Contact usThe Operating System
for Physical Safety.
Averiom learns what not to do from expert operator micro-corrections, predicts boundary violations 150–300ms ahead, and enforces safe envelopes at the edge — in under 5ms.
Rule-based safety systems trip on legitimate operations. False positives breed workarounds. Workarounds become incidents.
Skilled operators carry constraint knowledge in their hands. On their worst day — fatigued, distracted, under pressure — that knowledge fails. There is no backstop.
Governed agency. Learn the Negative Manifold from expert micro-corrections. Predict violations 300ms ahead. Intervene proportionally — with cryptographic evidence of every decision.
How it works
KERNEL v1.0
Beachhead markets
SEEKING DESIGN PARTNERS →Autonomous Mining Equipment
Hydraulic arm control on autonomous and semi-autonomous haul trucks and excavators. Boundary violations under multi-tonne load carry irreversible consequences. Averiom enforces operator constraint knowledge in real time — no cloud, no latency.
Composites Welding & Related Trades
Precision welding and carbon fibre layup on composite structures — including AUKUS submarine manufacturing — where sub-millimetre control is simultaneously a quality and safety requirement. Manufacturing defect rates of 3–8% in advanced composites result in costly rework and constrain critical defence production capacity. Portable .avm profiles encode expert constraint knowledge, preserving workforce IP as skilled operators retire.
Cryptographic Truth.
The world's first "Black Box" for human agency. Every intervention is signed, hashed, and immutable — contemporaneous tamper-evident records that survive post-incident examination.
Core Governance Stack
High-frequency ingestion (1kHz) of Vision, IMU, and Force data. Fused locally to detect Correction Events — micro-hesitations and tremors signalling impending error states. No network hop.
Continuously evaluates the Negative Manifold — the 3D boundary of failure space. Kalman filtering forecasts tool trajectory 150–300ms ahead. Acts on where the operator is going, not where they are.
Tiered physical response at the edge: Tier 1 haptic warning → Tier 2 viscous damping → Tier 3 hard lock via solenoid. Proportional to risk. Communicates through the tool itself.
Every intervention cryptographically signed and saved to the .avm profile. Tamper-evident, portable, auditable. Turns physical action into verifiable data.
ICL paper accepted for publication in Springer CCIS Proceedings and poster presentation at HCI International 2026, 26–31 July. Reviewed as “strong and ambitious… high-impact systems contribution.”
Working with an advisor from the Australian Government Industry Growth Programme (DISR) to develop Averiom's commercialisation strategy and validate market fit in the Australian manufacturing sector.
In active discussions with the Australian Manufacturing Capability Network (AMCN) as an Industry Partner Organisation to scope deployment pathways in composites welding and precision fabrication.
Working towards partnership with the Australian Manufacturing Growth Centre Ltd (AMGC) to accelerate Averiom's universal safety architecture across advanced manufacturing applications.
Technical writing
Inverse Constraint Learning: What It Is, Why It Is Different, and Why It Matters for Physical AI
Most AI safety systems tell machines what to do. Averiom's ICL approach learns what not to do from expert operator behaviour — a distinction that turns out to matter enormously in physical environments.
The Cage Problem: Why Industrial Safety Has Been Trading Productivity for the Illusion of Control
For 40 years, the dominant paradigm in industrial safety has been exclusion zones and hard stops. This has kept workers safe by keeping them away from the work. It is time for something better.
We are seeking design partners and investors who understand that physical AI governance is foundational infrastructure.
If you operate autonomous mining equipment or precision fabrication in Australia, we want to hear from you. If you are an investor who sees the gap between physical AI capability and safety governance — so do we.