Measuring the Behavioral Context of AI Systems at Runtime
Toward a common metrological layer for heterogeneous AI systems
NeoMundi develops and tests a behavioral metrology framework designed to measure the state, stability, and variations of AI systems under defined conditions.
This independent, timestamped, and comparable measurement provides the context required to interpret observations, detections, audits, diagnostics, comparisons, insurance assessments, and evidence.
Independent Observatory · Open Protocols · Longitudinal Data · Applied Research
One measurement. Multiple uses. Multiple infrastructures.
NeoMundi Research experimentally studies the scope and limits of this hypothesis.
270, 907 observations · 18 systems · 7,320 measurements per week
The AI Oservatory’s latest publications
- NeoMundi Invited to Present AI Weather to Students at Champlain College Saint-Lambert
- AI Barometer 12, August 31 to September 6, 2026
- 11th AI Barometer – August 24 to 30, 2026
- What a Stable Average Can Conceal in Production AI
- 10th AI Barometer – August 16 to 23, 2026
The AI Weather: the daily behavioral bulletin for leading AI models
AI systems change. NeoMundi makes those changes visible, verifiable, and consumable by your infrastructure.
Daily measurement • Machine-readable signal • Timestamped trace • Independent verification
OPEN SCIENCE
NEOMUNDI INSTRUMENT / CONTROLTOWER
→ Real-time AI behavior measurement and monitoring instrument
INDUSTRIAL SPECIFICATIONS
→ NeoMundi GitHub repositories
SCIENTIFIC PROGRAM
→ Observatory GitHub repository
PILOTS & USE CASES
→ Documented cases of infrastructures using NeoMundi signals
SCIENTIFIC & ADVISORY COMMITTEE
→ Discover the members