Products
Five products. One substrate. One evidence model.
Products developed in, and surfaced from, one QTT-native, physics-native, GPU-native computational substrate: verifiable evidence, quantum-ready cryptography, thermal-fluid simulation, engineering design optimization, and the AI inference data plane.
Availability
Turns important digital artifacts into verifiable evidence: signed receipts, evidence bundles, and offline verification. Two surfaces, one verification standard: a hosted Cloud API, or a sealed appliance where the customer holds the keys.
Quantum-ready cryptographic lifecycle: free cryptographic inventory and CBOM, signed migration quotes, migration execution, runtime protection, and continuous assurance.
GPU-native thermal-fluid simulation and evidence engine for AI factories and high-density datacenters: decision-grade thermal outputs with signed, offline-verifiable evidence artifacts per run.
QTT-accelerated multidisciplinary design, analysis, and optimization toolkit for high-dimensional engineering design campaigns.
Structural Data Platform: the AI inference data plane as one platform. A single signed runtime replaces the eight to thirteen services of a typical inference stack on customer GPU or CPU infrastructure.
Where to start
Two products have a front door that costs nothing and requires no entitlement.
HX-AIFactoryTwin, HX-MDAO-TP, and HX-SDP enter through a scoped pilot. The engagement starts with a qualification call on workload, deployment environment, and constraints, and ends in a validation report the customer team can run themselves.
One evidence model
Every product ends in an artifact someone outside HolonomiX can check.
The products differ in what they compute. They do not differ in whether the result can be verified. Each one signs its output, and verification runs without calling HolonomiX.
ML-DSA-65 signatures, canonical receipt schema, and a bundle an auditor verifies offline with the issuer public key.
A CycloneDX inventory at the front, deterministic signed quotes, ledgered migration and runtime events, and signed posture snapshots at the back.
Checksum-addressed and Ed25519-signed, carrying config, validation, certificate, quantities of interest, and the benchmark card. A failed gate blocks attestation.
Formal certificates where invariants are provable, measured verification where they are not, and the two are never conflated.
The exact-receipt schema verifies under the HX-Provenance verifiers, which is the cross-product integration path.
Where each product runs
Marketplace listing, customer environment, or sealed appliance.
Private Appliance as AMI, VMware OVA, KVM, Proxmox and OpenStack QCOW2, or an OCI runtime kit for Kubernetes. Customer holds the signing keys.
Cloud, Self-Hosted, and Air-Gapped editions. Every surface is available on every edition; the buyer chooses the deployment boundary.
CUDA container on NVIDIA GPUs, or Helm against the NVIDIA GPU Operator with GPU resource limits and probes.
Private offer and pilot for design campaigns, including government channels.
A single signed runtime on customer GPU or CPU infrastructure, single-node or distributed.