AxoDen compositional safety
Work on deterministic controls, admissibility checks, replayable evidence and bounded trust for AI-enabled systems, grounded in topology, graph theory, information physics and formal methods.
Research that supports product judgement.
Research and publications
Luminesce uses research as a working discipline: to define concepts clearly, test assurance claims, and decide which AxoDen capabilities deserve commercial pilot work.
20DOI-registered publications on Zenodo
29theorem-level results in the current register
24results currently Established with proof or empirical evidence
ORCID0009-0008-6435-3530
Research themes
The public research record should support commercial clarity. Each theme below maps to a service or product direction on the Luminesce site.
Evidence architecture
Research into how incomplete, contradictory or noisy evidence can be structured for audit, investigation and assurance.
APASES reasoning discipline
A governed human-AI reasoning architecture for traceable analysis, structured retrieval, decision decomposition and cross-domain synthesis.
RF-AI boundary
Assurance patterns for systems that convert radio-frequency observations into AI-mediated decisions.
Forensic intelligence
Evidence graphs, contradiction handling and replayable investigation records for cyber and technical incident review.
Commercial option synthesis
Structured assessment of where a research engine fits real buyer pain, procurement maturity and pilot feasibility.
How the research is used
The research record is a source of method, vocabulary and evidence. It does not replace buyer validation, operational testing or procurement due diligence.
For commercial work, Luminesce treats research claims as inputs to be checked against system context. A theorem, paper or benchmark can support a pilot hypothesis. It still has to be translated into a scoped product question, a dataset, an acceptance test and a decision about what happens if the evidence is weak.
Research-to-service map
This is the page structure for buyers who want to understand what the research becomes in practice.
| Research area | Commercial expression | Typical output |
|---|---|---|
| AxoDen Kernel | AI assurance and bounded trust review | Control map, evidence schema, refusal logic and pilot plan |
| APASES reasoning discipline | Executive reasoning and decision-decomposition engagement | Structured question map, evidence plan, assumptions register and decision brief |
| Structural gap inference | Automated pre-audit and compliance diagnostics for supplier dossiers | Missing-evidence report, contradiction register, supplier-risk briefing and remediation sequence |
| Forensic graphs | Cyber incident and root-cause support | Evidence graph, contradiction log and replayable investigation pack |
| AxoDen Defence | Refusal-capable autonomy and spectrum-derived AI decision review | Threat model, admission criteria, refusal-cascade test harness |
| Grid Flight Recorder | Cyber-physical resilience and infrastructure event review | Evidence model, replay timeline, uncertainty register and control recommendations |
| Enterprise transformation evidence | AI Assurance Readiness Review | Current-state evidence map, governance gaps, readiness decision and control recommendations |
| Commercial option synthesis | Challenge response or executive advisory | Option map, assumptions register, buyer-risk analysis and recommended next step |