📍 See us at Wind Europe 2026, Stand: 9-D46
BLADE NEW AI

Board Logging and Automated Data Extraction

Every inspection board tells a story. But for most blade service teams, that story never makes it into a system. One photograph. Every field. Structured data in seconds.

The Problem

Data captured. Data lost.

Re-keying data from a board is slow, error-prone, and simply not practical at height on a mobile device with a small keyboard. So in practice, technicians do the sensible thing: they take a photograph and move on. The data is captured, but it stays trapped in the image, unstructured and unsearchable.

It sits there until a compliance request arrives three or four months later, asking you to prove you followed a specific step in the OEM work instructions. Or until someone needs to run a damage report across a campaign and realises the data was never there to begin with.

Time on blade is too valuable to spend on data entry. Your first priority is the repair.

BLADE™ changes that.

Screenshot coming soon
Built on Research

A decade of AI investment

In 2019, Railston & Co Ltd sponsored a PhD research programme at Loughborough University to explore how vision AI and machine learning could be applied to wind turbine blade inspection. At a time when the industry was not yet thinking about AI in this context, that work produced four peer-reviewed papers and a deep learning pipeline built for blade defect detection.

BLADE™ is a direct product of that understanding, applying the same multimodal AI principles developed through years of academic and industrial collaboration to one of the most practical problems in blade service: getting structured data off an inspection board and into a system, quickly and accurately, with minimal effort from the technician.

Read about our AI research programme →
86.74%

Mean weighted accuracy achieved by the IE-MRCNN pipeline, trained on real blade inspection data from Railston & Co operations and published in the Journal of Imaging, 2021. The same research foundations underpin BLADE™.

How It Works

One photograph. Every field.

BLADE™ is built into Collabaro Field and requires no additional hardware or setup. The full capture workflow takes under a minute.

1

Photograph the board

The technician takes a photo of the completed inspection board using Collabaro Field on their mobile device. Standard camera quality is sufficient. No specialist equipment required.

2

AI extracts every field

The image is processed by our vision engine, which reads every field on the board. Inspection boards vary in layout across contractors and companies. BLADE™ reads for meaning as well as position, making it compatible across the industry. For complex extractions, you can describe as well as guide BLADE™ AI if necessary.

3

Review confidence-scored results

Extracted values are returned to the technician with a per-field traffic-light confidence score. Fields that need attention are immediately visible. Fields that are reliable require no action.

4

Confirm and store

The technician confirms the record. Data is stored instantly in structured format against the correct Project, Job, and Task, ready for reporting and downstream use.

Screenshot coming soon
Confidence Scoring

Every field is scored. Nothing is assumed.

Offshore lighting conditions, worn boards, angled photographs, and handwritten entries all affect readability. BLADE™ accounts for this with per-field confidence scoring. Every extracted value is assessed internally and displayed to the technician as a simple traffic-light indicator: a green, amber, or red circle beside each field.

The thresholds that determine each band are fully configurable. If your operation requires a higher bar for green, or a tighter tolerance on amber, you can set those values to match your own quality standards. The defaults are designed to work well across most conditions, but the decision of what constitutes reliable enough is yours to make.

Green — 85% and above
The extraction is reliable. No action required from the technician.
Amber — 50 to 84%
The value is plausible but warrants a quick visual check against the board before confirming.
Red — below 50%
The field was unreadable. The technician enters the value manually. This is the only keyboard interaction required.
A note on high-stakes identifiers. A small number of fields, including site name, turbine number, blade number, and asset ID, are always flagged for human confirmation regardless of confidence score. A wrong value stored against the wrong asset is a data integrity problem that is difficult to correct downstream. BLADE™ is designed to prevent it.
Screenshot coming soon
Data Value

Captured once. Available everywhere.

The wider significance of BLADE™ is not just speed. It is the data that was never being captured at all.

Inspection boards contain structured information about damage location, category, blade position, environmental conditions, and repair scope. In a manual workflow, much of this data never reaches a database. It lives on paper, in photographs, or in someone's memory. Across a multi-turbine campaign, that represents a significant gap in operational intelligence.

Board photo
Technician photographs completed inspection board
BLADE™ extraction
AI reads and structures every field
Confirmed
Technician reviews confidence-scored results
Stored
Structured against Project, Job, and Task in Collabaro

With BLADE™, every confirmed record is stored in fully structured format against the relevant Project, Job, and Task in Collabaro, immediately available for report generation and downstream integrations. Data that was once too time-consuming to capture is now captured as a matter of course.

See BLADE™ in action

Book a demo and we will show you how BLADE™ turns a photograph of a completed inspection board into structured, searchable operational data in under a minute.