Use cases
Who uses the server and how. The questions are examples of what you type into the conversation; the assistant picks the tools listed under each case itself.
Advert → address → everything else
Buyers, renters, agents
Adverts almost never give the address. Paste the advert's text into the conversation and the server reads it the way a valuer would: the area and its parts (balcony, cellar), the floor in every notation (“4/6”, “VP/2”, “pritličje”), year built, renovation year, lift, room count, street and place. Then it finds the building part that fits and says how sure it is. Once the address is known, everything else is available.
- Here is an advert: “…”. Which address is it?
- Is the price in this advert reasonable against sales in the building and the neighbourhood?
- What has happened to this flat before — sales, lettings?
- What is asked for similar flats in this part of town, and what is actually paid?
- the address and building part, with a confidence and the runners-up
- which details of the advert matched and which did not
- a value with its measured error, asked and paid prices with their n
resolve_propertyget_propertyproperty_historyestimate_valueasking_pricesmarket_statsneighbourhood
What the server watches for
A candidate's area corroborates, it does not select. When two buildings are nearly identical the answer is “ambiguous”, and the server says which detail would separate them. The advertised unit may not be registered on its own; an off-plan flat has no record at all.
Companies: at an address, and by name
Businesses, suppliers, journalists, buyers of commercial space
The AJPES register in both directions. From an address: which companies are registered there, by legal form, organisations before sole traders. From a name or a registration number: where the company is registered — and from there, everything the server knows about that address, building and surroundings. Business units are flagged and linked to their parent, so a village shop does not read as a chain's head office.
- Which companies are registered at Dunajska cesta 5, Ljubljana?
- Where is the company “Danes je nov dan” registered?
- What is behind this registration number, and how many business units does it have?
- Which companies are on Trubarjeva, house number by house number?
- What kind of building is this company's office in, and what is around it?
- the companies at an address, with legal form and counts by form
- a registered office by name (with or without diacritics) or by registration number
- whether an entry is a business unit, and which company is its parent
find_businessesfind_companystreet_profilebuilding_managersfind_ownerget_propertyneighbourhood
What the server watches for
A registered office is where the post goes, not proof anyone works there — the register holds no floor, unit or tenancy. Sole traders are people: the server shows them at an address, and never by name or registration number. Business activity (SKD) is not in the register's export.
Buying or renting a home
Buyers and renters
Before a viewing, learn what the advert leaves out: the building's year and renovations, the lift, utilities, gas and fibre, flood and landslide areas, what is nearby and how far the school the address belongs to is.
- What kind of building is at Cesta v Mestni log 55 in Ljubljana?
- Is this house in a flood or landslide area?
- Which school does this address belong to, and how far is it?
- Find me buildings like this one, but a house with a garden outside Ljubljana.
- the register's facts on the building and the unit
- utilities with the state of each fact (recorded or inferred)
- surroundings with two counts: OSM and the register
find_addressget_propertyterrain_and_utilitiesneighbourhoodsimilar_buildingsbuilding_permits
What the server watches for
The flood map is a protection instrument, not a risk assessment. A zero in OSM is not a zero. A fibre route is not a connection.
Valuation
Valuers, agents, banks
A sale value and market rent for a dwelling or a commercial unit, with comparable sales, the reason each excluded one was excluded, the GURS generalised value, and an error measured on past sales — for the segment, not one percentage for everything.
- What is this flat worth, and what is its market rent?
- Show the comparable sales and why some were excluded.
- What is this commercial unit in Celje worth?
- How far is the generalised value from market prices in this municipality?
- an estimate with its spread and the segment's measured error
- the individual transactions behind a statistic
- the price index, and old prices restated in today's money
estimate_valueestimate_commercial_valuefind_comparablesvaluation_basecarry_price_forwardprice_index
What the server watches for
Spread is not error. A commercial valuation anchors on the ratio to the generalised value, not on a price per m², and its measured error is on every response.
Investigative journalism
Journalists, analysts, NGOs
Who owns what, where the court auctions are, who manages buildings, where building is happening and which deals were the largest. Institutional ownership is visible; private owners carry no identifier in the registers and the server does not reveal them.
- Which institutions hold the most flats in Maribor?
- Where does the Housing Fund hold property?
- Which court auctions are near this address?
- What were the largest property deals in Koper?
- institutional ownership with the share that is visible at all
- auctions with the basis of each price, location and area
- the largest contracts and what each covered
institutional_ownersowner_holdingsfind_ownercourt_auctionsbuilding_managersportfolio_transactionsbuilding_permitsfind_company
What the server watches for
A lot leaving the auction board did not necessarily sell. The visible share of ownership is not one number. The server never returns a natural person by name.
Market research
Researchers, students, economists
Price and rent levels by area, market activity, the index, the rental register against the dwelling stock, and the current advert stock — all with n, period and a flag when a cell is too thin.
- How have flat prices in Ljubljana moved since 2015?
- How often do flats sell in Kranj compared with Ljubljana?
- Where are filed lettings fewest relative to the number of dwellings?
- How do asking and paid prices differ by size?
- statistics with n, period and flags
- asked kept apart from paid
- where to look for unfiled lettings — never how many there are
market_statsmarket_activityprice_indexletting_gapasking_pricescommercial_rentcommercial_sale
What the server watches for
The latest period is still filling and revises downward. The index has a known bias, which it states. A ratio between two registers is not a discount.
Land and planning
Councils, planners, investors
Designated use from the cadastre and the OPN, spatial planning units, the OPPN obligation, unbuilt building land, utilities and street lighting, building permits.
- Which unbuilt building plots over 1,000 m² are there in Kamnik?
- What does the spatial plan allow on this parcel?
- What is Trubarjeva in Ljubljana made of — dwellings, offices, businesses?
- How many building permits were issued in this district?
- parcels by what the plan allows
- a whole street, house number by house number
- permits and inspections
find_parcelsproperty_landstreet_profilebuilding_permitsterrain_and_utilitiesfind_businesses
What the server watches for
A parcel search result is never “available”. The OPN covers 192 of 212 municipalities, so it is served beside the cadastral use, never instead. An empty buildability parameter is not a limit.