Public capability statement
Fionn Labs LLC
Applied research and evidence infrastructure for model-change assurance in safety-critical AI.
Company profile
Public company facts
A focused research organization working at the intersection of learned-model change, assurance evidence, and technical review.
Legal entity
Fionn Labs LLC
Company profile
Independent U.S. research company
Location
Washington State, United States
Contact
reid@fionnlabs.com
Research capabilities
Evidence that can be examined
Fionn Labs develops research practices for examining model change while preserving a clear line between observed evidence, unresolved questions, and decisions held by authorized parties.
Model-change assurance research
Studies how a change to a learned component affects the evidence associated with an established baseline and identifies assurance questions that must be revisited.
Reproducible benchmark research
Uses public benchmark materials to support repeatable research and independent review. Configurations, results, and implementation details remain private.
Evidence discipline
Keeps claims bounded to the model, property, inputs, and conditions examined. Unsupported, timed-out, or inconclusive cases remain visible rather than becoming passes.
Standards-informed translation
Frames research questions and review artifacts using established assurance concepts and the current public record for aviation AI, without asserting compliance or approval.
Public research boundary
Public inputs, private implementation
Public benchmark materials provide non-proprietary context for the research. This site does not publish technical configurations, benchmark verdicts, or the protected method.
Public research posture
- Non-proprietary public benchmark materials
- Clear labeling of current research maturity
- Primary sources linked for public statements
- Independent replication remains in progress
Claim controls
- No benchmark verdicts are published here
- No certification findings are claimed
- No protected method details are disclosed
- No component result becomes a system conclusion
Standards literacy
Research framed by the public record
Standards and agency materials inform vocabulary, research questions, and evidence organization. Familiarity with them is not a claim of conformity, delegated authority, or regulatory status.
Airborne software and systems
DO-178C, DO-330, and ARP4754A provide reference points for software, tool, and system-development assurance concepts.
AI risk management
NIST AI RMF 1.0 provides public risk-management vocabulary for organizing questions, evidence, limitations, and accountability.
Evolving aviation AI work
FAA, EASA, and EUROCAE materials are followed in their stated status. Drafts, proposals, roadmaps, and working-group activity are not presented as settled requirements.
Authority disclaimer
Research evidence is not approval
Fionn Labs is an independent research company. It is not a regulator, certification authority, airworthiness authority, or standards-issuing body. Nothing in this statement constitutes certification, regulatory approval, a conformity determination, or authorization to operate. Those decisions remain with the relevant authorized organizations and public authorities.