Skip to content
Aesthon Labs logoAesthon Labs
Aesthon Featherline — the decision layer for plants that still run OEE by hand

Meet Featherline.
Three layers above your MES.

Featherline turns existing plant data into plant decisions. A decision layer, a digital twin, and an agent engine — designed to fit on top of the stack you already operate.

  • Pillars03
  • Stages / loop06
  • Duty cycle24/7
  • ModeREAD-ONLY DEFAULT
SENSEREASONRECOMMENDACTFEATHERLINE
SENSEREASONRECOMMENDACT

01 · The problem today

Your plant runs in real time.
Your OEE meeting runs once a week.

Every Monday, engineers hunt through MES exports, pivot OEE spreadsheets, and argue over which deviation came first. By the time a corrective action ships, three more shifts have run with the same loss.

  1. 01 · Mon · 08:00

    Monday meeting

    Engineers and supervisors decode last week's OEE by hand. The cycle restarts before the action shipped.

    Cost · Once a week

  2. 02 · +90 min

    Spreadsheet export

    MES exports pivoted, joined with QA logs, normalised by line and shift. The same query, rebuilt every Monday.

    Cost · 8h / engineer / week

  3. 03 · +2 days

    Root-cause hunt

    Eight emails, three meetings, a Teams thread. The deviation is still unconfirmed by Wednesday.

    Cost · Cross-functional ping-pong

  4. 04 · Next week

    Action deferred

    The decision gets parked. Nothing writes back to the line. The loop never closes.

    Cost · The cycle severs here

  5. 05 · Ongoing

    Loss continues

    By the time anyone re-opens the file, the same loss has run for three more shifts on the same line.

    Cost · 3 shifts repeat the fault

Cycle break · 4d 7h open-loop

  • 8 hrs

    per engineer per week lost to data preparation, root-cause hunts, and meeting prep

  • 4 days

    typical lag from a line anomaly to a documented corrective action

  • 68%

    of plant insights live in slide decks and Excel, not in workflows that act

Estimates based on conversations with Tier 1/2 plants in Southeast Asia. Replace with your own.

Pillar 01

Decision Layer

Agents that read your MES events, reason over context, and surface the next correct action. Built so an on-shift engineer can approve or override in seconds, with a full audit trail behind every move.

Live · L4 incident

Standby

Evidence

scanning…

  • Torque driftTORQUE
  • MES cycle deltaMES
  • Vibe signatureVIBE
  • Shift handoverOPS
  • SOP-447 §3.2SOP

evidence pending

Confidence

0% CONF

Awaiting signal

Status

Connected · plant 02 · awaiting signal

Featherline is watching L1–L4 across MES, SCADA and historian. When something drifts outside SOP, the loop wakes up.

Audit log · this decision

    • 01 · Capability

      Root-cause hypotheses with confidence scoring

    • 02 · Capability

      Ranked alternatives, not single-answer black boxes

    • 03 · Capability

      Approval gates per plant, per shift, per line

    • 04 · Capability

      Audit-grade write-back to MES with rollback

    Pillar 02

    Digital Twin

    A live, queryable mirror of every line, asset, and OEE metric in your plant. Stream from MES, SCADA, and historian into a single observability surface, and ask plain-English questions of the result.

    Live capture

    Plant L4 · Twin view

    Aesthon Featherline · Digital Twin · live channels

    16:9 · MP4

    Live channels

    4 / 312 signals

    Δt 250ms

    • OEEL4 · 4h
      62.1%Δ −11.2
    • TEMPL4 · 4h
      78.3°Δ +1.2
    • VIBEL4 · 4h
      0.42gΔ +0.18
    • FLOWL4 · 4h
      144/mΔ −6

    Plain-English query

    Grounded · plant context

    why did Line 4 stall at 06:00?

    Twin · streaming

    Line 4 stalled at 06:00:04 — Press 7 torque drifted to 142 Nm over the last 14 days.

    Drift matches the pattern logged in SOP-447 §3.2. Recommend reducing torque by 4% and resuming.

    Confidence 87% · 5 evidence sources · ranked candidates available.

    Sources · 5

    Time-travel · replay

    • 01 · Capability

      Real-time line, asset, and OEE views

    • 02 · Capability

      Plain-English query across structured + event data

    • 03 · Capability

      Time-travel: replay any shift, any incident

    • 04 · Capability

      Per-asset signal anomaly detection

    Pillar 03

    Agent Engine

    Customize agent behavior, KPIs, SOPs, and approval gates per plant. Featherline agents learn your team's actual decision logic (not a generic AI template), and stay inside the lines you draw.

    Agent roster

    3 running · 1 standby · 1 paused

    AgentKPIStatus
    • OEE-WATCH-AAll shifts
      OEE · plant 02ACTIVE
    • DOWNTIME-CAUSEDay · 06–18
      MTBF · L1–L4ACTIVE
    • SOP-DISPATCHAll shifts
      SOP-447STANDBY
    • QA-DRIFTDay · 06–18
      Defect · paintACTIVE
    • CHANGEOVER-OPTNight · 22–06
      Setup · cell BPAUSED

    Config · OEE-WATCH-A

    v3.2 · locked

    • KPIOEE · target ≥ 85%
    • SOPsSOP-447, SOP-220, SOP-118
    • Approval gateEngineer · L2+
    • Write-backMES · audited
    • NotifyShift lead · email + line

    Versioned · diff v3.1 → v3.2

    Rollback ready

    Audit log

    Stream · last 60m

    1. 06:00:04agent:oee-watch-aanomaly · L4 · OEE −11%
    2. 06:00:09agent:oee-watch-acontext built · 14d window
    3. 06:00:14tool:torque_historydrift · 142 Nm
    4. 06:00:18tool:sop_lookupSOP-447 §3.2 matched
    5. 06:02:18agent:oee-watch-arecommend · Press 7 −4% torque
    6. 06:03:42human:eng_4421approved · write-back authorized
    7. 06:03:43agent:oee-watch-aMES action A-8814 dispatched
    8. 07:00:00agent:oee-watch-arecovery confirmed · OEE 88.1%
    • 01 · Capability

      Per-plant SOPs and KPI definitions

    • 02 · Capability

      Plant-aware reasoning grounded in your data

    • 03 · Capability

      Versioned workflows with shift-level granularity

    • 04 · Capability

      Full prompt + decision audit log

    04 · Where it fits

    We sit on top of what you already have.

    Featherline is a decision layer, not a data platform. We plug into your MES, SCADA, and historian (read-only by default), and surface actions to the engineers who already run the line.

    • 01No new hardware. No new sensors.
    • 02No replacement of your MES or dashboards.
    • 03Read-only integration, audit-graded write-back.
    • 04Deployable in your VPC or our managed cloud.
    03
    HumansPlant team retains control
    Illustrated portrait of a plant director holding a clipboard

    Plant Director

    Approvals

    Illustrated portrait of a process engineer reviewing a holographic schematic

    Process Engineer

    Oversight

    Illustrated portrait of a senior engineer with a ruggedized tablet on the line

    Senior Engineer

    Real-time

    02
    Featherline · Decision LayerDigital twin · Agent engine
    Live
    SENSEREASONRECOMMENDACTFEATHERLINE
    Loop · 24/7Sense → Reason → Recommend → Act
    01
    Your stackExisting plant systems

    MES

    SCADA

    Historian

    PLCs

    WonderwareIgnitionRockwellOPC-UA

    Integrations

    Speaks the protocols your plant already uses.

    • MESWonderware / AVEVA System Platform
    • MESSiemens Opcenter
    • MES / SCADARockwell FactoryTalk
    • SCADAIgnition by Inductive Automation
    • ProtocolsOPC-UA / Modbus / MQTT
    • HistorianOSIsoft PI / AVEVA Historian

    Don't see yours? We integrate per pilot. Ask us.

    Security & data

    Aesthon Featherline — Security & data: Built for the way automotive operates.

    • Read-only by default

      Write-back requires explicit human approval and is fully audited.

    • Your data stays put

      Deploy in your VPC or our managed cloud. Data never trains shared models.

    • Versioned and reversible

      Every agent decision is traceable, replayable, and rollback-safe.

    05 · A Monday morning

    06:00 · Line 4 stalls.
    By 07:00, it's already fixed.

    The agent's path through the incident — from plant signal to restored OEE — drawn as the flow it actually is.

    inputs · always-on

    MESSCADAHistorian
    1. step 01

      Sense

      anomaly detected

      L4 · OEE −11%

    2. step 02+5s

      Reason

      evidence + tools

      context ranked

      • query_mes14d · 4,218 cycles
      • torque_historydrift · 142 Nm
      • sop_lookupSOP-447 · §3.2
    3. step 03+2m 9s

      Recommend

      action proposed

      SOP-447 §3.2

    4. step 04+1m 24s

      Approve

      engineer sign-off

      eng_id=4421

    5. step 05+0s

      Act

      write-back

      MES action A-8814

    6. step 06+56m 18s

      Outcome

      OEE 88.1%

      standup cancelled

    1. · step 1 · Sense · anomaly detected · L4 · OEE −11%
    2. · step 2 · Reason · evidence + tools · context ranked
    3. · step 3 · Recommend · action proposed · SOP-447 §3.2
    4. · step 4 · Approve · engineer sign-off · eng_id=4421
    5. · step 5 · Act · write-back · MES action A-8814
    6. · step 6 · Outcome · OEE 88.1% · standup cancelled
    7. tool · query_mes · 14d · 4,218 cycles
    8. tool · torque_history · drift · 142 Nm
    9. tool · sop_lookup · SOP-447 · §3.2

    Illustrative scenario · ~30s compressed from a 1-hour incident. Numbers, SOP refs and station IDs are placeholders for the plant context Featherline learns during onboarding.

    Frequently asked

    Aesthon Featherline — Frequently asked: The questions we hear first.

    Walk us through your stack.

    We'll map where Featherline fits, what we'd integrate first, and what a 90-day pilot would look like on your data.