Injection molding · shop-floor control

Injection molding, in control.

Machines, molds, parameters and maintenance in one system on your network. Full mold history, changeovers checked against the setpoint, cycle-based Wartung and OEE — with live MES data and early scrap warnings on the way.

Request a demo Explore modules
  • FULL MOLD HISTORY
  • MES LIVE · NEXT
  • SETPOINT CHECK
  • CYCLE-BASED WARTUNG
  • OEE · DASHBOARD
Deployment
On-premise
Stack
Flutter · FastAPI
Status
Pilot testing
01 / OBSERVED

Where the process leaks

3 failure modes
01

Half of the process lives on paper. Checklists in a drawer, MES in another system, the rest in people's heads.

AUDIT TRAIL · OFFLINE
02

When a defect surfaces, root-cause chasing eats a shift. Not minutes — a shift.

ROOT-CAUSE · 1 SHIFT
03

Fault history is written like a diary. Filter, group, count — none of it works.

LEARNING LOOP · NONE
02 / WHAT WE BUILT

One system for the shop floor

8 + 4 modules

What setters, toolmakers, quality and admins use every shift. Point at a card for the detail.

Available

Hall map & machine panel

One card per machine. Tap it and it opens into exactly the actions its state and your role allow.

Hall map & machine panel

Available

The main screen of the shop floor. Each card shows whether the machine is free, idle or producing, with the mold and part on it — one look at the wall screen tells you where to go.

  • Panel shows mold, production start, material, mold cycles and cycles left to Wartung
  • Open mold findings visible before anyone touches the machine
  • Machine is locked while a colleague fills a checklist
  • A PC next to a machine can be bound to it as a station
Available

Guided changeover & parameter check

Mount, start-up, unmount and parameter checks as step-by-step wizards — every value checked against the setpoint.

Guided changeover & parameter check

Available

Checklists come from the admin, not from code. Each parameter is compared with its setpoint and tolerance; an out-of-tolerance value needs a comment and goes to the responsible role.

  • Mount, start-up and unmount wizards ending in an editable summary
  • Material checked against the mold's allowed list
  • Start-up and preventive checks — an overdue check becomes a task
  • Suspiciously fast confirmations are logged
Available

Mold history & findings

Every Wartung, repair, finding and unmount of a mold on one timeline — who, when, what.

Mold history & findings

Available

Findings are reported per category and cavity, with a photo. An open finding belongs to the mold: it survives unmount and the next setup, and a new report joins it instead of creating a duplicate.

  • Status per finding: in progress · can't fix · fixed — with who started and who closed it
  • Top fault categories of the last year in the history header
  • Filter by category and cavity
  • Colour always means status, never author
Available

Toolroom & cycle-based Wartung

A mold map for the toolroom. Cycle limits per mold and a task before the limit is crossed.

Toolroom & cycle-based Wartung

Available

Wartung 1 and Wartung 3 have cycle limits per mold; the mold card shows how close it is and turns amber from 75 %. Over the limit, the toolroom is notified automatically.

  • Wartung 1 · 2 · 3 recorded straight from the mold panel
  • Three map layouts: categories, compact, free drag-and-drop
  • The counter can't be lowered by mistake
  • Automatic counter from the MES — in preparation
Available

PD Analyzer

Drop a process-data export from the machine and get median, sigma and Cpk in seconds.

PD Analyzer

Available

Reads CSV / XML exports from ENGEL and KraussMaffei machines, 2K included. Downtimes and sessions are detected automatically, NOK cycles highlighted, and the result goes to quality as a report.

  • Median, sigma, Cp / Cpk against your LSL / USL
  • Downtime and session detection
  • Stacked, overlaid, table and Cpk views
  • Report export to PDF, DOCX or PNG, in English or Slovak
  • Live feed from the MES — in preparation
In pilot test

OEE per machine and hall

Availability × Performance × Quality for every machine and the whole hall — by day, week, month or year.

OEE per machine and hall

In pilot test

The admin imports the MES daily report (6:00 → 6:00); the numbers are stored in your database, so history stays even without the MES. Every machine card on the hall map carries its OEE — tap it for downtime and losses.

  • Machine panel: day timeline with app events, downtime Pareto, trend
  • Hall tile with OEE, utilisation and TEEP
  • Targets and lower limits set by the admin
  • Numbers match the MES screen in the hall, by design
In pilot test

Dashboard & KPIs

One page for admins: KPIs with trends, top molds, fault rates — and what needs attention now.

Dashboard & KPIs

In pilot test

The Overview is the second view of the Dashboard, for the same period as the OEE map. It shows where the plant loses time and quality, and which molds, parameters and people need attention.

  • 5 KPI cards with a 30-day trend
  • Top molds, fault rate, most-changed parameters, toolroom reaction
  • Admin notes chat next to the numbers
  • Works on a phone
Available

Configurable, not hardcoded

Machines, molds, checklists, setpoints, fault categories, shifts and roles — all set by the admin.

Configurable, not hardcoded

Available

When the plant changes, the admin changes the app — no developer, no new release. Every change to master data is recorded with the old and new value and its author.

  • Checklist editor with step types and per-machine definitions
  • Fault categories with icons and colours
  • Users, roles and shift groups
  • Hall layout and company logo
ALSO IN THE APP

Tasks & notifications

One pill for everything waiting on you — tasks, repairs, confirmations.

Available

Admins assign tasks in a five-step wizard; recurring tasks are created automatically on a schedule. Each role sees only what is addressed to it.

  • Recipients by role, shift group or person — with place, deadline and photo
  • Recurring tasks, e.g. the monthly moisture measurement
  • Quality confirms the last parts at unmount
  • Chat rooms per role

Materials, dryers & shifts

Dryers with their materials and moisture limits, plus the shift schedule.

Available

Each dryer card shows its materials, temperatures and maximum moisture, and measurements are recorded per dryer. The schedule generates the rotation of shift groups A–D.

  • Dryer, non-dried and additive cards
  • Moisture measurement with a task that stays open until OK
  • 3-shift (8 h) and 4-shift (12 h) rotation
  • Calendar view of the schedule

Built for the shop floor

Gloves, wall screens, shared PCs — the app is designed around them.

Available

Large touch targets, a numeric keypad that works in gloves, 125 % default text and an idle screen readable from across the hall.

  • Glove-friendly keypad and touch targets
  • Idle screen with open tasks per role
  • Station mode for a PC next to a machine
  • Trusted-device login, auto-lock after inactivity

Nothing gets lost

A Wi-Fi drop doesn't lose a record, and every change has a name on it.

Available

Records saved during an outage wait in an offline queue for up to 24 h and are sent in order when the connection returns. The server spots replays and won't overwrite a colleague's newer entry.

  • Offline queue, no duplicates on retry
  • Change history: who, when, old → new
  • Deletes are audited too
  • Recording on behalf of someone is logged as such
03 / COMING NEXT

Where the app goes next

in preparation

Built on what already runs. The items with a written plan come first.

  1. NEXT · TIG FIRST

    Live MES connection

    Machine state, alarms, the mold shot counter and live parameters, read straight from your MES.

    In preparation

    Read-only, from the backend, through the MES's official interface — never its database, and nothing is written back. First connector: TIG authentig.

    • Live machine state, mounted mold and alarms on the hall map
    • Automatic shot counter in its own column — Wartung limits use whichever is active
    • Actual parameters loaded for the check against the setpoint
    • Order pieces and scrap per machine
  2. NEXT · PLAN READY

    Scrap cost & limits per part

    What scrap costs per machine and hall — and which part is over its limit.

    In preparation

    The admin sets a price per piece and a scrap limit per part number, each valid from a date, so history stays right when prices change.

    • Scrap cost per machine and for the hall
    • Part status: OK · near limit · over limit
    • Machine card shows its worst part
    • Part totals per period and the hall's top 5
  3. PLANNED

    Drift & early scrap warning

    A warning when a process starts walking away — before parts start failing quality.

    In preparation

    Alerts on drift and trends, not only on tolerance breaches — designed against alarm fatigue. Root cause traceable through PD Analyzer and the mold history.

    • Trend and drift detection on live parameters
    • Linked to findings and mold history
    • A few alerts that matter, not a wall of red
  4. CONCEPT

    AI assistant per mold

    Ask about a mold and get an answer from its own history.

    Concept

    A read-only advisor that sees a mold's findings, Wartungs and parameter history and suggests where to look. It runs on your company's own AI through the same API — your data stays under your control.

    • Read-only: suggests, never changes data
    • Runs on your corporate AI, not ours
    • Uses the same REST API as the app
ALSO ON THE LIST
  • Export from OEE / KPI
  • Real-time notifications
  • Permission matrix
  • Granulate history
  • PD Analyzer zoom
  • Mold MTBF statistics
04 / PROVENANCE

Built inside the shop floor

who built it · why

Built by a process technologist with hands-on injection-molding experience. The product grew inside the shop floor — not written to a spec by a vendor who read about the shop floor.

— author of MedianMolding

05 / FOR IT / DIGITIZATION

Made to sit inside your network

on-premise · PostgreSQL · JWT

The technical answers your digitization lead will ask before signing off.

DEPLOYMENT

On-premise

Application, database and files stay inside your network. No outbound internet calls.

  • Backend as a Windows service, web build served from your own web server
  • PostgreSQL with Alembic migrations — the backend refuses to start on an outdated schema
  • Fonts and web engine bundled — no CDN, restricted-network safe
  • Backups and a restore test agreed with your IT
INTEGRATION

MES, read-only

Reads through your MES's official interface. Never its database, never writes back.

  • Today: MES daily report import for OEE
  • Next: live machine state, alarms, mold shot counter
  • Backend-to-MES only — clients never talk to the MES
  • Manual counter stays as a fallback
SECURITY

Security

JWT, bcrypt, roles, rate-limited login and a full audit trail.

  • Every API route needs a token, except login, version and public startup data
  • Tokens revoked on password change
  • API documentation off in production
  • No passwords or personal data in API responses
STACK

Stack

Flutter for Windows, Android and Web from one codebase. FastAPI and PostgreSQL behind it.

  • Flutter · Dart — Windows desktop, Android, Web; iOS from the same code
  • FastAPI · SQLAlchemy · Python 3.12 — 157 REST endpoints
  • 1 300+ automated tests, backend and frontend
  • No GPL / AGPL in the dependency tree
DATA FLOW · read-only

How data moves

The app keeps its own data in your PostgreSQL. From the MES it only reads — through the official interface, never its tables — and nothing is written back.

READ from MES: machine state · mold counter · orders · OEE · WRITE to MES: nothing
SECURITY · COMPLIANCE

Speaks the language your auditors ask about

Answers to the questions on every automotive-supplier gate check.

  • TISAX automotive supply-chain
  • IEC 62443 OT / IIoT security
  • GDPR no PII outside the plant
  • EU residency on-premise · your box
WHERE THIS FITS

MES

MES runs orders, materials and machine data. We're the shop-floor layer above it — checklists, setpoints, mold history, findings, Wartung. It reads from the MES; it doesn't replace it.

Excel

Excel is dead by 5 pm Friday. Cpk in a spreadsheet is a snapshot, not a system. Here every record is searchable and tied to the mold, machine and person.

Build in-house

A shop-floor system takes 12–18 months to build well. This one is already built, in pilot testing and shaped by a process technologist. In-house is fine — if you have a two-person team free for two years.

06 / QUESTIONS

The questions you'll actually ask us

7 answers
Q01 How do you integrate with our MES?

Read-only, through your MES's official interface — never its database, and nothing is written back. Today the MES daily report is imported for OEE; the live connection (machine state, alarms, mold shot counter, parameters) is in preparation, starting with TIG authentig. Other MES are mapped during discovery.

Q02 Where does the data live?

On your server, in your PostgreSQL, inside your network. The app makes no outbound internet calls — no cloud in the operational path. You own the data end-to-end.

Q03 What runs today and what is coming?

Every module marked Available runs today; OEE and the Dashboard are in pilot testing. The live MES connection, scrap cost and drift warnings are in preparation. Each card on this page carries its status, so you know exactly what you get on day one.

Q04 How long from kickoff to production?

Typically 4–6 weeks: discovery call · pilot on 1–2 machines · plant rollout. Setter training runs in parallel with the pilot.

Q05 Can we customize checklists and workflows?

Everything is configurable in the admin surface. Checklists, fault categories, machines, molds, setpoints, roles, shifts — no source-code change when reality changes.

Q06 What happens when the network or the MES is down?

The shop floor keeps working. Records saved during a Wi-Fi outage wait in an offline queue for up to 24 h and are sent in order when the connection returns — no duplicates, no overwriting a colleague's newer entry. The MES is a data source, not a dependency.

Q07 Who runs it in the plant? Do you provide support?

You run it — the backend installs as a Windows service. We onboard your team, train setters and toolmakers, and ship improvement releases. Support tiers are discussed during discovery.

07 / HOW IT LANDS

From kickoff to production, in weeks

4–6 weeks
  1. 01 week 01

    Discovery call

    One 90-minute call. We map your machines, molds, MES endpoints and current workflow. You get a scoped proposal before we touch anything.

  2. 02 weeks 02–04

    Pilot deploy

    Full stack live on 1–2 machines in your plant. Setters and toolmakers are trained in parallel. Real cycles, real molds, real feedback loop.

  3. 03 weeks 05–06

    Plant rollout

    Remaining machines onboarded. Dashboard live. MES connection switched on where available. Handover to your internal admin.

08 / TALK TO US

See what this looks like
on your floor.

One short call, then a demo tailored to your machines and molds. No sales script.

Request a demo
or write directly to [email protected]