Serumon BOOK A PILOT

STAGE 05 // SIMULATE

Twynex
runs it first.

Serumon’s simulation and optimisation brain: an Omniverse-based make-and-fill digital twin that simulates compounding, emulsification, shade and filling to hit target spec before the batch, plus a GPU-accelerated optimiser for campaign sequencing, changeover, cleaning validation, filler balancing and raw-material allocation.

BOOK A PILOTSEE PRICING
PRE-BATCHSPEC PREDICTION
cuOPTCAMPAIGN SOLVER
OMNIVERSETWIN PLATFORM
DGX + OVXGPU PROFILE
twynex — campaign W41 optimiser 
$ twynex optimise --week 41 --site MILAN

SKUs 34  |  VESSELS 6  |  FILLERS 4
CONSTRAINTS  allergen seq, colour seq, CIP validity

BASELINE SEQUENCE   changeover 61.5 h
OPTIMISED SEQUENCE  changeover 44.2 h  (-28%)
CLEAN CYCLES        22 -> 17
FILLER BALANCE      utilisation 71% -> 84%

PRE-BATCH SIM  SPF50 rev14 on V-07
  predicted visc 41.2k  IN SPEC
  predicted dE   0.38   IN SPEC
  recommend homogenise 3160 rpm / 11 min
MAKE-AND-FILL TWINPRE-BATCH SPEC PREDICTIONCAMPAIGN OPTIMISATIONNPI SANDBOXBIDIRECTIONAL SYNCSYNTHETIC SCENARIOS

THE PROBLEM

Scale-up is trial and error paid for in pigment

Short-run SKU proliferation and constant launches explode changeovers and cleaning validation, while off-spec first batches waste expensive pigments, actives and packaging.

Plants cannot cheaply answer the only question that matters before committing raw materials: will this formula, on this vessel, hit target viscosity, stability and shade — and how should the campaign be sequenced to minimise changeover and giveaway?

So scale-up and new-product introduction stay trial and error, and campaign scheduling stays a spreadsheet.

WHAT THE TWIN DE-RISKS

FIRST-BATCH SUCCESS
OFTEN <60%
NPI TRIALS
3–8 BATCHES
CHANGEOVER / WEEK
40–70 h
CLEANING VALIDATION
MANUAL
SCHEDULING
SPREADSHEET

ILLUSTRATIVE SITE FIGURES [ASPIRATIONAL UNTIL MEASURED].

CAPABILITIES

What Twynex does

MAKE-AND-FILL TWIN

Simulates compounding, emulsification, shade and filling on the specific vessel and line the batch will actually run on.

PRE-BATCH SPEC PREDICTION

Predicts viscosity, particle size, stability and ΔE before raw materials are committed, and proposes the process to hit them.

CAMPAIGN OPTIMISATION

cuOpt-based sequencing across changeover, cleaning validation, filler balancing and raw-material allocation.

NPI SANDBOX

Scale-up and new-product introduction explored in simulation before a vessel is booked.

BIDIRECTIONAL SYNC

Validated recipes and schedules push to Emulson and Fillix; real outcomes flow back and correct the twin.

SYNTHETIC SCENARIOS

Rare fault scenarios generated to stress-test control policies without breaking real product.

THE LOOP

Four beats, every batch

  1. 01

    MIRROR

    The twin is built from real vessel geometry, line configuration and the process traces Emulson and Fillix have accumulated.

  2. 02

    SIMULATE

    Candidate formulas and processes are run against physics-informed surrogates for emulsification, rheology and shade.

  3. 03

    OPTIMISE

    cuOpt sequences the campaign under allergen, colour, CIP and due-date constraints, balancing fillers and allocation.

  4. 04

    PUSH

    The validated recipe and schedule go to the plant; the resulting batch outcome is fed back to reduce twin error.

AI CAPABILITIES

What is real today, and what is not

Serumon labels its own maturity. Anything not yet validated in production is marked aspirational, on this site and in the room.

REAL / FULL FIDELITY ASPIRATIONAL

PINN SURROGATES

Physics-informed emulsification, rheology and shade surrogates. A full-fidelity twin is a multi-year build, not a launch claim.

REAL

SHARED WORLD MODEL

The same learned process world model Emulson uses, so the twin and the controller do not disagree about physics.

REAL

CONSTRAINED OPTIMISATION

cuOpt over campaign, changeover and allocation constraints at plant scale in interactive time.

REAL

SCENARIO GENERATION

Cosmos and Replicator-based synthetic scenarios for rare fault coverage.

NVIDIA ALIGNMENT

GPU is must-have, not nice-to-have

DGX + OVX

Training and Omniverse simulation workloads for twin fidelity at plant scale.

OMNIVERSE

The make-and-fill twin environment: vessels, lines, robot cells and material flow.

MODULUS

Physics-informed surrogates that make emulsification simulation interactive rather than overnight.

cuOPT

GPU-accelerated constrained optimisation for campaign sequencing and allocation.

MEASUREMENT

What we get judged on

FIRST-BATCH HITPRIMARY METRIC
CHANGEOVER HSECONDARY METRIC
OEETHIRD METRIC
$9k/moLINE TIER PRICE

PILOTS ARE CONTRACTED AGAINST A NAMED ROI METRIC AND A BASELINE WE RECORD BEFORE ANYTHING CHANGES.

THE MOAT

A twin is only as good as the traces behind it

Simulation vendors sell environments. The hard part is calibration: a twin that has never watched ten thousand real batches on your vessels is a rendering, not a prediction.

Twynex is fed continuously by Emulson traces, Shadera spectra, Fillix telemetry and Stabion outcomes. Its error shrinks with every batch the plant runs, which is exactly the asset a standalone simulation product cannot assemble.

  • Twin calibrated on your vessels, not a generic model.
  • Error corrected batch by batch against real outcomes.
  • Campaign constraints learned from actual changeovers.
  • Shared world model with the live controller.

DEPLOYMENT SHAPE

TIER
LINE / PLANT / ENT
LINE PRICE
$9,000 / mo
PLANT PRICE
$60,000 / mo
ENTERPRISE
$400k–$5M ACV
EDGE NODE
PER VESSEL CLUSTER
AUTONOMY
GRADUATED

ECOSYSTEM

How TWYNEX plugs into the loop

Every agent hands the next one a better decision. Buy one; the others make it sharper.

TWYNEX FAQ

Straight answers

WEDGE PILOT

Run Twynex on one line

Instrument, baseline, shadow, then assist — against a metric you name before we start. If the numbers do not move, you have lost a quarter of telemetry and gained a very detailed picture of your own process.

BOOK A PILOTBACK TO PRODUCTS