Construction Incident Photo Sorter & Claims Review Queue

Turn hundreds of unorganised construction incident photos into a ranked, structured review queue for insurance claims — sorted by damage type, location zone, and reviewer priority, with the highest-risk photos on top.

Built for insurance claims adjusters, self-insured construction firms, and loss-adjustment teams who receive multi-hundred-photo batches per incident and sort them by hand.

What does the Incident Photo Sorter do?

1. Classifies each photo into one of eight damage types — structural collapse, fire, water, weather, equipment, surface, fall debris, electrical — using its filename, description, tags, and notes (metadata, not pixels).
2. Identifies the location zone — roof, foundation, facade, interior, excavation, elevated works, and more.
3. Assigns reviewer priority (high / medium / low) from damage severity and urgency keywords.
4. Groups photos by damage type, zone, and priority for instant drill-down.
5. Produces a ranked review queue so the highest-risk photos reach a human reviewer first.

All outputs are metadata triage only. Every damage assessment must be confirmed by a qualified assessor.

How do I sort incident photos with it?

1. Provide the photo metadata list (filename, description, optional tags/notes) and a claim_id.
2. Choose an output_format: full, queue_only, or summary_only.
3. Run. Every photo classified becomes one dataset row; your chosen output shape lands in the run's OUTPUT key-value record.

Input example

{
  "claim_id": "CLM-2026-001",
  "photos": [
    {"filename": "collapse_north.jpg", "description": "roof truss collapsed after storm, severe damage"},
    {"filename": "flood_basement.jpg", "description": "flooding in basement, burst pipe"}
  ]
}

Output example (one dataset row per photo)

{
  "photo_id": "collapse_north",
  "filename": "collapse_north.jpg",
  "damage_type": "structural_collapse",
  "location_zone": "roof",
  "reviewer_priority": "high"
}

Pricing (pay-per-event)

Event	Price
Actor start	USD $0.10 per run
Photo classified	USD $0.03 per photo (one dataset row each)

You pay only for photos actually classified. A small builder's weekly incident batch of ~100 photos works out to roughly USD $13/month; a 1,000-photo loss-adjustment batch costs $30.10. No subscription.

FAQ

Does it analyse the actual images? No — it classifies from photo metadata (filename, description, tags, notes). That makes it fast, cheap, and private: no images ever leave your storage.

Is this AI image recognition? No. It is deterministic keyword and rule matching — auditable and repeatable, which matters in a claims file.

Does it decide liability or WHS compliance? No. This tool organises photos for human review only. No damage assessments, WHS compliance judgements, or liability determinations are made by the software. All findings must be confirmed by a qualified assessor before any claim decision is taken.

What if a photo has no useful metadata? It still gets classified (to the closest match or a low-confidence default) and counted, so nothing silently disappears from the batch.

Support & feedback

Spotted a damage type or zone the rules miss? Open an issue on this Actor's Issues tab — rule additions are cheap and ship fast.