Working today
Everything in this list is built and runs on the V2 field node - on the bench and on a live test deployment. What a certified product would additionally be held to is in the section below, named rather than glossed.
- The V2 field node - a sealed field housing carrying eleven measurement channels
- Sensor acquisition with per-device baseline capture at boot
- On-device threshold evaluation classifying every reading SAFE, WARN, DANGER or CRITICAL
- Local alarm - display, LED matrix and buzzer - driven by the node itself, independent of the network
- Authenticated cloud ingest, with per-device credentials
- A real-time operator dashboard, pushed rather than polled
- Deterministic prediction: linear-regression trend, time-to-threshold estimates and z-score baseline deviation
- Cost-gated AI analysis with a per-device cooldown and a threshold-only deterministic fallback
- Incident and alert creation on DANGER and CRITICAL transitions, merged into one threat timeline
- Remote emergency command path from dashboard to node
- Public technical documentation covering architecture, wiring, firmware and the ingest API
In development
These are known gaps, identified in the engineering work rather than discovered by a customer. Naming them is deliberate: each one is a thing a buyer would eventually ask about, and the answer is better given up front.
- Traceable gas calibration - the v2 node does compute ppm, but from an Ro auto-learned at boot on the assumption that the air is clean. That is a self-calibration against ambient, not a traceable calibration against a reference gas, and the difference matters for any figure a buyer would rely on
- False-alarm validation - no controlled measurement of false-positive and false-negative rates has been performed, so no accuracy figure is published
- Stronger edge intelligence - more analysis on the node itself, reducing what a network outage costs in capability
- Independent verification of the enclosure's ingress rating
- A certification pathway - scoping which standards apply and what compliance would require
What we will not claim
Alfa HazardX holds no ATEX, IECEx, SIL or intrinsic-safety certification. It is not certified for use in hazardous or explosive atmospheres. It is not a replacement for a certified safety instrument, and it should not be specified as one.
No accuracy percentage, detection rate or response-time figure appears anywhere on this site, because none has been measured under a documented methodology. When that testing is done, the setup, hardware, dataset, conditions, metrics, limitations, results and date will be published together - a number without its methodology is not evidence.
There are no customer references, installation counts, testimonials or partner claims on this site, because there are none to report.
What is protected
The method is the subject of Indian patent application 202641098309, filed at The Patent Office, Chennai on 13 August 2026 as an ordinary application with a complete specification. It was published on 21 August 2026 under Section 11A and appears in the public register, where anyone can look it up by application number.
Why publish this at all
Because the alternative is worse. A system whose limits are hidden survives exactly until the first technical question from someone who knows the field, and industrial safety buyers know the field. Stating the boundary makes everything inside it more credible, not less.
