Our Focus

Structured markers
Encodable markers that represent context, internal state, and meaningful change. These markers evolve over time under clear rules, rather than being guessed or inferred.

Validation infrastructure
A measurement layer that updates marker values and states over time. It tracks evidence, stability, and regression using normalization, smoothing, and timing rules.

Decision checkpoints
A gate layer that determines whether the system is allowed to act at a given moment. It evaluates policy-defined conditions against the current measured state and either permits or blocks a transition.

Role and mode control
A protocol for changing system behavior, where transitions are explicitly proposed, approved, or reversed rather than drifting implicitly.

Continuity primitives
Foundational mechanisms that support time-aware interaction across sessions, including selective persistence, boundary enforcement, and intentional closure.

Applied prototypes
Early-stage prototypes and technical specifications exploring long-duration, context-sensitive workflows such as reflection, journaling, and collaborative reasoning.

Together, these elements form a decision spine that allows AI systems to measure their internal state, govern when action is permitted, record why decisions were made, and maintain continuity over time without losing clarity or control. 

When AI systems move beyond single prompts and short interactions, accountability can no longer be limited to inputs and outputs alone. Long-lived systems introduce new responsibilities. They must decide when to act, when to hold back, how to adapt over time, and how to remain understandable as context accumulates.

Without internal structure, continuity becomes unreliable. Systems drift. Adaptation becomes opaque. Confidence is expressed even when it is unwarranted. Over time, trust erodes not because of malicious intent, but because the system lacks the means to govern itself across time.

This becomes critical in domains where context and duration matter.

In research and knowledge work, continuity determines whether an AI can support long-term inquiry without distorting context or weakening methodological integrity.

In education and instruction, systems must respect learning trajectories, role shifts, and developmental boundaries rather than operate as static tools.

In reflective and journaling workflows, continuity shapes whether memory, insight, and closure are handled intentionally, preserving agency and healthy boundaries over time.

In these settings, continuity isn’t optional. Without a structure that tracks state over time, governs when action is permitted, and records why decisions occur, systems lose accountability. That structure has to persist — not just in memory, but in responsibility over time. 

Liminal Research focuses on building that structure at the foundation level, so AI systems can adapt over time while remaining transparent, restrained, and accountable. 

Liminal Research is in an active research and development phase focused on establishing a rigorous foundation for governed AI decision systems.

The core architecture is defined and has been worked through end to end, including decision flow, state progression, transition logic, and auditability. 

Two patent applications have been filed, covering symbolic markers, decision gating, evidence recording, and continuity mechanisms. 

Early prototypes and reference implementations have been developed to validate key ideas and system behavior. These provide a concrete basis for ongoing testing, refinement, and applied research.

The project is now positioned to move into pilot testing and grant-supported research with aligned partners. Current work is focused on refinement, documentation, and preparing the system for limited real-world evaluation in controlled contexts.

We welcome collaboration with:

Individuals or research teams exploring AI ethics, symbolic systems, continuity, or applied cognition

Clinicians and practitioners interested in journaling frameworks and state continuity 

Funders supporting next generation explainability, AI trust, or mental health innovation

To request a research summary, framework overview, or discuss collaboration:

liminalresearchllc@gmail.com