Platform

From a reading
to a closed job.

A chart is where the question starts. BeconixAI carries a signal all the way through — from the sensor that produced it to the work order that closes it, and the baseline that proves it worked.

What it is

AI diagnostics, energy performance and maintenance management — in one platform.

Three things normally bought separately. One here, running on the systems already installed. The AI validates the data and works out what is failing and why. The energy layer measures what it is costing and verifies what a fix actually returned. Maintenance turns that into a work order and closes it with the evidence attached.

Connects to
BMS, controllers and meters already on site, through an open Tridium Niagara edge — and fire panels through a separate listen-only gateway that never writes back to them.
Used by
FM leads, operations managers and asset owners — not only whoever can read a control graphic.
Instead of
A front-end somebody has to sit and watch. The platform watches; people decide and act.
01

Connect

Meet every site where it is

Already have a BMS, RTU or monitoring system? It gets integrated whatever the brand, read-only where required — over BACnet, Modbus or a vendor driver, which covers most of what is installed. Where a system is genuinely closed, we meter around it instead. Gaps get instrumented with sensors and meters. Either way you end up with one normalised data layer instead of a rack of dialects.

  • Vendor-agnostic integration on an open Niagara framework
  • Wireless sensors, meters and controllers where nothing exists yet
  • Edge gateways that buffer locally and forward when the link returns
  • Fire panels joined listen-only, without touching the certified system
02

Monitor

The estate on one screen

Portfolio, site, building, asset, point. Every level is a live page rather than a static rollup, and each one drills into the level beneath it.

  • Live estate view ranked worst-first, not an alphabetical list
  • Energy, comfort, asset health and open work in one picture
  • Interactive graphics, floor plans and digital twins
  • The full history behind any number on the screen
03

Detect

Alerts that arrive already diagnosed

A threshold being crossed is the start, not the answer. The reading is validated first, the alarm is confirmed as genuine rather than a spike, and AI fault detection and diagnosis names the likely fault — so what lands in front of an engineer is a diagnosis with its evidence, not a line in a list.

  • Configurable rules per asset, with delay and severity that suit the site
  • Sensor anomaly detection for drift, stale readings, out-of-range and flapping
  • AI fault detection and diagnosis that names the likely fault and traces it through the equipment chain
  • Confirm, partial or reject on every diagnosis, so the verdicts are corrected by the people who know
  • Predictive alerts from recurring patterns and weather ahead of the event
  • Real-time delivery, with the suspect points and their history attached
04

Analyze

A verdict you can argue with

Scoring only helps if you can see how it was reached. Every analyzer shows its formula, its gates and the data quality behind the call — so an engineer can agree or push back.

  • Asset performance scored on real behaviour, analyzer by analyzer
  • Comfort measured as unmet hours per space, not a satisfaction survey
  • Peer, baseline and trend signals shown next to every verdict
  • AI investigation of a performance result, and fleet-wide search for the same root cause
05

Act

One queue, SLAs attached

Here an alert becomes a work order natively — same system, same asset record, nothing re-keyed and nothing lost between two products.

  • Work orders carrying the asset, the alarm and the evidence that raised them
  • Preventive maintenance driven by real runtime rather than the calendar
  • Technicians, vendors and requesters working in their own portals
  • Compliance and calibration tracked to closure, not just logged
06

Prove

Evidence, not estimates

Anyone can claim an improvement. The question is whether it survives a hot summer, a change of occupancy and a finance review. What counts as proof depends on what changed, so the platform carries both kinds and shows the workings for each.

  • Energy verified IPMVP-style against a weather-normalised baseline, with the model and its fit visible
  • Operational fixes verified by recurrence, time between failures and SLA performance
  • Compliance and calibration evidenced per schedule, rather than asserted at year end
  • Dashboards and reports built on the live data layer, not exports
  • Board-ready energy and ESG reporting
  • Full history retained, exportable, and traceable to the point
Agentic AI

It doesn’t stop at the alarm.

Raising an alarm is the easy part. Here it is the halfway point. Each agent does one job and hands on what it found — validate, verify, diagnose, act — so by the time work is raised you can see exactly why, and who decided what.

Intelligence on bad data is not intelligence — it is a confident guess. So the raw signal is cleaned first: drift, stale values, out-of-range readings and flapping are identified before anything is allowed to become an alarm.

Sudden driftStale valueOut of rangeFlapping
Assistant

Ask anything

The assistant reads the same live data the dashboards do. Ask in plain language and it answers with the chart, the table or the report — not a paragraph of guesswork. Every answer keeps the trail back to the points behind it.

  • Which assets are costing me the most this quarter?
  • Show unmet comfort hours by floor for Tower One
  • What changed on CH-04 before the alarm?
  • Which vendors are breaching SLA this month?
AssistantAsk anything about the estate

Illustrative exchange. Responses are generated from your connected data.

Mobile

The whole platform, in a pocket.

Not a cut-down companion that only shows alarms. The same estate, the same assets and the same live points as the desktop — because the person who needs an answer is usually standing in front of the plant, not sitting at a desk.

  • Portfolio, site, asset and point — every level, live
  • Asset health with the readings behind it, not just a score
  • Alarms as they happen, with the evidence attached
  • Scan an asset to open its record on the spot
9:41
MR
Tower OneAll systems
94healthDX Unit — Kitchen 01Central Plant · updated 40s agoCommunicating
8Points
0Alarms
2Open jobs
98%Online

Live points

Space temperature30.2 °C
Setpoint22.5 °C
Fan speedHigh
Space humidity51.4 %
Unit statusRunning
Autonomous operations

Autonomous should mean fewer surprises, not more.

Occupancy schedules, group setpoints and rule-driven actions written back to the plant — on points explicitly marked writable, by users who explicitly hold the permission, with overrides that expire and every change attributed.

Energy management

Knowing the bill went up is not energy management.

Which system, on which floor, at which hour, and what is it costing against what it should. That needs metering that reconciles, attribution down to the asset, and the tariff that actually applies — not a single blended rate across a portfolio.

Metering
Incomers, sub-meters, tenant and plant meters in one hierarchy, so a building total reconciles with the parts underneath it.
Attribution
By system, floor, tenant and — where a sub-meter or an asset’s own power point reaches — by asset. Anything estimated is labelled as an estimate.
Cost and carbon
Priced per country and per utility, with Scope 2 emissions from the same figures and the source of each factor recorded.
Work management

A full CMMS, not a ticket list

An alert is only half the job. This is a complete maintenance system in its own right — the one your team would otherwise buy separately — already holding the asset the alarm came from.

01Trigger

A job starts from one of four places — an alarm the agents confirmed, a maintenance schedule falling due, a request raised by an occupier, or a defect found on an inspection. Not every alarm earns a job, so the queue does not fill with noise.

Confirmed alarmSchedule dueOccupier requestInspection finding
Measurement & verification

Measurement and verification, done to the standard

Claiming a saving is easy. Proving one against a building that changed occupancy, weather and operating hours over the same period is the hard part — and it is the only version a finance team will accept. This is a full IPMVP module, not a chart with a line on it.

Whole-facility baseline · IPMVP Option C Baseline period Reporting period
Outside air temperature (°C)Energy (kWh/day)2428323640model · R² 0.94avoided energy

Illustrative. The fitted model, its variables and its goodness of fit are all inspectable in the platform.

The baseline
Regressed on the drivers that actually move consumption, with outliers flagged rather than deleted.
The test
R², CV(RMSE), mean bias error and an F-test, each scored against its published acceptance threshold.
The result
Avoided energy against metered actual — then priced and converted from that same figure.
The difference

You shouldn't be investing in two platforms

Buy monitoring and maintenance separately and that is two licences, two asset registries, two sets of users to administer, an integration project to make them agree with each other — and someone maintaining that integration for as long as you own both. BeconixAI is one product that does both.

Two platforms

  • ·An IoT platform for monitoring
  • ·A separate CMMS for maintenance
  • ·Two asset registries that drift apart
  • ·Middleware and an integration project to join them
  • ·Two licences, two vendors, two support contracts
  • ·Findings re-keyed from one system into the other

One platform

  • Monitoring, analytics and maintenance in one product
  • An alert becomes a work order natively
  • One asset registry behind every module
  • Nothing to integrate between them
  • One licence, one vendor, one support line
  • One permission model and one audit trail

One licence, not two

You are not paying twice for the same asset list, the same users and the same estate. The commercial model follows the product: one platform, one subscription, one renewal to argue about.

No integration project

The costly part of the usual approach is rarely the licences — it is the middleware, the mapping work and the person who maintains it forever. There is nothing to join here, so none of that exists.

Configured, not custom-built

Work order types, statuses and workflows, auto-raise and auto-planning rules, checklists, failure-code trees, SLA policies, custom fields, notification rules and dashboards are all configuration. Your operation shapes the platform without a code change.

Dashboards built by you

Thirteen widget types over your own points and work-order data, arranged on a drag-and-drop grid, kept private or shared with the organisation. Nobody has to raise a ticket to get a chart.

Branded per tenant

Client, portfolio and site scoping with per-tenant identity, so a managing agent can run several brands from one deployment without them ever seeing each other.

Adds sites, not systems

Taking on a new building means connecting it — not procuring another monitoring product and wiring it into what you already run. Growth stays a configuration exercise.

Built to run estates

Scoped, permissioned and deployed the way you need it.

Multi-tenant
Client, portfolio and site scoping, with per-tenant branding.
Role-based access
Granular permissions gate every module and action.
Mobile
A native app for the field, not a shrunken desktop page.
Deployment
Cloud, private cloud, or fully in-country where sovereignty is required.

See it on your own estate

A pilot baselines your sites before anyone quotes a number. Connect what is already there, instrument the gaps, and judge the platform on your data rather than a demo.