Skip to content
MAÂT
frRequest a Demo
MAÂT — Industrial Physical Intelligence

Your Machines. Their Digital Twins. Their Full Potential.

Connect industrial machines, understand their physical behavior, simulate optimized configurations and compare real performance with engineering-validated possibilities.

Request a Demo Explore the Platform
  • OPC UA · Modbus TCP · MQTT · CSV
  • 16+ physics & machine models
  • Non-invasive deployment
Real machine and its digital twinLeft, a pick-and-place gantry; right, its wireframe digital twin and the optimized trajectory computed by the MAÂT engine; centre, the data flows.REAL MACHINEx(t)τ(t)v(t)DIGITAL TWIN
◇ Demo DataVirtual machine DEMO-PP-01, synthetic data computed by the MAÂT engine. Not a customer result.
Explore twin →
Actual cycle time
3.96 s
field stream (A)
Simulated optimized cycle
3.06 s
-22.8 %
Energy per cycle
231 → 208 J
-9.8 %
Motion precision
0.047 → 0.004 mm
placement error
Vibration level
0.66 → 1.42 mm
peak deflection · limit 2.0 mm
Optimization opportunity
+268 parts/h
simulated, to validate by trial

The industrial challenge

Your Machines Generate Data. MAÂT Helps Turn It Into Engineering Decisions.

Performance losses rarely show on a dashboard. They hide in the physics of motion.

  • 01
    Excessive cycle times
    Cautious motion profiles, tuned once and never revisited.
    MAÂT MOTION
    Multi-objective optimisation of motion profiles under torque, speed and deflection constraints.
  • 02
    Under-used machines
    Nobody knows what the machine could really do.
    MAÂT REAL vs OPTIMAL
    Continuous comparison of the real machine with a calibrated and an optimized twin.
  • 03
    High energy consumption
    Energy is metered at the panel, not per cycle phase.
    MAÂT ENERGY
    Energy per phase, per axis and per recipe, simulated and compared.
  • 04
    Vibrations and mechanical stress
    Accelerating harder can cost accuracy and wear.
    MAÂT TWIN
    Structural model: tool deflection, settling time, wear indicator.
  • 05
    Unplanned stops
    Slow drifts go unnoticed until failure.
    MAÂT HEALTH
    Drift detection (CUSUM) and governed twin recalibration.
  • 06
    Fragmented data
    PLCs, historians, files: no coherent physical view.
    MAÂT CONNECT · DNA
    Read-only connectors and normalised units in a machine passport.
  • 07
    Changing a parameter is risky
    Without evidence, nobody touches a running machine.
    MAÂT INTELLIGENCE
    Time-stamped validation dossier, four-eyes approval, supervised trials. MAÂT never writes to a PLC.

How it works

Six steps, from the real machine to a measured gain.

Every step produces verifiable evidence before the next one.

STEP 01 — CONNECT

Connect the real machine or import its data.

OPC UA, Modbus TCP, MQTT, read-only; signals mapped, units converted to SI.

Real machine vs optimized digital twin

See the Gap. Understand the Cause. Explore the Potential.

Three synchronised behaviors: the real machine, its calibrated twin and its optimized twin. Pick a quantity, a scenario, show the constraints.

◇ Demo Data
Scenario summary
ScenarioReal cycleSimulated optimized cycle
Pick-and-place gantry — nominal conditions3.96 s3.06 s
Same gantry — increased friction (simulated wear)3.96 s3.06 s
Heavy cartesian palletiser8.31 s6.29 s

The platform

Twelve suites, one shared physical model.

Every suite relies on the same calibrated twin: what one computes, the others can check.

“Available”: implemented in MAÂT and proven on virtual machines, synthetic and simulator data. “Pilot”: offered with a validation plan. No gain has been measured at a customer yet.

Industries

Where motion drives performance.

Typical use cases per sector. These are intended applications, not customer references.

Discrete manufacturing

Automotive

  • Transfer gantries between presses
  • Assembly-line cadence balancing
  • Axis energy in continuous cycles
MAÂT MOTIONMAÂT PERFORMANCEMAÂT ENERGY

Machines modelled today are cartesian gantries (pick & place, palletisers, CNC routers). Other architectures require a dedicated model.

Solutions by role

Designed for Every Industrial Role.

One physical truth, presented in each profession's language.

ECONOMICS · FLEET

CEO & Executive Leadership

Economic vision and investments

  • Estimated value per site, with its evidence basis
  • Investments ranked by expected return
  • No promises: gains measured, then validated

Technology & trust

Built for demanding industrial environments.

What is in place today, and what is in development.

  • Read-only on equipment

    Available

    MAÂT neither controls nor writes any machine parameter. Every change remains a human decision.

  • Secure industrial connectivity

    Available

    OPC UA in SignAndEncrypt with verified certificate thumbprint; Modbus TCP and MQTT read-only.

  • IT / OT separation

    Available

    Connectors declared by an administrator, disabled by default, traced activation.

  • Access control

    Available

    Roles and per-site scopes, strong authentication (MFA), OIDC SSO, separation of duties.

  • Multi-tenant isolation

    Available

    Per-company isolation at database level (Row-Level Security), hosting in the European Union.

  • Traceability

    Available

    SHA-256 chained audit log; time-stamped validation dossiers.

  • Model governance

    Available

    Twin versions, validation gate (R² ≥ 0.95, NRMSE ≤ 5%), activation by another engineer, rollback.

  • Uncertainty management

    Available

    Validity domain, constraint margins, model-vs-real gap shown with every recommendation.

  • Edge / Cloud architecture

    In development

    Execution next to the machines for isolated sites.

Optimisation recommendations are validated before any physical application: validation dossier, approval by another authorised person, supervised trial, measurement. MAÂT claims no certification.

ROI & business value

Estimate the value, with its assumptions.

Every formula is shown. The result is indicative: only a pilot measures a real gain.

Additional theoretical capacity
+175,518 parts / year
Estimated energy saving
2,970 kWh · €600
Potential economic value
€351,600 / yr
Solution cost
€35,000 / yr
Indicative ROI
905 %
Indicative payback
1.2 months
Formulas and assumptions
  • cycles/yr = h × 3600 / cycle = 330,560
  • extra parts = machines × cycles/yr × (1/(1−g) − 1)
  • output value = extra parts × value/part = €351,000
  • kWh = machines × kWh/yr × e ; € = kWh × price
  • ROI = (value − cost) / cost ; payback = cost / value × 12
  • Assumption: additional output can be sold. Otherwise, value the freed hours.

▲ Indicative estimate based on your assumptions. No result is guaranteed.

Pricing

Start small. Prove it. Then scale.

A diagnosis, a pilot on one machine, then a subscription sized to your fleet.

MAÂT Discover

Diagnosis and feasibility

from €3,900
excl. tax, fixed fee
  • Inventory of available machine data
  • Digital-twin feasibility opinion
  • Prioritised use cases and value hypotheses
Request a quote

MAÂT POC

One machine, one use case, evidence

from €24,900
excl. tax, scoped pilot
  • Calibrated twin with fidelity score (R², NRMSE)
  • Optimisation and validation dossier (SHA-256 fingerprint)
  • Report of gains MEASURED on the machine, if a physical trial takes place
Request a POC

Subscriptions

MAÂT Starter

First digital twins, up to 3 machines

€990HT / month
  • ◇ Up to 3 machines
  • ◇ 1 site · 5 users
  • ◇ 200 simulations / month
  • ◇ E-mail, answer within 2 business days
Request a quote

MAÂT Professional

Recommended

Multi-machine optimisation, up to 10 machines

€2,490HT / month
  • ◇ Up to 10 machines
  • ◇ 2 sites · 15 users
  • ◇ 1,000 simulations / month
  • ◇ E-mail and video, answer within 1 business day
Request a quote

MAÂT Industrial

Industrial fleet, up to 30 machines

€5,900HT / month
  • ◇ Up to 30 machines
  • ◇ 5 sites · 50 users
  • ◇ 5,000 simulations / month
  • ◇ Priority support, answer within 4 business hours
Request a quote

MAÂT Enterprise

Multi-site, central governance — on quote

On quote
  • ◇ Up to 100 machines
  • ◇ 50 sites · 500 users
  • ◇ 50,000 simulations / month
  • ◇ 24/7 support on critical incidents, account manager
Request a quote
Compare all plans
Compare all plans
StarterProfessionalIndustrialEnterprise
Machines31030100
Sites12550
Users51550500
OPC UA connectors251560
Collection frequency1 Hz10 Hz100 Hz1,000 Hz
History12 mo24 mo36 mo60 mo
Optimisations / month201005005,000
AvailabilityAvailability target 99.5%Availability target 99.5%Availability target 99.7%Target 99.9% — negotiated contractual SLA

Prices excl. tax. Initial proposals, subject to HORUS commercial validation. (PB-2026.11-1)

Industrial Pioneer Program

Become a MAÂT Industrial Pioneer.

For manufacturers, machine builders and integrators ready to run a rigorous technical pilot on a real machine.

  1. 01Machine selection
  2. 02Objectives definition
  3. 03Data analysis
  4. 04Twin construction
  5. 05Simulated optimisation
  6. 06Controlled trials
  7. 07Results measurement
  8. 08Roll-out evaluation

Pilot terms (scope, duration, price) are set out in a written proposal. No free offer or exclusivity is promised.

Apply for a Pilot

* required fields

Partners & OEM

Build the Next Generation of Intelligent Machines.

Machine builders, integrators, engineering firms, automation specialists, mechanical experts and technology partners.

Machine buildersIntegratorsEngineering firmsAutomation specialistsMechanical expertsTechnology partners

  • MAÂT integration

    Deploy MAÂT at your customers, with a partner simulation environment.

  • Model development

    Model new machine architectures with our engineers.

  • OEM solutions

    Deliver each machine with its calibrated twin.

  • Co-innovation

    Build industry modules together, each party's contributions protected.

  • Commercial partnerships

    Controlled distribution and resale.

Become a PartnerNo partnership is presented here as signed without a written agreement and an authorisation to communicate.

Resources

Frequently asked questions

Does MAÂT work with every machine?

Not yet. Available physical models cover 2-axis cartesian gantries (pick & place, palletisers, gantry CNC routers). Another architecture (6-axis robot, press…) needs a dedicated model, assessed during a Discover diagnosis.

Do we need to replace our PLCs?

No. MAÂT reads existing data read-only (OPC UA, Modbus TCP, MQTT) without changing the PLC program.

Can MAÂT be used without a real-time connection?

Yes. One-off acquisitions or data exports are enough to calibrate a twin; a continuous connection is used to track drifts.

What data do we need to provide?

Axis characteristics (strokes, masses, drives), the motion recipe and, ideally, position, speed and torque traces over a few cycles. The exact list is set during the diagnosis.

Can a twin be built without a CAD model?

Yes. A MAÂT twin is a dynamic model (masses, friction, stiffness, drives) identified from data, not a geometric mock-up.

What is the difference between a 3D model and a physical twin?

A 3D model shows the shape. A physical twin computes torques, deflection, energy and cycle time, and shows its measured fidelity against the real machine.

How does MAÂT optimise machines?

Through multi-objective optimisation (NSGA-II) of motion parameters on the calibrated twin, under torque, speed, deflection and accuracy constraints. Each proposal states its margins, validity domain and risks.

Can MAÂT change parameters automatically?

No. MAÂT never writes to a machine. A recommendation goes through a validation dossier, approval by another authorised person and a supervised trial.

How is industrial data protected?

Per-company isolation at database level, role-based access with MFA, chained audit log, hosting in the European Union. Data remains the customer's property.

How long does a pilot project last?

Typically 8 to 12 weeks for one machine and one use case, depending on data access and the possibility of trials.

How is MAÂT billed?

A fixed-fee diagnosis and pilot, then a monthly subscription based on machines and sites. No overage is billed without a signed amendment.

Can several plants be connected?

Yes, with per-site isolation and scopes. Multi-site comparison (FLEET) is currently offered as a pilot.

The Next Performance Breakthrough Could Be Inside Your Machines.

Discover how MAÂT can help your engineering teams understand, simulate and explore better machine performance.

Request Your MAÂT DemoDiscuss an Industrial Pilot
* required fields

Company

MAÂT, by HORUS.

HORUS builds MAÂT, an industrial physical intelligence system linking data, models, simulation, optimisation and engineering decisions. Automate • Optimize • Scale.