PLCs, sensors, meters and SCADA in one model, on one clock. No more stitching spreadsheets together.
Without replacing equipment
It integrates with the SCADA, ERP and MES you already run. It reads from them and adds the common layer they lack.
Data even when the network drops
Captia Connect buffers at the edge and synchronises once the connection is back. Acquisition does not depend on the cloud.
Who we work with
A real case, with altered figures. Nine dry contacts on eight machines, two weeks of baseline and the analysis that found the real constraint where the availability percentage misled everyone.
A plant does not suffer from a lack of data: it suffers because the data lives in systems that do not share a model, a clock or a history.
A plant generates data everywhere: PLCs, sensors, machines, meters, SCADA, ERP and MES. Each system keeps its own, in its own format, in its own history, and nobody can see the whole plant without exporting spreadsheets and stitching them together by hand.
An industrial data platform solves exactly that. It is the software layer that captures data at source, normalises it into a common time series model and makes it available to the whole organisation: visualisation, rules and alerts, workflows, reporting, APIs and artificial intelligence. It is also the foundation of industrial plant monitoring: without that common layer, each system shows its own fragment and nobody sees the whole. It does not replace existing systems: it connects them and adds the common layer they lack.
This page describes the Captia industrial data platform as a product. If what you need is the concept in the abstract, before looking at any tool, the reference guide is what an industrial data platform is and how it is built: layers, architecture decisions and selection criteria, with no product in the way.
Data is born on the floor, Connect captures and normalises it, and Captia.ai turns it into operations.
Where it leads
The foundation of physical AI
Physical AI does not start with the robot. It starts with the data from the machine you already have.
Physical AI is AI that perceives, reasons and acts on the real world. In a factory it starts by capturing the data from the equipment that already exists. Robots, autonomous vision and agents acting on the plant floor are where the sector is heading, and none of them works without reliable machine data, with a timestamp and context, that many plants still do not have.
The Physical AI Ladder orders that path into five levels, defined by what the plant can do with its data rather than by the technology it buys. The platform covers levels 1 to 3: Captia Connect connects the equipment and normalises the data, Captia.ai turns it into OEE, energy, downtime and quality in real time, and workflows, Energy and Service close the loop under supervision. Level 4 belongs to the sector and is only possible once the first three are solved.
The Physical AI Ladder. Captia covers levels 1 to 3. Level 4, drawn as an outline with no fill, describes the state of the art of the sector.
Separating acquisition from exploitation is not a preference: it is what lets the plant keep capturing data even when the cloud is not available.
Plant
Machines
PLC
Sensors
Meters
SCADA
Captia ConnectEdge
Multi-protocol acquisition
Normalisation
InfluxDB · local buffer
Secure transport
Outbound connection
No open ports on the floor
Captia.aiPlatform
Web SCADA
Dashboards
Rules and alerts
Workflows
Reporting
APIs
Operations
Production
Maintenance
Energy
ERP · MES
How to read the diagram: plant sources feed Captia Connect at the edge, secure transport carries the data to Captia.ai and the platform returns it as industrial operations.
1
Captia Connect: the edge layer
Deployed on the plant floor, next to the equipment. It acquires data from PLCs, sensors, machines, meters and SCADA systems over multiple protocols, reads the signals, normalises them into time series and buffers them locally with persistence without connectivity: if the network drops, no data is lost.
2
Secure transport
Normalised data travels from the edge to the platform via secure delivery. Acquisition on the floor does not depend on cloud availability: Connect keeps capturing and synchronises once the connection is restored.
3
Captia.ai: the platform
The layer where data becomes operations: web SCADA, dashboards by asset, line, role or plant, history, rules, alerts and events, workflows, reporting, APIs, user and role management, an Operator Layer and integration with ERP, MES, CRM and SAP. On that foundation run the artificial intelligence modules.
The data infrastructure is the territory of Captia Connect. The analytics, industrial models and AI modules that exploit that data are the territory of our industrial artificial intelligence unit.
Sizing
Size your data platform
You do not size a platform by number of machines but by data points: signals times frequency times time.
Enter how many assets and signals you want to connect and how often. The result is transparent arithmetic over your inputs: data points per day and which capabilities of each layer your scenario would activate. The calculation is illustrative, based on your own numbers.
Platform sizer
Estimated data points per day
172,800
120 signals x 1,440 readings/day = 172,800 points/day
Illustrative calculation over your own numbers; it is not a product metric.
Layers and capabilities your scenario would activate
Captia Connect (edge)
Multi-protocol acquisition and normalisation into time series
PLC integration
Sensor integration
Secure transport
Secure delivery from the edge to the platform
Captia.ai (platform)
Web SCADA, dashboards and history over the connected signals
The diagnosis turns this scenario into a real inventory of signals and phases.
A protocol is not a checkbox in a table: each one implies a different topology, data model and polling strategy. Connect implements all nine.
Alongside the nine protocols sits the stack the edge layer runs on: Docker for deployment, InfluxDB for time series and Tailscale for secure connectivity to the platform.
Lightweight pub/sub protocol over TCP designed for telemetry: central broker, hierarchical topics and QoS levels 0 to 2.
Where it is used
IIoT sensors, gateways and devices that publish their readings to a broker.
How Connect handles it
Connect acts as a client that subscribes to the relevant topics, normalises the payloads into time series and applies local buffering if the broker or the network uplink goes down.
Lightweight pub/sub protocol over TCP designed for telemetry: central broker, hierarchical topics and QoS levels 0 to 2.
Where it is used
IIoT sensors, gateways and devices that publish their readings to a broker.
How Connect handles it
Connect acts as a client that subscribes to the relevant topics, normalises the payloads into time series and applies local buffering if the broker or the network uplink goes down.
The multi-drop serial variant of the same Modbus protocol, over an RS-485 bus.
Where it is used
Legacy equipment and existing field buses without an Ethernet connection.
How Connect handles it
Connect attaches to the serial bus, manages slave addressing and bus timing, and lifts the readings into the same time series model as every other source.
If your equipment speaks another protocol, we evaluate it during the diagnosis. The full acquisition layer is explained in Captia Connect and in the data acquisition system guide.
Artificial intelligence
Captia.ai AI modules
Models do not run on promises: they run on a consolidated, normalised history. That is why AI is the last layer of the architecture, not the first.
The plant network should not have to expose ports to the internet to get a data platform: the edge initiates secure outbound connections and acquisition does not depend on the cloud.
Secure connectivity, no inbound ports
Outbound connection from the edge
How to read the diagram: the OT network is never exposed to the internet; Connect initiates the outbound Tailscale connection and, if the network drops, capture continues against the local buffer.
Secure delivery
Connectivity between the edge and the platform is established with Tailscale: Connect initiates encrypted outbound connections and the plant network opens no inbound ports.
Persistence without connectivity
Connect stores time series in a local buffer on InfluxDB: if the network link drops, capture does not stop and the data syncs once the connection returns.
Users and roles
Captia.ai manages access with users and roles: each person sees the dashboards, assets and functions that match their role in the organisation.
Deployment models: on premises, hybrid or cloud
The same secure route works in all three models: what changes is where the exploitation layer lives.
On premises
The whole data layer lives at the plant. The usual choice when company policy requires that data never leaves the premises.
Reference
Hybrid
Acquisition and buffering at the edge, exploitation on the platform. This is the reference model: the plant keeps its autonomy and the organisation gains centralised web access.
Cloud
The full platform in the cloud, with the edge as the capture point. It simplifies IT operations when there are no local data residency requirements.
Deployment
How it is deployed: project phases
A data project is not deployed in one go: it is deployed in phases that deliver value from the first connected signal.
1
Diagnosis
Inventory of data sources (PLCs, sensors, meters, SCADA, ERP, MES), available protocols and the plant network architecture. This defines the scope of the first phase.
2
Edge deployment
Captia Connect is installed on the floor and the first signals are connected: acquisition, normalisation, buffering and secure delivery up and running.
3
Platform go-live
Captia.ai is configured: web SCADA, dashboards by asset, line, role or plant, users and roles, and the first rules and alerts.
4
Operationalisation
Workflows, reporting, APIs and integration with ERP, MES, CRM or SAP, so plant data reaches management without manual steps.
5
Intelligence
With a consolidated history, the AI modules are activated: energy optimisation, anomaly detection, predictive maintenance, forecasting, demand prediction, recommendations and optimisation of production and consumption.
Comparison
Data platform vs MES vs traditional SCADA
The three concepts are often confused. This comparison is conceptual: it describes the role of each type of system, not specific third-party products.
Dimension
Industrial data platform
MES
Traditional SCADA
Primary purpose
Common data layer for the whole plant: capture, normalisation and exploitation
Manage production execution (orders, routings, times)
Supervise and control specific processes in real time
Data scope
Cross-cutting: machines, sensors, meters, SCADA, ERP and MES
Focused on the production process and its orders
Focused on the equipment and signals it supervises
Access
Web based, from the browser, with users and roles
Typically client workstations on the floor and in the office
Typically local operator stations
Analytics and AI
History, reporting and AI modules across the whole plant
Production indicators within its own scope
Trends and history of its own signals
Integration
Built to integrate: multi-protocol on the floor and APIs towards ERP, MES, CRM or SAP
Typically integrates with the ERP and with the data platform
Becomes one more data source for the platform
How they relate
Does not replace MES or SCADA: it connects them and adds a common layer
Coexists with the platform as the execution system
Coexists with the platform as the control system
Industrial data platform
The common layer
Primary purpose
Common data layer for the whole plant: capture, normalisation and exploitation
Data scope
Cross-cutting: machines, sensors, meters, SCADA, ERP and MES
Access
Web based, from the browser, with users and roles
Analytics and AI
History, reporting and AI modules across the whole plant
Integration
Built to integrate: multi-protocol on the floor and APIs towards ERP, MES, CRM or SAP
How they relate
Does not replace MES or SCADA: it connects them and adds a common layer
MES
Primary purpose
Manage production execution (orders, routings, times)
Data scope
Focused on the production process and its orders
Access
Typically client workstations on the floor and in the office
Analytics and AI
Production indicators within its own scope
Integration
Typically integrates with the ERP and with the data platform
How they relate
Coexists with the platform as the execution system
Traditional SCADA
Primary purpose
Supervise and control specific processes in real time
Data scope
Focused on the equipment and signals it supervises
It is the software layer that captures data from plant equipment (PLCs, sensors, machines, meters, SCADA), normalises it into a common model and makes it available to the operation: visualisation, rules, alerts, workflows, reporting and integration with management systems. It replaces a collection of silos (one historian per machine, spreadsheets, systems that do not talk to each other) with a single data layer the whole factory works on.
What is the difference between Captia Connect and Captia.ai?
Captia Connect is the edge layer: it is deployed next to the equipment, acquires data over multiple protocols, normalises it, buffers it locally and sends it securely. Captia.ai is the platform: web SCADA, dashboards, rules, alerts, workflows, reporting, APIs and artificial intelligence modules. Connect solves the data; Captia.ai turns it into operations and intelligence.
Which industrial protocols does Captia Connect support?
MQTT, OPC UA, Modbus TCP, Modbus RTU over RS-485, OpenWebNet, IEC 870-5-102, REST APIs, webhooks and CSV files. This allows PLCs, sensors, meters and existing SCADA systems to be integrated without replacing plant equipment.
What happens if the plant loses its internet connection?
Captia Connect includes local buffering and persistence without connectivity: data keeps being captured and stored at the edge, and is sent once the connection is restored. Acquisition does not depend on cloud availability.
How is the connection between the plant and the platform secured?
Delivery from the edge to the platform uses Tailscale: Captia Connect initiates secure outbound connections, without opening inbound ports on the plant network. If connectivity drops, local buffering at the edge preserves the data and synchronises it once the connection is restored.
Which artificial intelligence modules does Captia.ai include?
Seven modules: Energy Optimization, anomaly detection, predictive maintenance, forecasting, demand prediction, recommendations and optimisation of production and consumption. All of them work on the consolidated, normalised history captured by Captia Connect.
Is this platform physical AI?
No. The platform is not a robot or an autonomous vision system. It is the data foundation any physical AI in a factory rests on: physical AI is AI that perceives, reasons and acts on the real world, and in a plant it starts by capturing the data from the equipment that already exists. Captia Connect and Captia.ai cover levels 1 to 3 of the Physical AI Ladder (Connected, Aware, Acting). Level 4, robots, vision and autonomous agents, is where the sector is heading and is only possible once the first three are solved.
Does a data platform replace my SCADA or my MES?
Not necessarily. The platform integrates with existing systems: it can read from the current SCADA and integrate with ERP and MES. What it adds is the common layer those systems do not provide on their own: a unified data model, web visualisation, cross-cutting rules and alerts, workflows, reporting and AI modules over the full plant history.
How is it deployed and how soon do you see data?
The project starts with a diagnosis of the data sources and the plant network architecture. Captia Connect is then deployed at the edge and the first signals are connected, becoming visible in Captia.ai. From there the scope grows in phases: more signals, rules, alerts, workflows, reporting and ERP or MES integration according to business priority.
How many signals can the platform capture?
It depends on the plant sources and the sampling frequency: reading 50 signals every minute is not the same as reading 500 every second. The volume is sized during the diagnosis, based on the inventory of PLCs, sensors, meters and existing systems. The sizer on this page lets you estimate the daily data points of your scenario with your own numbers.
Do you know what data your plant generates today?
Tell us which systems you have on the floor and we will tell you which signals can be captured, over which protocols, and how the project would be phased.
No commitment. You get a reply with a signal inventory and the project phases.
