit an inquiry here.