Platform
The HOMAI property intelligence platform
HOMAI turns a Norwegian address into a roof and property risk report. Official register data, flood, landslide, precipitation and storm exposure, and an AI description of the roof are combined into a score you can trace back to its sources.
What is the HOMAI platform?
HOMAI by CUATRO is property intelligence software for Norwegian buildings: a web dashboard built around an insurance-oriented risk report. Its three modules use the same analysis of each address and the same deterministic scoring.
Roof analysis
Ten roof risk factors from up to seven aerial, satellite, street-level and LiDAR images and the byggeår in Matrikkelen. The AI describes the roof; arithmetic sets the score.
Physical climate risk
Flood and landslide from NVE hazard maps, precipitation and storm from ten years of MET Norway statistics, combined into a climate score from 0 to 10.
Portfolio analytics
Risk bands and the highest-risk properties across your latest analyses, with analysis history per property and CSV import.
What can you analyse for one property?
You choose the property by searching for an address, clicking on the map or drawing an area. HOMAI first collects what the public registers know about the building and the plot.
| Part of the analysis | What it adds | Main source |
|---|---|---|
| Building and plot | Byggeår, usable floor area (BRA), floors, building type, gnr/bnr and plot boundary | Kartverket, Matrikkelen |
| Terrain | Ground and surface elevation, building height, slope and aspect | Kartverket national height model (LiDAR) |
| Planning | Zoning plan, hazard zones and consideration zones, shown as registered | Geonorge |
| Heritage | A flag for buildings in the SEFRAK register | Riksantikvaren, via Geonorge |
| Climate hazards | Flood, landslide, precipitation and storm exposure | NVE, MET Norway |
| Roof | Roof type, pitch, material, condition, visible defects, chimneys, roof windows, vents and drainage | Up to seven images read by a vision model |
The analysis runs as a background job and can take minutes. A synthesis step turns the facts into around 40 report fields, including a summary, an insurance classification with recommendations and a maintenance plan. The score itself is plain arithmetic, as described in our scoring method.
How do you receive the results?
Every analysis ends in the same property intelligence report, which you can use in three ways.
- Interactive report. Scores, a factor table with the source of each value, hazards with cited sources, registry data, terrain, planning zones, an imagery gallery and the formula filled in with the property's own numbers.
- A4 PDF. Generated from the same HTML as the screen report. Admins choose which sections it includes and can hide the weight columns in client PDFs.
- Dashboard. Address search, map click and area drawing, portfolio pins, the analysis history of each property and the report viewer. Admins manage users, roles, categories and per-user access.
To load many properties, import a CSV file; the HOMAI portfolio view then summarises your latest analyses by risk band. For data inside your own systems, ask us about API access when you book a demo.
Which data sources are behind HOMAI?
Every official input comes from a Norwegian public source. These agencies supply data; they are not partners of CUATRO.
- Kartverket: Matrikkelen (the cadastre), address search, the national height model and Norge i bilder orthophotos, current and historical.
- Geonorge: zoning plans, hazard and consideration zones, and the SEFRAK heritage register (Riksantikvaren).
- NVE: flood zones with return periods, flood susceptibility and landslide susceptibility maps.
- MET Norway: ten years of precipitation and wind gust statistics from the nearest station, via the Frost API.
- NIBIO: AR50 land cover, as context for runoff.
Google and Esri imagery is part of the image fallback chain, and Open-Meteo adds the weather at the time of analysis. Each source's role is listed under data sources and methodology.
How reliable are the results?
HOMAI does not publish an accuracy percentage. Instead, each value in the report shows where it came from and how certain it is.
- Provenance on every value. Each derived value is labelled official, calculated, AI-estimated, AI-inferred or estimated, with the provider, a confidence from 0 to 1 and its basis. See how provenance labels work.
- Official values stay official. Registry values such as byggeår are never re-estimated by the AI.
- Nothing is filled in. Missing values show as "Not assessed" or "Not derivable", never as a guess.
- Repeatable scores. The same property, imagery and weights always give the same score, because the AI never sets it.
Uncertainty also counts against the result: low confidence adds up to 8 points to the risk index, and an analysis with fewer than three scored factors gets a 10-point coverage penalty.
What does HOMAI not do?
- It does not replace a physical inspection or a certified condition report. It is remote screening that helps you decide where a site visit is needed.
- It does not estimate building age, footprint or area from images. Those come from Kartverket's registers.
- It does not put a price on risk. There is no repair-cost, loss or property-value model.
- It does not monitor properties continuously or send alerts. Analyses run on demand, and the portfolio view summarises the latest ones.
- It is not instant. A full analysis can take minutes.
- It covers Norway only, because every official source it uses is Norwegian.
- Its imagery is not live. Aerial and satellite images are as recent as the provider's latest flight or capture.
Which teams use HOMAI?
The same report answers different questions depending on who reads it.
Insurance
Roof facts, climate exposure, an insurance classification and recommendations for each address.
Banks and lenders
Hazard and roof data for each collateral property, with the source of every value.
Real-estate investors
Building-level roof and climate data before an acquisition and across your holdings.
Property managers
Roof condition and maintenance plan fields for deciding which buildings to inspect first.
See HOMAI in a demo
We show you the dashboard, a full report and its methodology section, and answer questions about your use case.
Frequently asked questions
- Which properties can HOMAI analyse?
Any building registered in Matrikkelen at a Norwegian address can be looked up. The registry, terrain, planning and climate parts work without imagery. The roof part depends on the images available: with weak imagery, counts are marked as low confidence, and with no usable image the report states that the roof has not been assessed yet.
- How long does an analysis take?
A full analysis runs as a background job and can take minutes, because HOMAI gathers registry and hazard data, reads up to seven images and then computes the score. Results that are already cached come back faster. HOMAI does not promise answers in seconds, and there is no site visit to schedule.
- Does the AI decide the risk score?
No. A multimodal AI model describes what is visible on the roof and writes the summary text. The score comes from fixed arithmetic over weighted factors, so the same property, imagery and weights always give the same result. The formula, filled in with the property's own numbers, is shown in the methodology section of the report.
- Can we analyse a whole portfolio?
Yes. You can import a property list from a CSV file, and the portfolio view shows risk bands and the highest-risk properties across your latest analyses. HOMAI does not monitor properties continuously or send alerts; you run a new analysis when you need fresh results. See risk bands across all your properties.