A Unique Software Platform
to Collect, Unify, Control, Analyze, Model, Interpret and Operate
Industrial Data

Know more


At, we spend our days creating the tools needed to run factories using data. Our platform makes it possible to deploy virtual agents, trained with Artificial Intelligence, to render factories more efficient and drive them towards autonomy. Here are the principles guiding our work.

Start with data

Factory data is often underused, or only very locally on a workstation or a specific production line. Machines have been used for decades. Their recent digitization and greater connectivity create a tremendous opportunity to shift paradigms. Collected in large quantities and linked to information steaming from operations’ context, data can be used to feed complex algorithms and train AI-agents. Once deployed, they assist plant operators in their daily tasks and perform repetitive tasks autonomously.

Complement with field expertise

In order to build an Artificial Intelligence that has a real impact on operations, analyzing data does not suffice. You have to go out in the field, talk to the operators and the technicians, get a feeling of how workstations and production lines are physically organized. Too often data scientists and R&D are cut off from operations because they do not have the tools to collaborate with data from industrial sites. However, in order to make the most of industrial data, it is necessary to be able to translate the expertise of the teams into algorithms.

Reconcile IT and OT

Deploying an AI on an industrial process does not have the same impact as predicting a click rate on an online ad. Special care must be taken to ensure the security of industrial computing (OT) by deporting certain processing at the edge of the network (edge ​​analytics). Thanks to these special provisions, maintaining control of the data is possible, while allowing us to deploy the intelligence as close as possible to your operations.


Centralize Your Data


Analyze Your Data


Deploy Analytics in your operations

Asset Framework

Ever had trouble identifying your data source?
FieldBox helps you identify where your data is coming from.

Digital Twins must be built from the ground-up: In order to genuinely represent what is happening in a factory, one must start at the equipment level.

Feed models with meaningful data with the FieldBox :

  • Map data streams to the right source,
  • Assign tags through a flexible and powerful built-in hierarchy,
  • Organize data according to equipments, production lines, plants, sites,
  • Appropriately define units of measure to avoid headaches and common mistakes encountered when converting data

Data Entry

Go paperless with FieldBox extensive Data Entry capabilities.

Before computers were deployed in factories, daily data was always written down manually. Having to report and re-transcript everything by hand is without contest a time-consuming and fallible activity.

Digitization is a foolproof way to provide immediate productivity gains! By reducing the time spent on manually calculating and reporting your data, your operations will be streamlined. When direct connectivity is not possible, we provide solution for bridging the gap.

  • Configure your reports to make sure operators capture your data in the appropriate format and unit of measure.
  • Inspection, Maintenance, Machine Parameter Statement… Take the operational context into account to design interfaces that are both specific to the industry and user-friendly,
  • Set up custom validation workflows so you can implement the four-eyes principle, and only distribute qualified data to strengthen your decision process

Connectors & API

FieldBox supports the main industrial data format and protocols and is continuously enriching its sets of interfaces to let you promptly build an intelligence layer on top of your equipment and operations.

It is designed as an open platform that is extensible at every layer of the stack. From low-level data integration, import pipeline customizations, to building custom user interface, a wide set of APIs and other tools allow you to import and export data and easily integrate with third party application.

Our sets of entry points include :

  • Connectors to PLC, DCS, SCADA, SQL and Historian databases,
  • Industrial Communication Protocols : Modbus, Ethernet IP, OPC UA/DA,
  • Audio and video acquisition : FieldBox Computer Vision, GigE, Camera Link,
  • Specific Interfaces : FieldBus, FieldBox Computer Vision,
  • FieldBox Data Acquisition Modules : Manual Data Entry for Operators,
  • Import of documents in Standard Format : Excel, CSV, Words.

From the Factory to the Cloud

Smart data to Big Data, is built to scale to industry wide operations. How much do you want to get?

FieldBox typical setups include 1000-5000 tags per asset, which can lead your configuration in the hundreds of thousands of tags.

These core technologies are key to be able to gracefully handle various datasets, whose size may vary from small to very large. FieldBox can manage low frequency data or high frequency data, structured information or unstructured information. Connectors embed built-in filtering and auto-diagnosis capabilities that can be configured remotely.

These native features allow you to run complex algorithms smoothly, with a 100% reliability.



Integrating new AI-based technologies in the very specific industrial IT environment can be long and complex. It shouldn’t be. FieldBox Analytics is a complete suite for analyzing production and operations data and creating AI-agents to assist operators and managers in their daily tasks. FieldBox Analytics lays out the operational ground for deploying advanced algorithmics and lets you take control over your analytics through three pillars.


At the heart of FieldBox Industrial Analytics runs an engine that facilitates the training of model.

It can run on multiple CPUs, GPUs, and a wide range of machine learning and deep learning algorithms, as well as other data science routines.


From the outset, we’ve always wanted to give power to the users. We are therefore offering full access to the algorithms, that can be written in Python or R in the data studio.

The editor integrates a set of tools to ensure maximum efficiency in writing the models


An AI agent is composed of a datascience model – trained in the FieldBox or in a third party Studio – and all the FieldBox tools to leverage the Python/R code developed: data flow from live sources, robust code runners, custom web UI, third-party views and industrial actioners.



Anomaly Detection

In industrial operations, anomaly detection is key in a number of cases, from detecting sensor errors to identifying faulty production lots. Detecting outliers is part of the standard routine performed when digging into a set of industrial data, that may not have the quality required to train machine learning models.

There are several reasons why one may want to change machine parameters :

  • Take the the quality of an input product into account,
  • Adjust to operating conditions (think temperature, pressure, hygrometry),
  • Reflect an evolving condition of the machine itself.

Computer Vision

Modern vision systems are often a heterogeneous collection of interpretation layers.

We apply numerical techniques to transform images into meaningful signals. :

  • Image processing
  • Machine learning,
  • Pattern recognition

The FieldBox platform is built to integrate videos with other contextual information to find patterns that can not be detected through sensor data only.

Machine Parameter Adjustment

As opposed to human operators, virtual agents can constantly interpret the inflow of data to adjust machine parameters depending on operating conditions in real-time.

In our quest to give birth to autonomous manufacturing, adjusting machine parameters based on algorithms is key. There are several reasons why one may want to change machine parameters :

  • Take the the quality of an input product into account,
  • Adjust to operating conditions (think temperature, pressure, hygrometry),
  • Reflect an evolving condition of the machine itself.

Avoid errors, save time compromise and reach operational excellence by automating these operations using AI.

Quality Control Automation

Make sure to reach the desired quality standards with our Quality Control Automation Agent.

Quality inspection often relies on human senses, to check if the production output « looks » good and fits right. There are often hidden criteria that guide his judgment and need to be captured to help him better perform this task. The challenge in automating Quality Control is to put in place numerical measures of these feelings.

The benefits for Quality Control Automation are huge :

  • Reduce stock by processing your output in real time,
  • Reduce sampling rate,
  • Avoid over quality and costly product recall: your shipments always correspond to clients needs.

Predictive Maintenance

In industrial operations, stakes are often too high to allow for failure. As a result, maintenance policy tends to be very conservative, inducing additional costs. Maximize uptime and equipment efficiency with AI.

Our Predictive Maintenance Agent combines all available operations informations to recommend optimal maintenance intervention, namely :

  • Load forecasting for better planning of downtime
  • Prediction of failure to optimize when replacement pieces should be ordered and installed.
  • Scoring of Maintenance operations to qualify operations that are too often limited to ticking a checkbox, to say the job is done.

Virtual Metering

With FieldBox Virtual Metering Agent, you can use process conditions to estimate flow rates instead of using a physical meter. Virtual Meters are implemented :

  • To avoid costs of installation and maintenance associated with physical meters,
  • To save space, in particular in physically constrained environment (offshore platforms, underground facilities…).

Our Virtual Meter Agent is a complex combination of physical and statistical approaches that is self-calibrating and easy to implement.



A rich set of applications to leverage your modelling effort.

Native applications offer the set of features that are the most common to industrial use cases. They are also leveraged to display and analyze results from tailor-made algorithms in the office or on the production line.

They include

  • Monitoring: a standard applications for visualizing data streams connected to the FieldBox, and easily comparing them.
  • Reports: publish results in .csv, PDF, and MS Office formats, or to third party systems you can easily interface with such as your accounting and ERP systems.
  • Alerts: depending on the severity of issues, receive e-mails or SMS to be made aware of critical situation as they arise.

On Premise or in the Cloud

FieldBox‘s flexible architecture lets you choose the most appropriate setup so you can get the most out of your data.

Many use cases require code to run inside a private network, behind complex firewall rules. Many people favour cloud setup as they provide quicker turnarounds. We want to provide our customers the best of the two worlds and are therefore relying on a flexible architecture that relies on virtualization and containers.

FieldBox is bullt to accommodate setups that can be hybrid and versatile. A typical project may start with a cloud installation, before it migrates to private networks within our customers industrial sites.

Permissions and Security

FieldBox provides government grade security through sophisticated permissions at every level.

Get the right level of security through multi-layer security mechanisms. From multi-factor identification to managing data access at low levels, users and administrators have access to detailed access control so that the appropriate rules can be defined.

In order to facilitate communication within teams and between them, establishing one source of truth is key. A first step is to make data is accessible to the people who need it. A second step is to make sharing analysis and results easy.

  • Notebooks, dashboards can be shared directly to teammates and other parts of the organization.
  • Reports and configuration can be assigned for a group of user.

Audit Trail

Monitoring activity is critical to deploying efficient software solutions.

Every exception anywhere in the platform is captured, including errors from algorithms defined by users. FieldBox also registers every API call showing what user and API key are used on a given call. Exceptions can be configured to capture error message, stack trace, and partial or full input.

Admins can configure the logging module to capture more or less information depending on the use case.


I authorize to inform me about its offers via the contact details collected on the form. (required)

I authorize to inform me about its offers via the contact details collected on the form. (required)

I authorize to inform me about its offers via the contact details collected on the form. (required)