The industrial MLOps platform
AI closed-loop automation at industrial scale
Connect your production data, upload your own model or build one, and deploy it on every asset you run, in the cloud or at the edge. Wizata turns models into set points your control system applies.
Built for process manufacturing: cement, lime, steel & metals, mining, chemicals & pharma, food & beverage, oil & gas.
Trusted by
How it works
From your assets to an optimised process
Most industrial AI stops at a dashboard. Wizata runs the whole chain in one platform: the same platform that reads your controllers writes the optimised values back to them.
Connect
Connect your data: controllers, historians, OT networks and cloud streams, through Wizata Edge on site or directly in the cloud.
Organise
Every signal lands in a digital twin of your plant, labelled by asset, unit and category, with batches and events alongside.
Build or bring
Upload a model your team already trained, build one in a visual pipeline, or start from the built-in library.
Deploy
Ship the pipeline to production on one asset or many, and choose cloud or edge for each one.
Operate
Run in open loop with recommendations, or in closed loop with set points sent straight to the control system.
Three modules, one platform
One place for the process team, the data scientists and IT
An industrial MLOps platform: data, models, deployment and operation share the same digital twin, so a model that works on one asset is ready for the next.
Your plant as a digital twin, connected from the edge
A unified namespace for every signal, asset and batch, and the place where you configure and control every edge connection.
- Digital twin hierarchy from plant to line to asset, with data points attached where they belong.
- Edge connections configured and monitored in one place, with live status in the mobile app.
- Events and batches as first-class data: query by batch ID, compare durations, preview before you run.
- Data Explorer to find anomalies and patterns without writing code.
For process engineers · data scientists · OT
Bring your model, or build one. Run it everywhere.
Many teams arrive with models already built in R&D. The AI Lab takes them to production: upload, wrap in a pipeline, version, deploy.
- Upload your own model trained anywhere, or use the built-in library: isolation forest, autoencoder, Hotelling T², Mahalanobis, gradient boosting and more.
- Pipelines: Query → Script → Model → Write, Plot or Alert, on a canvas your whole team can read, on top of real Python.
- Experiments and versioned models so every production result can be traced back to its training run.
- Pipelines in your own Git: sync pipelines and scripts as plain JSON and Python files to GitHub, GitLab, Azure DevOps or Bitbucket.
For data scientists · digitalisation teamsLearn more in the docs →
Operate in open or closed loop
The Control Panel is where you run the system day to day, on the desktop and in the Wizata mobile app.
- Open loop: operators receive recommended set points and decide.
- Closed loop: set points go straight back to the control system.
- Anomaly detection and alerting on push, email, SMS, WhatsApp, Slack and Teams, grouped into incidents so nobody gets spammed.
- Dashboards per asset, with live KPIs and alert states for every line and machine.
For operators · plant managers · process engineers
Scale
Pilot on one asset. Roll out to many.
A template describes a type of asset once. A pipeline built on a template runs on every matching asset, from a handful of kilns to thousands of rolling-mill stands, and the deploy wizard sets cloud or edge for each one.
Every new asset means a new project, a new model and more maintenance.
One pipeline for every similar asset, replicated in about a week.
Open by design
Your data, your models, your IP
You own what you build
All data, models and intellectual property developed on Wizata belong to you. Wizata does not access or use your data for any other purpose.
Open architecture
You get access to the underlying resources of the platform, not only to its interface.
Secure, compliant and explainable
Data stays in protected environments, on your own infrastructure or in a dedicated Wizata tenant, and complies with GDPR. Models are explainable, so your teams understand every recommendation.
import wizata_dsapi # Last 24 h of kiln data, 1-minute means df = wizata_dsapi.api().query( datapoints=["kiln1_burning_zone_temp", "kiln1_o2", "kiln1_fuel_rate"], start="now-1d", end="now", agg_method="mean", interval=60000 ) # Same data the pipeline editor sees: explore in Jupyter, # then deploy the script as a pipeline step.
Integrations
Works with the infrastructure you already run
Independent of any machine brand or cloud. Wizata reads from your control layer, works with your data science tools and shares results where people already look.
Connect your data
Write set points back over OPC UA for closed-loop control.
Work with your tools
Bring existing notebooks, libraries and models. Nothing has to be rebuilt.
Share your results
Alerts reach people where they are, and results flow into the BI tools your managers already use. Native iOS and Android app included.
Licensing
One platform, three licences
Every licence covers the full workflow from data to model. Higher tiers add edge, dedicated resources and deployment in your own cloud.
Standard
Cloud- Cloud connectivity
- Digital twin and Data Explorer
- Pipelines, triggers and simulation
- Model storage and training
- Custom Python code
- Dashboards
- Users and permissions
- Custom integrations
Pro
Cloud + edge- Everything in Standard
- Edge connectivity
- Runners management
- Custom Python libraries
- Dedicated resources
- Git repository sync
- Closed-loop automation on edgeAdd-on
Enterprise
Your cloud + edge- Everything in Pro
- Development environment
- Custom deployment in your cloud
- Closed-loop automation on edgeAdd-on
Results in production
Measured on real lines, then scaled
on several continents running autonomously with limited human intervention.
Read the lime case →“We decided to work with Wizata because they have a better business knowledge of process manufacturing than the competition and their collaborative approach allows us to stay in control of our IP.”Head of Operations Support, Carmeuse
Get started
Start with one asset. Prove the value. Scale.
Talk to one of our engineers about your process, or spend one or two days with our team on your own data. We only propose a pilot when we can see the return.
- Free
Free assessment workshop
One or two days with our team: your goal, your data, the expected impact.
- 3–6 months*
Pilot
One asset connected, one model in production, ROI measured.
- ~1 week per asset
Scale
The same pipeline replicated to every similar asset.