The inference API for physical AI.
Connect your systems to Arthedain — edge-deployable,
fault-immune, connectivity-denied.
Arthedain is Enotrium's edge inference engine — designed to run on forward-deployed hardware where connectivity is unavailable, unreliable, or unacceptable. The Enotrium API brings that engine to developers building autonomous systems, manufacturing intelligence, drone payloads, and sovereign edge infrastructure.
This API is an interface to physical AI — inference that closes the sense–act loop on-device, adapts in real time to sensor drift and hardware fault, and operates within the power envelope of embedded hardware.
What Is the Enotrium API
The Enotrium API is a programmatic interface to three capabilities that do not exist in any commercial cloud AI platform:
First:Arthedain inference — a proprietary inference architecture that runs fully on-device, adapts continuously to sensor drift and environmental change on-device, and maintains classification integrity under hardware fault conditions that would degrade or crash conventional inference engines.
Second:Icarus hyperspectral sensing pipelines — multispectral and hyperspectral imaging fused with high-density LiDAR for contaminant detection, material characterization, and anomaly detection, running on edge hardware deployed on UAVs, ground vehicles, and fixed sensor arrays.
Third:Cryptographic supply chain provenance — zero-knowledge proofs that allow downstream operators and procurement officers to verify domestic-content compliance without exposing underlying supply chain data.
These three capabilities are available today as early-access API endpoints for vetted partners. General availability follows technology-transition milestones.
Who This API Is Built For
Drone and autonomous systems manufacturersintegrating perception and adaptive reasoning into UAV payloads for operation in comms-denied and GNSS-degraded environments. Arthedain handles sensor fusion and distribution shift recovery on-device — no ground station, no retraining cycle, no data leaving the aircraft.
Defense primes and system integrators building autonomous platforms that must satisfy domestic-content requirements from materials through software. The provenance API provides compliance verification at the line-item level.
Industrial operatorsdeploying adaptive process control at manufacturing edge nodes — biorefinery throughput optimization, fiber grading, contaminant screening — where latency-deterministic inference is required and connectivity is intermittent. Arthedain's fixed memory footprint and physical AI profile make it deployable on constrained industrial hardware.
Allied-nation governments and sovereign infrastructure programsrequiring AI inference that stays on sovereign hardware. Enotrium's edge deployment model keeps inference, adaptation, and data local by architecture — not by policy.
Arthedain Inference Engine
Arthedain is Enotrium's inference engine. It runs fully on-device, adapts continuously to sensor drift and hardware fault in real time, and operates within the power envelope of embedded and FPGA-class hardware — air-gapped by design, no retraining cycle, and no data leaving the deployment node.
Fault immunity
Arthedain maintains classification integrity under hardware fault conditions that cause conventional inference engines to degrade. Designed for forward-deployed and physically degraded environments where hardware reliability cannot be assumed.
On-device adaptation
Sensor drift, environmental shift, and distribution change are handled locally and continuously. No ground station. No retraining. No round-trip.
Power and footprint
Validated on embedded hardware for deployment in constrained power envelopes: UAV payloads, industrial edge nodes, implantable and wearable systems.
Where Arthedain Deploys
UAV and Drone Payloads
Arthedain's profile fits the power envelope of small unmanned airframes. The same adaptation mechanism that handles sensor drift generalizes directly to UAV perception — shifting illumination, atmospheric interference, sensor wear — with no data leaving the aircraft and no ground station connection required. UAV-specific sensor fusion integration is the current engineering milestone.
Manufacturing Edge Nodes
Industrial process control at biorefinery and manufacturing edge nodes — fiber grading, throughput optimization, contaminant screening — where latency-deterministic inference is required and hardware operates in electrically noisy, physically demanding environments. Arthedain's fault immunity and fixed memory footprint make it the appropriate inference architecture for this deployment class.
Sovereign and Air-Gapped Infrastructure
For allied-nation governments and cleared programs requiring inference that stays on sovereign hardware. Enotrium's deployment model is edge-native: inference, adaptation, and data remain on sovereign hardware by architecture. Dedicated deployment packages available for vetted partners under appropriate confidentiality terms.
Implantable and BCI-Adjacent Hardware
Arthedain's original development target was implantable brain-computer interface decoding under power constraints. The architecture has been validated on decoding benchmarks.
Hyperspectral and LiDAR Sensing Pipeline
The Icarus sensing layer combines multispectral and hyperspectral imaging with high-density LiDAR — deployed on autonomous platforms, ground vehicles, and fixed sensor arrays. Icarus pipelines detect soil contaminants, characterize biological material at the field scale, and provide the sensor inputs that Arthedain processes on-device.
The same sensing architecture developed for agricultural biosecurity generalizes to aerial reconnaissance, material characterization, and anomaly detection from UAV and ground platforms.
Supply Chain Provenance
The Enotrium provenance layer uses zero-knowledge proofs to allow downstream operators to verify supply chain compliance without exposing underlying counterparty data, operational details, or farm-level information.
Selective disclosure with cryptographic guarantees. Built under post-quantum primitives.
Production deployment of the provenance API requires engineering investment and IC validation partnerships. Early access available for qualified procurement programs.
Integration and Compatibility
The Enotrium API uses a REST interface with JSON payloads. It is not OpenAI-compatible — the input and output schema reflect physical sensor data, inference results, and adaptation state, not language model completions.
SDKs for Python and C++ are in development for the early access program. Direct REST integration is available immediately for vetted partners.
For deployment packages, Verilog synthesis targets, and hardware integration support: contact the engineering team directly.
Apply for Early API Access
The Enotrium API is available to vetted partners. We review applications on a rolling basis and prioritize by deployment context and mission alignment.
Priority access tracks:
- Autonomous systems and drone manufacturers targeting comms/GNSS-denied operation
- Defense primes and system integrators with domestic-content compliance requirements
- Industrial operators deploying edge inference at manufacturing or biorefinery nodes
- Allied-nation sovereign infrastructure programs
To apply: send your organization, deployment context, and intended use to access@enotrium.org.
For sovereign and government deployments: contact us directly.