Industries
Manufacturing & IoT
MES/ERP integrations, telemetry platforms, quality control with computer vision and predictive maintenance — from the sensor to the dashboard.
What are the typical challenges in manufacturing and IoT software?
Factories generate data that nobody stores, defects that are found at the end of the line, maintenance that happens only after a breakdown, and three systems that disagree about stock. According to the IEA (2024), industry accounts for roughly a third of global final energy consumption, so waste in energy and downtime is a cost line as well as an efficiency one. The regulatory side is also moving: the EU Machinery Regulation (Regulation (EU) 2023/1230) applies from 2027 and explicitly covers software and self-evolving behaviour of machinery, and the Cyber Resilience Act (Regulation (EU) 2024/2847) adds security obligations for connected products.
- Machines produce data nobody stores or reads
- Quality defects found at the end of the line, not at the source
- Reactive maintenance and expensive unplanned downtime
- ERP, MES and spreadsheets disagree about stock and orders
What do we build for manufacturing?
We build telemetry ingestion platforms on queues and time-series stores, MES/ERP/PLM integrations and production dashboards, operator tablets and technician mobile apps, and supplier and order portals. The architecture is the same one we use for our event-data products: ingest millions of events a day, store them cheaply, and make them queryable in seconds. Industrial security follows the ISA/IEC 62443 zone-and-conduit model (IEC, 2023 edition of the core parts), so OT networks are segmented from the IT services that read from them.
- Telemetry ingestion platforms on queues and time-series stores
- MES/ERP/PLM integrations, production dashboards
- Operator tablets and technician mobile apps
- Supplier and order portals
Where does AI move the numbers in manufacturing?
AI moves manufacturing numbers in inspection, maintenance and planning. Deloitte (2017) estimated that predictive maintenance can cut maintenance costs by 5–10% and reduce unplanned downtime, and McKinsey (2017) reported reductions in machine downtime of 30–50% and longer machine life where it is done well. Computer-vision inspection finds defects at the station instead of the end of the line, planning models balance demand against capacity, and Document AI reads drawings, specifications and certificates. All of it depends on telemetry that is actually captured, which is why we start with the data platform.
- Computer-vision quality inspection
- Predictive maintenance on sensor data
- Demand and production planning models
- Document AI for drawings, specs and certificates
Scheduled vs predictive maintenance: which do you need?
Scheduled maintenance is the right baseline for cheap, uniform assets and where sensor data does not exist; predictive maintenance pays off on expensive, instrumented equipment whose unplanned stops are costly. The prerequisite is months of sensor and failure history. We usually keep the schedule as a safety net and let the model move interventions earlier or later within it.
| Criterion | Scheduled (time-based) | Predictive (condition-based) |
|---|---|---|
| Data needed | Manufacturer intervals | Sensor telemetry and failure history |
| Unplanned downtime | Reduced but not eliminated | Reduced further where models are accurate |
| Parts and labour | Replaced on schedule, sometimes early | Replaced on condition |
| Up-front investment | Low | Sensors, data platform, models |
| Fit | Uniform, low-cost assets | Critical, instrumented assets |
Why Glanit for manufacturing?
Because our event-data platforms already ingest millions of events a day in production, and a factory floor is the same shape with different sensors. Sixteen years and 500+ projects have taught us to build the data platform first, integrate MES and ERP through adapters rather than point fixes, and add models only where the telemetry can back them. We align security work with the NIST Cybersecurity Framework 2.0 (NIST, 2024) so that IT and OT teams share one vocabulary. See cases and AI & machine learning.