Kindynos research develops the ideas beneath Event Intelligence: consequence chains, endogenous risk, determinism, provenance, value of information and model governance in the presence of generative AI.
Each paper develops concepts that inform Kindynos architecture, engineering and application design.
Examines how the same structural event can produce materially different consequence when identity, state, institutions, contracts and stakeholder positions differ. The paper motivates a world-aware rather than scenario-only approach to analytical systems.
Explores why institutional systems can be internally competent yet collectively fail when information must cross boundaries. It frames integration as preservation of causal, contractual and evidentiary meaning—not merely movement of data.
Develops governance boundaries for systems in which language models assist analysis but deterministic or otherwise governed machinery must remain authoritative for numbers, state transitions and consequential decisions.
Asks which missing fact is worth obtaining before a decision is made. The work links observation cost, uncertainty, model sensitivity and decision consequence rather than treating more data as automatically valuable.
Examines systems in which apparent trust is established before the underlying claim can be independently reproduced. It motivates provenance, forced verification, named refusal and evidence packages as product behaviors.
The program links systemic behavior, event representation and analytical trust.
Endogenous amplification, concentration, dependence, physical-financial feedback and event risk in investment systems.
Event-to-stakeholder consequence, chain operations, value of information and decisions under interacting actors.
Determinism, replay, model identity, refusal, provenance and independent challenge of AI-assisted systems.
The research program defines design requirements that appear in the technology: distinguish zero from absence, carry model identity, value information, preserve chain evidence and make refusal visible.
Kindynos is interested in work that can sharpen a consequential model, validate a world, test a governed path or improve the evidence boundary.