BRENNEN

Industries / Construction and property

Damage restoration

Damage restoration is the work of making a building and its contents dry, clean, safe and whole again after water, fire, smoke or mold damage, from the first emergency call through drying and cleaning to the insurance estimate and the rebuild.

1867

New York's Fire Patrol was chartered as a paid salvage service, run by the New York Board of Fire Underwriters and paid for by an annual levy on every company insuring property in the city. Its patrolmen worked beside firefighters to protect property and records from fire, smoke and water and so to cut the cost of insurance claims.

nycfiremuseum.org

41

Milestones achieved across 7 domains.

8

To watch for: announced, not yet confirmed outside.

15

Open deliverables. Nobody has shipped them.

In one minute

Where it started

Insurers paid for the first damage control: salvage crews tied to New York's fire insurers date to 1803, and the city's Fire Patrol carried covers, brooms and buckets to fires until the underwriters shut it down in 2006, when a New York insurers' association said the companies funding it "no longer find any value in it". The modern trade grew after the Second World War out of rug cleaning and fire repair. Ten rug-cleaning firms formed the group that became the Restoration Industry Association in 1946, Martin King brought 45 damage restoration companies into it in 1971 with the aim, in his words, "to see the practice of insurance damage repair become a profession", and the Institute of Inspection Cleaning and Restoration Certification (IICRC) followed in 1972.

Where it stands

Water and fire are among the largest costs insurers pay on homes. In Verisk's US figures, published by the insurance trade group Triple-I, water damage and freezing made up 22.2% of homeowners insurance losses in 2024 and fire and lightning 19.9%, and from 2020 to 2024 the average water claim was $17,240 and the average fire claim $101,092. In Germany, by the insurers' association GDV's figures, building insurers paid more than €4.9 billion for leaking and burst pipes in 2024, over half of all their losses, and Swiss Re counted US$107 billion of insured natural catastrophe losses worldwide in 2025, US$40 billion of them from the Los Angeles wildfires.

What is open

Much of the trade runs on numbers nobody has published a test of: how many air movers a wet room needs, what a restoration dehumidifier removes once the air is nearly dry, how well a nailed tarp holds, how much soot each cleaning method lifts, and when old siding stops matching new. Other open jobs are shared records and test sets: drying logs from real jobs, many estimators pricing the same loss, assessors reading the same mold reports, and smoke-testing zones scored against lab results. Cheap sensors and load cells, simulation libraries, and AI models that read documents and video make each of them a job for a small team, and none requires entering a damaged or contaminated building.

Domains

AchievedWatch forOpen

Advances to build with

1 research note

Open in damage restoration

3 of 15 open deliverables

Open · Emergency response

Tarp fastenings pulled apart

A bench test of how temporary roof coverings hold. Fasten reinforced plastic sheeting to new plywood roof-deck pieces in the common ways (cap nails through the sheeting, nails through wood strips laid over its edges, and adhesive-backed covers), pull each to failure, and convert the forces into the wind each fastening spacing can resist on a one-story house.

Built with what exists

New plywood and shingle offcuts; sheeting sold for temporary roofs; a digital force gauge on a pull jig, or a universal testing machine at a college materials lab; the ASCE 7 wind load method and the free ASCE Hazard Tool for design pressures; Python for the tables.

Credit · unclaimed

Open · Water extraction and drying

Open drying logs

A public dataset of drying records from closed water-damage jobs, with addresses removed: daily temperature and relative humidity inside and outside the drying area, moisture readings for each material, equipment counts and run hours, the water's category, and the day each material reached its dry goal. Add a baseline model that predicts the days to dry from the first day's readings.

Built with what exists

Exports from the job-documentation software firms already use; an AI model to read scanned paper drying logs, with a person checking a sample; Python with pandas and scikit-learn; Hugging Face Datasets for release under an open licence.

Credit · unclaimed

Open · Mold remediation

Spore results, many readers

A test set of past post-remediation air-sample reports, with names and addresses removed, each giving the lab's indoor and outdoor spore counts by type. At least ten assessors, working separately, and three AI models each call every report pass or fail and give their reasons. Publish how often they agree on the same numbers.

Built with what exists

Reports donated by assessment firms with their clients' permission; a simple online form or Label Studio for the calls; AI models such as Claude, GPT and Gemini given the same report and prompt; Fleiss' kappa, a measure of agreement among many raters, computed in Python with statsmodels.

Credit · unclaimed