The situation
Solar plants fail quietly. A cracked cell or a hot spot costs output for months before anyone notices, and walking a utility-scale site panel by panel does not scale. The imagery from a drone survey is only useful if something turns thousands of frames into a list of what to fix, where.
Stack
- Computer vision
- GIS mapping
- Drone imagery pipeline
- React
- Reporting
What we did
- 01Computer-vision defect detection over drone survey imagery, producing a report rather than a folder of photographs.
- 02GIS-based interactive maps so every detected fault has a location a maintenance crew can walk to.
- 03Coverage across the three plant shapes that behave differently: utility farms, commercial rooftops and floating solar.
- 04Outputs shaped for O&M teams — prioritised, actionable findings instead of raw model confidence scores.
- 05Client-owned data and transparent plans, so the asset owner keeps the record of their own plant.
The outcome
Chainfly runs at chainfly.co offering AI defect reports, GIS asset monitoring, smart solar layouts, agrivoltaics planning and C&I maintenance as a single service suite.