Case study
Fire compartmentation reports built from pins, not memory
Earl Kendrick's fire compartmentation surveyors spent 3–4 hours on site, then 2–4 hours per report rebuilding what they had found from hundreds of photos and memory. Gigabyte built a mobile app that pins each defect to the floor plan with a standard category and photos, and a web platform that checks the pins and produces the report.
2026
- 3–4 hours
- on site per surveybefore the app
- 2–4 hours
- in the office rebuilding each reportbefore the app
- 100+
- defect types in one standard taxonomy2026
The challenge
Earl Kendrick's surveyors walk a building floor by floor, taking hundreds of photos with nothing to record where each was taken. Back in the office the report was reconstructed from memory, taking as long as the survey itself. Defect descriptions varied between surveyors, and every visit produced a standalone document with no link to the last, so changes and remediations could not be tracked.
What we did
- March 2026 — sessions with EK's fire safety surveyors on defect lists, measurements, observations and how a pin should behave on site
- March 2026 — proposal: capture the data first, then turn it into deliverables
- Phase I — mobile app: pins on floor plans, 100+ defect types, photos, notes and audit fields, with AI labelling the plans
- Phase II — web QA, PDF reports and filterable defect exports
- Next — AI that recognises defects from photos, once enough labelled surveys exist
What we built
Floor plans are set up in the office: upload a multi-page PDF and AI reads each page's title block and suggests the floor label for a surveyor to confirm. On site the app remembers the zoom position, so working along a corridor is a swipe and a tap. Each pin takes a category from the defect list, found by keyword, plus notes and photos, with who recorded it and when filled in automatically.
- Defects pinned on uploaded floor plans
- AI floor labels from plan title blocks
- 100+ defect types with keyword search
- Photos, notes and measurements on every pin
- QA stamp on each pin before issue
- PDF reports and filterable Excel exports
The result
Every defect now has a location, a category, photos and an audit trail from the moment it is recorded, so the report is generated from the survey rather than rewritten after it. Floor plans set up for a building serve every later survey type, fire doors included, and each survey adds labelled photos towards AI that suggests the defect from the picture.