Case Study
Go4Green
Mobile application, OCR system, and predictive maintenance for Belgium's leader in cogeneration unit monitoring.
100%
Digitized meter readings
~0
Manual data entry errors
Real-time
Anomaly detection
Go4Green, part of Idex, is the Belgian leader in the installation of cogeneration units in boiler rooms and monitoring of their data.
The company operates large networks of technical installations requiring frequent, reliable meter readings under real-world field conditions, often in poorly lit basements with aging equipment.
Prior to this project, meter readings were collected manually on site, involving handwritten notes and photos stored locally. The process was time-consuming, error-prone, and made data centralization nearly impossible at scale.
The transformation
From manual chaos to automated precision
Before BRAFT
- Photos sent manually via messaging apps
- Data re-entered by hand into scattered Excel files
- Manual verification and correction of every reading
- No centralized database: information lost across devices
With BRAFT
- One photo via the app: data extracted automatically by AI
- Centralized database with full traceability
- Standardized, auto-generated maintenance reports
- Predictive models detect anomalies before they escalate
Mobile application
Custom app for field technicians
Operators identify meters via unique code or QR scan, capture readings and photos in seconds, and upload everything to a centralized database. The structured workflow eliminates manual lookup and reduces on-site time.
- Identify meters via QR code, no manual lookup needed
- Capture readings + photos in seconds, even in the field
- Upload & sync to a centralized database for full traceability
Optical character recognition
Automatic index extraction from photos
BRAFT developed a custom OCR pipeline that detects index values from on-site photos, even with imperfect lighting, angles, and device quality. Readings are validated, sorted, and integrated into the database automatically.
- Works with low-quality photos, reflections, and angled shots
- Adapts to different meter models and digit placements
- Robust zone detection isolates digits before parsing
Predictive maintenance
Detect anomalies before they escalate
Mathematical models predict plausible consumption values for each boiler and compare them against actual meter data. When measured values deviate significantly from predictions, the system flags the anomaly automatically.
This allows maintenance teams to intervene early, before a small inefficiency becomes a costly breakdown.
Automated reports
Standardized, self-generated maintenance reports
Based on checklists filled in by maintenance inspectors via the mobile app, reports are generated automatically in a standardized format. Decision trees identify defects and produce predefined comments.
Defect reports are stored in a database and shared with the responsible staff, preventing information loss and enabling faster, more efficient maintenance.
Start your own project with us
From mobile apps to OCR pipelines and predictive maintenance: we build field automation tools designed for sovereign, reliable operations.