Frontier Cognitive Benchmark — Ianudor CNL vs GPT-4o vs Claude Sonnet
Fixed-scale and scaling benchmark results comparing the Constraint Neural Lattice against frontier LLMs on adversarial cognitive tasks.
Three products that compute the exact shape of what your rules allow and forbid.
See our products
Ianura builds constraint geometry engines for the domains where rules are dense, overlapping, and consequential. Given any rule population — security policies, trade controls, AML frameworks — our engine computes the complete geometry of permitted and forbidden space, then reports what matters: the controls doing real work, the gaps attackers can use, the regulatory changes that will move the line.
All three products are powered by the same constraint geometry engine. Given any set of rules, it computes the complete geometry of permitted and forbidden space — then reports what matters: the redundant controls, the keystone rules, the evasion paths, the impact of any change before it is made.
Ianura builds engines that compute the exact shape of what your rules allow and forbid. Three products, one engine, across security, trade, and financial crime.
See products →Security policy geometry. Find redundant controls, keystone policies, and attack paths across NIST, CIS, PCI-DSS, and ISO 27001. 6M+ states analysed per policy set.
Explore Vigilia →Trade compliance and route intelligence. Check any shipment, find evasion paths, pick the best trade agreement. 1.6B trade states analysed in 27 seconds.
Explore Cordon →AML and financial control intelligence. Identify compliance dead weight, keystone rules, hidden transaction paths across OFAC, FinCEN, FCA, AMLD6. 3.6M states per control set.
Explore Remit →A post-Platonic physics ontology where constraint is the sole primitive. Derives fundamental constants from pure topology with zero free parameters.
Explore CGF →CGF applied to navigation and agency. Traction measures when action correlates with outcome. Bearing points toward viability.
Explore VTT →