Skip to content
WINGMANAUTOMATION
How it works

It watches, then it acts, then it proves it was right.

Wingman connects to weighers you already own over their standard industrial protocol, reads every head's state every second, and continuously retunes the feeders to hold target pack weight with the least possible giveaway.
01The sequence
  1. 01

    It watches first

    Installed in monitor mode, it reads every head, every cycle, and records what the machine does under your operators. It writes nothing at all, for as long as you want.

  2. 02

    It takes over gently

    When you arm it, it starts from your existing settings and walks them in small, bounded steps. Nothing lurches. The chart below is a real feeder setpoint being walked from where the plant had it to where the controller wanted it.

  3. 03

    It solves the whole machine at once

    Every control cycle it works out the settings for every head together, inside the machine's own limits, in a fraction of a second. It balances the whole machine rather than chasing one head at a time.

  4. 04

    It learns your heads

    In the background, it learns how each individual head on your machine behaves. No two heads are quite the same, and it stops treating them as if they were.

  5. 05

    It only acts on learning it can prove

    What it learns can only improve on the proven control, never overrule it. When it can't improve, the proven control is used unchanged.

  6. 06

    It locks in success

    When it beats the recipe's previous best, it writes a new record. When it doesn't, the old record stands. Tuning is captured per product and never lost to a reboot or a changeover.

  7. 07

    It teaches the next machine

    The second machine to run that product starts from what the first one learned, not from zero.

The design principle
Every architectural choice in this system serves one idea: capture the upside of machine learning without ever inheriting its downside.
02Step 02, in the data

What “walks, doesn't lurch” looks like.

It walks, it does not lurch
010203040MINUTES · DISPERSION SETPOINT
Swipe sideways to see the whole chart
The dispersion feeder setpoint (the top feeder that spreads product out to the heads) over the armed window. It starts from the setting the plant was already running and walks it in small bounded steps. There is no jump anywhere in this line.Measured session, 23 Jul 2026Dispersion feeder setpoint across all logged control decisions, in bounded steps.
03Steps 03 to 05, in the data

What solving the whole machine looks like.

Efficiency through the session
406080100ARMEDRECORDRECORDMODEL010203040MINUTES · EFFICIENCY %
Swipe sideways to see the whole chart
Rolling efficiency across the whole session, derived from the machine's own counters rather than estimated. The short grey stretch is the machine before the controller took over; green is under autonomous control. Gaps are periods when the line was stopped or settling and no measurement was taken. RECORD marks a new recipe best; MODEL marks the first learned model going live. The machine was starved on four of fourteen heads throughout.Measured session, 23 Jul 2026Rolling window, counter-differenced from the weigher's good-pack, missed-cycle and overweight counters. Counters are ground truth; logged per-cycle scalars are not used.
04It tunes what a person can't

What that means on your line.

Continuous per-head control

Every head, every cycle, every shift, with the same attention the best operator gives once an hour.

Whole-machine control

It balances the whole machine, not one head at a time.

Per-head learning

Your machine's heads are not identical, and it stops pretending they are.

05It keeps what it learns
Recipe memory

Tuning is captured per product and never lost to a reboot or a changeover.

Ratchet-protected records

A recipe's stored settings are only ever overwritten by settings that provably performed better.

Warm start

Product changeover resumes proven settings instead of relearning them.

Fleet seeding

The second machine to run a product starts from the first machine's learning, not from zero.

Model versioning

New models deploy only when they beat the model in service; old versions are retained for rollback.

06It fits a plant, not a lab
Fully offline

No cloud, no internet, no data leaving your site. Ever.

One box, many machines

A single on-premise computer drives up to 20 weighers.

Standard protocol

Connects over the weigher's standard industrial interface. No machine modifications.

Single application

One installed program. No database server, no runtime to install, no IT project.

Per-machine licensing

Licence what you want to automate; the rest can run in monitor mode.

Offline licensing

Activated by exchanging a short code. No licence server, no phone-home.

Runs on the OT network

Designed for an isolated plant network from the ground up.

Book a line assessment.

We'll tell you what your line is actually doing before you buy anything.