Sensing
Multi-modal PAT fusion
Raman, FBRM, PVM in-situ imaging, and NIR combined into one live state estimate of the reaction mass and slurry — with disagreement between instruments treated as signal, not noise.
Features
Everything needed to control a crystal in real time.
Sensing, prediction, control, simulation, and record-keeping — designed together, because in API manufacturing they are the same problem.
Core capabilities
Built for the physics of drug substance
Not a generic industrial AI platform with a pharma skin.
Prediction
Polymorph prediction
In-line form identification and transition risk, with confidence bounds and cited spectra.
Prediction
Particle size & habit
Live D10/D50/D90 and habit classification from chord-length and image data.
Control
Adaptive profiles
Cooling, antisolvent, and seeding profiles that adapt per batch rather than replaying a fixed ramp.
Prediction
Impurity trajectory
Chemical, chiral, and genotoxic impurity trends predicted before offline HPLC confirms them.
Simulation
Digital twin
Reactor and crystallizer simulation for profile design, scale-up, and tech transfer.
Quality
Review by exception
Batch records that surface deviations instead of demanding a page-by-page read.
Sensing
Four instruments, one truth
Each PAT modality sees a different part of the crystallization. Fusing them is what turns four trend lines into a decision.
- Raman: chemical composition and polymorphic form
- FBRM: nucleation, growth, and attrition dynamics
- PVM: habit, agglomeration, and oiling-out
- NIR + telemetry: solvent, moisture, and thermal state
Plant-edge console
One screen for the batch that is running right now
Supersaturation, particle size distribution, predicted polymorph, impurity trajectory, and the next control move — with the evidence behind each number.
Polymorra plant-edge console showing live supersaturation, particle size distribution, predicted polymorphic form, and impurity trajectory for a running crystallization, with the recommended next control move.
Control
Adaptive profiles instead of fixed ramps
A recipe assumes every batch is the same. Polymorra assumes the opposite — measuring supersaturation and particle dynamics in-run, then adjusting cooling rate, antisolvent addition, seeding, and agitation to keep the batch inside the metastable zone.
- Setpoint moves bounded by validated control windows
- Instant handback to the chemist at any moment
- Every move logged with the evidence that triggered it
Simulation
A twin that knows your crystallizer
Geometry, agitation, heat transfer, and solubility behaviour for your equipment — so a profile designed in simulation survives contact with the plant, and a lab-to-plant transfer starts from a prediction instead of a guess.
- Scale-up prediction from lab to pilot to plant
- Rare-fault synthesis for training and operator drills [ASPIRATIONAL]
- What-if analysis on solvent, seed load, and cooling rate
Quality
The record writes itself
Every perception, recommendation, approval, correction, and control action lands in an immutable, Part 11-ready audit trail. Review by exception then surfaces only what deviated — with the underlying spectra and images one click away.
- E-signature-ready review flows
- Deviation clustering across a campaign
- Exportable evidence packages for inspection
Accelerated computing
An NVIDIA-accelerated stack, from the train to the fleet
Perception at the edge, reasoning in the cloud, and a reactor-and-crystallizer twin in between.
Stack
Jetson at the train
Plant-edge inference per reactor or crystallizer train: PVM crystal-image models, FBRM feature extraction, Raman and NIR soft sensors, anomaly and safety classifiers. [ASPIRATIONAL]
Stack
TensorRT + Holoscan
Low-latency perception and sensor-stream preprocessing for in-situ imaging and PAT spectra inside validated control windows.
Stack
Triton + NIM
Route-specific polymorph, impurity, particle-size, endpoint, and batch-reasoning models served with validated version locks and rollback.
Stack
Omniverse + Cosmos
Reactor-and-crystallizer twin plus synthesis of rare crystallization faults — oiling-out, fouling, seed drift, wrong-polymorph nucleation. [ASPIRATIONAL]
Stack
RAPIDS + cuOpt
Telemetry ETL at campaign scale, plus scheduling, solvent recovery, and crystallization-profile optimization.
Stack
NeMo + DGX
Fine-tuned chemistry-process and batch-record reasoning trained on de-identified design-partner data and chemist corrections. [ASPIRATIONAL]
Integrations
Connects to the systems already running your plant
Polymorra reads and writes through the reactor DCS, PAT instruments, crystallizer and isolation skids, historians, and MES you already validated.
Reactor control & DCS
Emerson DeltaVSiemens PCS 7Honeywell ExperionOPC UAModbus/TCP
PAT & analytics
Mettler-Toledo FBRMPVM in-situ imagingReactIREndress+Hauser RamanNIRHPLC / XRPD imports
Isolation & drying
Nutsche filter-dryersCentrifugesAgitated dryersConical millsMicronizers
MES, historian & quality
Körber PAS-XEmerson SyncadeOSIsoft PILIMSeBR / QMS exports
Trust & compliance
Validated for GMP manufacturing
Polymorra is engineered for regulated drug-substance production: grounded outputs, immutable audit trails, graduated autonomy, and validation documentation from day one.
21 CFR Part 11
Immutable, validated audit log for every agent perception, recommendation, and control action, with e-signature-ready review flows.
GMP / ICH Q7
Change control, validated model versioning, and deployment documentation designed for drug-substance manufacturing.
SOC 2
SOC 2 Type I in progress, Type II on the roadmap; SSO/RBAC, encryption in transit and at rest. [ASPIRATIONAL]
Route & IP protection
Per-tenant isolation of routes, recipes, spectra, and crystal images. No cross-tenant training. On-prem option.
Graduated autonomy
Shadow → assist → bounded auto-control. Every autonomy level is explicitly configured, bounded, and revocable.
Grounded outputs
Every recommendation cites the spectra, images, telemetry, SOP, or specification it was derived from.
Plant-edge first
Inference runs on-site so control loops survive network loss; cloud is for training, fleet, and analytics.
Human-in-the-loop
Chemist and QA checkpoints are first-class: approve, correct, or reject — and every correction trains the models.
FAQ
Frequently asked questions
Both — in a deliberate order. Polymorra starts in shadow mode (perceive only), moves to assist (recommend, chemist approves), then to bounded auto-control on qualified trains where setpoint moves stay inside validated control windows. Every level is configured per train and revocable.
Polymorra is built for GMP (ICH Q7) and 21 CFR Part 11: immutable audit trail, validated model version locks with rollback, change control, grounded and citable outputs, and human-in-the-loop checkpoints. We ship a validation package and support IQ/OQ/PQ activities with your quality unit.
API or OPC access to your reactor DCS, PAT instruments (Raman, FBRM, PVM, NIR), crystallizer and isolation skids, and MES or historian — plus a train with enough campaign volume for clear ROI.
No. Per-tenant isolation of routes, recipes, spectra, and crystal images is enforced at the data and model layer. Cross-site benchmarks are opt-in and de-identified.
A typical wedge pilot instruments one train, runs shadow mode over a baseline campaign, then moves to assist mode with a defined polymorph, yield, or reprocessing success metric agreed up front.
Perception and control inference run at the plant edge. Cloud handles training, fleet management, and analytics — losing it degrades reporting, not the batch.
Feature lists are cheap. Batches are not.
Start with one crystallizer or reactor train. Prove polymorph, yield, and reprocessing ROI in a validation-friendly pilot. Then expand across the plant.