PHLAI
Request a DemoExplore the Platform

01Industrial digital twins

Understand the machine. Model the system. Improve the decision.

Connect industrial assets, operational data and digital representations in one environment built for equipment visibility, engineering analysis and better-informed operational decisions.

Explore the PlatformView the Digital Twin Demo

Read-only by design · Starts with the data you already have

Illustrative demo cell — not a customer facilitySelect an asset in the drawing
CELL 3 · MACHINING & PACKINGMachiningstationnot monitoredMTR-001Conveyor drive motorCNV-001Transfer conveyorPMP-001Coolant pumpPKG-001Case packerDatum 0,0 · schematic, not to scale · illustrative demo cellTwin v3 · Published · Fidelity: Structural + Visual

PMP-001

Warning

Coolant pump

Centrifugal pump · Criticality A · Cell 3

Live (demo) · 14:20
  • Vibration4.9 mm/s
  • Bearing temp71 °C
  • Pressure3.6 bar
  • Flow182 L/min
  • AL-0007 Vibration > 4.5 mm/s for 10 min
  • AL-0010 Bearing temp > 70 °C for 30 min
Open PMP-001 in the demo →

02The visibility problem

The equipment is connected. The information isn’t.

A single pump’s story is split across a historian, a maintenance system, a criticality spreadsheet, an OEM manual and someone’s memory. Each part is real. None of them sit together where a decision gets made.

  • Readings OBS

    Historian trend screens

  • Maintenance notes NOTE

    CMMS free text

  • Criticality NOTE

    A spreadsheet on a shared drive

  • OEM limits RULE

    PDF manual, §6.2

  • Shift knowledge NOTE

    A supervisor’s memory

PMP-001

Coolant pump

Each fact exists. None of them meet here.

Example uses the fictional demo asset PMP-001. Output-class tags show whether a fact is a reading (OBS), a rule (RULE) or human knowledge (NOTE).

03One connected industrial model

Connect. Model. Observe. Analyze. Improve.

Five stages, one model. Each stage adds to the same asset record, so nothing is re-entered and every finding points back to its data.

Connect

Bring readings in from CSV or Parquet exports, the bundled sample data, or MQTT and OPC UA sources through a gateway on your network.

Stored on the asset
Signals · quality codes · timestamps
Where it lives in the product
Data sources

Next: Model →

  • CSV / Parquet exportAvailable in MVP
  • Sample datasetAvailable in MVP
  • MQTT via gatewayRequires adapter
  • OPC UA via gatewayRequires adapter

{ "signal": "p301-vib", "ts": "2026-10-14T14:20Z", "value": 4.9, "unit": "mm/s", "quality": "good" }

04Flagship product · PHL TwinCore

One view of each asset, its data and its history.

The digital-twin workspace for maintenance and engineering teams. Useful from day one with a structured asset model; 3D is optional. Select an asset in the hierarchy, or switch tabs.

Northstar Manufacturing Cell 3 PMP-001Illustrative data · demo time 2026-10-14 14:20

PMP-001 Coolant pump

Centrifugal pump · Criticality A · Live (demo) · 14:20

Warning

  • Vibration

    4.9 mm/s

    warning > 4.5 · live

  • Bearing temp

    71 °C

    attention > 70 · live

  • Pressure

    3.6 bar

    attention < 3.2 · live

  • Flow

    182 L/min

    attention < 170 · live

  • Current

    14.8 A

    warning > 17 · live

  • Power

    4.60 kW

    no rule · live

Recent events

  1. 2026-10-14 10:50 AL-0010 openedBearing temp > 70 °C for 30 min
  2. 2026-10-14 05:00 AL-0007 openedVibration > 4.5 mm/s for 10 min
  3. 2026-10-09 12:40 Twin v3 publishedBearing temperature attention rule (70 °C, 30 min) added.
  4. 2026-10-06 12:00 Gateway buffer gapNo readings 2026-10-06 10:00–12:00. Interval excluded from analysis.
  5. 2026-09-28 05:00 Gateway buffer gapNo readings 2026-09-28 02:00–05:00. Interval excluded from analysis.

Related equipment

  • coolsSupplies coolant to the machining station upstream of CNV-001
Open TwinCore in the demoTwinCore details

05From data to digital twin

From a tag list to a working twin in six steps.

What your twin can show depends on the signals, integrations and engineering information you provide. We start with what exists and show you exactly what’s missing.

Cell3.P301.VibDE → Drive-end bearing · mm/s

Each source tag binds to an asset, an optional component and a canonical unit.

Twin fidelity after this step
  1. Structural
  2. Visual
  3. Parametricneeds engineering parameters for a scenario model
  4. Validatedonly when an engineer records validation evidence

06Equipment intelligence

Every flag shows its working.

PHL AI separates what was measured, what a rule concluded and what a statistical check noticed, so your team can trust, question or tune each one. Not every fault is visible in the available data, and PHL AI does not claim to predict failures.

PMP-001 vibration velocity RMS (mm/s)10-minute readings
0.02.04.06.08.014:2020:2002:2008:20nowWarning 4.5Critical 7.1

Vibration velocity RMS over the last 24 hours: ranged from 4.0 to 5.0 mm/s, mean 4.5 mm/s. Latest reading 4.9 mm/s at 2026-10-14 14:20 demo time. Thresholds: warning above 4.5 mm/s; critical above 7.1 mm/s.

Reading warning > 4.5 mm/s critical > 7.1 mm/s

Why flagged

  1. OBS Reading

    Vibration velocity RMS: 4.9 mm/s at 14:20 demo time

  2. RULE Rule

    Vibration > 4.5 mm/s for 10 min · source: OEM manual §6.2 (demo)

  3. STAT Statistical context

    60-min mean 4.9 mm/s vs 30-day baseline 2.7 ± 0.91 → z = 2.3

  4. RULE Review Priority

    80 · High — 4 contributing factors

  5. NOTE Decision

    Recorded by a person. PHL AI does not act on equipment.

RULE Rules-based alerts

Deterministic conditions you configure and own, such as “above 4.5 mm/s for 10 minutes”. Available in the first release.

STAT Statistical checks

Deviation from an asset’s own recent baseline, with the window and limit visible. They point to something unusual, not to a cause.

MODEL Predictive models Future capability

Trained models need labelled failure history and published evaluation. PHL AI does not claim to predict failures today.

07PHL Simulate

Compare before you change.

Replay an alert policy against a month of recorded readings, or check what a conveyor speed change does to a cell’s bottleneck, with every assumption written down.

These are deterministic scenario models. Physics-based simulation requires an appropriate engine and validated engineering models. Future capability

PHL Simulate

SCEN S2 · Cell capacity model · deterministic

Illustrative scenario
%
OutputBaseline 100 %Alternative 115 %Change
CNV-001 conveyor rate (units/h)1,2001,380+180
Cell throughput (units/h)1,2001,214+14
Bottleneck station CNV-001PKG-001changed
Packer infeed buffer Not fillingFull in ≈ 43 minchanged
MTR-001 load (linear estimate) (%)6878.2+10.2
  • MTR-001 load ≤ 85 % (assumed limit)78.2 % estimated
  • Packer infeed buffer does not fill within 60 minFull in ≈ 43 min
Calculation and assumptions

throughput = min(1,200 × speed, 1,320 × 0.92) · buffer fill = 120 ÷ surplus · load = 68 % × speed

  • Steady state only: no start-up, changeover or stochastic stoppages.
  • PKG-001 effective rate = rated 1,320 units/h × 92 % planned availability, treated as a constant.
  • CNV-001 rate scales linearly with belt speed from its rated 1,200 units/h.
  • MTR-001 load scales linearly with conveyor speed from its 30-day mean. Not validated for this drive.
  • Packer infeed buffer holds 120 units and starts empty.

Results follow from the model and assumptions listed. This is not a physics simulation and not a validated prediction.

08PHL Maintain

From a signal to a recorded decision.

Review items gather the equipment, the condition that triggered them, supporting readings and a suggested next step. A person assigns, inspects and records the outcome. PHL Maintain exports work orders as files; it does not dispatch them.

  • RULE Priority comes from listed rules, never a hidden score
  • NOTE Decisions and outcomes are stored on the asset
  • DATA Stale data is flagged, not ranked

RV-0003 · Maintenance review

Open · unassigned
Equipment
PMP-001 Coolant pump · Criticality A
State
Warning
Triggering condition
Vibration > 4.5 mm/s for 10 min RULE
Supporting telemetry
4.9 mm/s now · bearing 71 °C · flow 182 L/min
First observed
2026-10-14 05:00
Suggested next action
Schedule a drive-end bearing inspection on PMP-001 (lubrication condition, bearing noise, coupling alignment). Compare with the 21 Sep inspection record. RULE

Review Priority RULE

80High

  • Active warning rule: Vibration (AL-0007)+35
  • Active attention rule: Bearing temp (AL-0010)+15
  • Rising 7-day trend: Vibration +115 %+20
  • Criticality A+10

Rules-based ranking to help order inspections. Not a probability of failure.

Illustrative review itemInvestigate in the demo →

09Industrial use cases

Built around the work, not the dashboard.

Four workflows the first release is designed for, each on the same demo cell.

A

Equipment monitoring

Workflow: a reliability engineer opens the cell at the start of a shift, sees which assets are normal, which need attention and which have stopped reporting, and opens the one that matters.

PHL AI’s role: keep identity, readings, thresholds and freshness on one record per asset.

Equipment monitoring →
AssetStateLast reading
MTR-001 Conveyor drive motorAttention14:20
CNV-001 Transfer conveyorNormal14:20
PMP-001 Coolant pumpWarning14:20
PKG-001 Case packerStale13:30

B

Maintenance intelligence

Workflow: a planner orders the week’s inspections by Review Priority, reads why each asset was flagged, assigns the work and records the outcome.

  1. PMP-00180
  2. MTR-00120
  3. CNV-0015
  4. PKG-001n/c

Rules-based ranking, illustrative data. “n/c”: not computed because data is stale.

PHL Maintain →
CELL 3 · MACHINING & PACKINGMachiningstationnot monitoredMTR-001Conveyor drive motorCNV-001Transfer conveyorPMP-001Coolant pumpPKG-001Case packerDatum 0,0 · schematic, not to scale · illustrative demo cellTwin v3 · Published · Fidelity: Structural + Visual

C

Digital-twin visualization

Workflow: an engineer traces what depends on what — the motor drives the conveyor, the conveyor feeds the packer — before planning a stop.

PHL AI’s role: a versioned model of assets and typed relationships. A 2D plan is the default; a 3D view is planned as another view of the same twin.

Digital twins →

D

Scenario analysis

Workflow: before raising conveyor speed, a production engineer checks which station becomes the bottleneck and how quickly the packer buffer fills.

throughput = min(station rate × availability)

Deterministic models with listed assumptions. Physics-based simulation needs an external, validated engine.

Scenario analysis →

10Integration architecture

From the machine to the screen, one direction.

Proposed architecture: readings leave the plant through a gateway you approve, are validated and stored, and become twin state. AWS services are named as the planned hosting design, not as a running deployment.

  1. A

    Physical equipment

    • Machines & sensors
    • PLCs
    • SCADA / historian
    • CMMS records

  2. B

    Edge connectivity

    • OPC UA read-only clientRequires adapter
    • MQTT bridgeRequires adapter
    • Store-and-forward bufferRequires adapter

  3. C

    Data ingestion

    • CSV / Parquet importAvailable in MVP
    • Validate · normalise units · dedupeAvailable in MVP
    • AWS IoT Core + queuePlanned

  4. D

    Industrial data platform

    • Asset metadata (PostgreSQL)Planned
    • Time-partitioned telemetryPlanned
    • Files & history (Amazon S3)Planned

  5. E

    Digital twins

    • Signal map & relationshipsAvailable in MVP
    • Versions & freshnessAvailable in MVP
    • 3D view (glTF)Planned

  6. F

    Analytics & simulation

    • Rules & statistical flagsAvailable in MVP
    • Scenario models S1 · S2Available in MVP
    • Trained predictive modelsFuture capability

  7. G

    Application

    • Twin explorer & telemetryAvailable in MVP
    • Alerts & review queueAvailable in MVP
    • ReportsAvailable in MVP

Control boundary. Readings flow from the plant to PHL AI over outbound TLS. There is no write path back: PHL AI does not send commands to PLCs, setpoints or production controls. People act through the systems they already operate.

Available in MVP
Designed for the first release and shown in the demo with sample data.
Requires adapter
Works through a gateway, adapter or customer configuration on your network.
Planned
On the roadmap, designed but not in the first release.
Future capability
Requires models, infrastructure or validation that does not exist yet.
Integration catalogue

11Enterprise architecture & security

Designed for industrial data. Labelled honestly.

The controls below describe the intended design. Nothing here is independently audited or certified yet, and we will only publish attestations once they exist.

CapabilityDesignStatus
Organization-level accessOrganizations contain workspaces; every record carries its workspace.Available in MVP
Role-based permissionsOwner, Admin, Engineer, Technician, Viewer.Available in MVP
Audit historyAppend-only log of rule, twin, alert and access changes.Available in MVP
Secure data handlingTLS in transit, encryption at rest, credentials in a secrets vault.Available in MVP
Integration boundariesOutbound-only gateways with per-device certificates and read-only adapters.Requires adapter
Data retention controlsPlan-based retention of raw readings and rollups.Available in MVP
Monitoring & recovery planningAlarms on ingestion lag and errors; backup targets defined, not yet tested.Planned
Single sign-onSAML / OIDC.Planned
Independent security attestationNo certification or audit has been obtained.Not yet

Security design in detail →

  • Flagship · first release

    PHL TwinCore

    Industrial digital twin & asset intelligence

    One versioned, data-connected model of each asset: identity, signals, relationships, thresholds and history.

    PHL TwinCore →

  • Review queue in first release

    PHL Maintain

    Equipment health & maintenance review

    Rank equipment for review with transparent, rules-based priority and record every decision against the asset.

    PHL Maintain →

  • Two scenario models in first release

    PHL Simulate

    Scenario analysis workspace

    Compare a baseline with alternatives using deterministic scenario models, with every assumption written down.

    PHL Simulate →

Bring your industrial systems into clearer view.

Start with a focused equipment workflow and expand toward connected digital twins and industrial simulation as your requirements evolve.

Request a DemoExplore TwinCore

Illustrative demo · no signup · not connected to equipment