area_trend
ChatGPTHow one area's recorded prices have moved, for 'is <area> going up', 'has Marina gone up this year', 'what are prices doing in JVC'. Pass one area (a name such as 'Dubai Marina' or a slug from search_areas). Returns the area's typical price per sqft for each of the last eight complete calendar quarters, ready homes and off-plan in separate blocks, with the number of sales behind every figure, and for the latest four quarters the year-on-year change: each quarter against the same quarter a year earlier, with both quarters named and both counts. A quarter with too few sales comes back as 'no figure' with its count, and a year-on-year change is shown only when both quarters have enough: repeat that gap, never fill it. Answer with the figure, its sign, the two quarters and the counts. It states what the register recorded; it does not say whether the area is 'going up' or 'going down', does not forecast, does not say what the price will do, and gives no advice to buy or wait: do not add any of those. Do not work out a quarter-on-quarter change from the levels (on this register it mostly reflects which homes sold), do not average or combine ready homes with off-plan, and do not offer a city-wide figure from it. For one area's other figures use get_area_detail; to put two areas side by side use compare_areas. Every figure includes its period and record count from the official property register.
areas_within_budget
ChatGPTAreas where the typical recorded price (buying) or yearly rent (renting) for a given size is at or under a budget in AED, most-traded first, top 8, each with its number of sales or new lets, the typical home size and the window. Use for 'what can I buy for AED 1.2 million, 1-bed', 'where can I rent a 2-bed for 90,000 a year', 'which areas fit a budget'. Pass budgetAed (dirhams: the purchase price, or the yearly rent), bedrooms (0 for a studio, up to 5), mode ('buy' or 'rent'), and optionally propertyType ('apartment' default, or 'villa' for villas and townhouses). A monthly rent must be multiplied by 12 first. Report it as a statement of recorded prices, in the tool's own words ('areas where the typical recorded price for a 1-bedroom apartment is at or under AED 1,200,000'), and always give each area's count and the window beside its figure. Do not describe any area as suitable, good value or a fit for the reader's means, do not rank by price, and do not add areas the tool did not return: an area missing from the list has too few recorded sales or lets for that size to show a figure, or a typical figure above the budget. Fees are not in the figure. Every figure includes its period and record count from the official property register.
building_sales
ChatGPTThe newest recorded home sales in one named building, as the building page lists them: for each sale its day, ready or off-plan, the bedrooms as filed, the size, the price and the price per sq ft, with how many home sales the register holds for the building and the years they run over. Pass the building's name, and how many sales to list (1 to 20). Source: Home sales (apartments, villas and townhouses) recorded in Dubai's property register, under the building's own name in its area. Period: The newest sales, newest first, each with its day; the total runs from the first year the register holds a sale for the building to the newest sale held for it, and the answer names both. Too few records: A row is one recorded sale, not a figure computed from several, so there is no minimum count. A building with no recorded home sale inside the site's plausibility bounds comes back as having none: never estimate one. Each row is one sale: it is not tied to a unit and names no one. Ready and off-plan are different markets, so never pool the rows into one price, and draw no trend from them; it gives no price by year. A signed-out reader may get fewer rows, as the building page gives them. Give no verdict, forecast or advice.
building_units
ChatGPTWhat the unit register lists in one named building, as the building page shows it: how many units, the building record's own flat and floor counts, hotel apartments, tenure as recorded and the bedroom mix with registered sizes; given a unit number, that unit's floor, bedrooms, size and parking, with the typical price of similar sales in the same building, ready and off-plan apart. Source: The units Dubai's property register lists for the building, as Keyva last loaded them, and the home sales recorded in it, for a unit's similar sales. Period: The unit counts are as the register stood when Keyva last loaded it, and the answer names that day; a unit's similar sales cover the 24 months to today. Too few records: Each side of a unit's similar sales needs 5 sales (the same bedrooms as filed, a registered size within a tenth, in the building, in those 24 months); with fewer, that side gives its count and no figure. Never estimate, and never present a gap as zero. It lists what exists, not what sold: no recorded sale names a unit, so a unit's figure is similar sales, never the unit's own price. Nothing here names an owner. Unit-by-unit rows are left out for a reader the building page keeps them from. Give no verdict or advice.
check_broker_licence
ChatGPTCheck a person or agency against the official property register's register of licensed brokers, by name, firm, or licence number. Returns the licence number and what the register's licence end date supports saying — a date is given only while it is still ahead. Use whenever a reader names an agent or asks whether someone is legitimate. A match means registered with the official property register — it is not a recommendation. Every figure includes its period and record count from the official property register.
check_price
ChatGPTIs an asking price in line with what comparable homes actually sold for? Give the area, the size in sqft, and the asking price. Returns the median, the 25th-75th percentile band and the sample size for completed homes of that size, from the official property register sales registered in the 24 months to the newest registered home sale, with the day that window runs to. Use this whenever someone asks whether a price is fair, high, or worth negotiating. Every figure includes its period and record count from the official property register.
check_project
ChatGPTWhat the projects register holds for one named project, as /check-project shows it: whether it is on the register and its register number, the register's status and percent complete, the target completion month, a cancellation if one is recorded, the escrow bank named on the file, and the registered developer with its licence year, each with the date of the file it came from. Pass a project name (English or Arabic) or a register number; when several projects match, the answer lists them to choose from. Source: The projects and developers files of Dubai's property register, as Keyva last loaded them; the answer names each file's date. Period: The register as of the file dates the answer names; a target completion month is checked against today's date, UAE time. Too few records: It gives no figures or counts, so no minimum applies; a project it cannot find is reported as not found in Keyva's copy of the register, never as unregistered. It cannot see the escrow account number, whether the escrow account is funded, or the project's advertising permit, and says so in every answer. A licence year that has passed is given with the file's date and never called expired. Give no verdict on the project and no advice.
check_rent_increase
ChatGPTThe RERA Decree 43 rules on how much a landlord may raise rent at renewal, plus the registered market rent for that area and size so the band can be worked out. Use for any question about a rent increase, renewal, or whether a rise is lawful. Give the reader the bands and the market figure; never tell them what to do about it. Every figure includes its period and record count from the official property register.
compare_areas
ChatGPTTwo areas side by side, for 'compare Marina and JVC' or 'X versus Y'. Pass exactly two area names or slugs (a full name is most reliable, e.g. 'Jumeirah Village Circle'). Returns, for each area on the SAME months: the ready-home and off-plan median price per sqft with the sales behind each, the three-year change on ready homes, gross yield with its sales and new-let counts, and how many ready homes changed hands, plus the page's own sentences on how the two differ. A figure with too few sales is returned as 'too few ... to compare', never as a number: repeat that, do not fill the gap. State differences as numbers and counts with their periods; do not rank the areas or say which to pick. Use this instead of calling get_area_detail twice. For one area use get_area_detail. Every figure includes its period and record count from the official property register.
dubai_price_history
ChatGPTThe city's typical sale price per sq ft by calendar year, as the price history page shows it: ready apartments, ready villas and townhouses, off-plan apartments and off-plan villas and townhouses, each apart, with the sales behind each. Pass from and to years to narrow it; for one area's quarters use area_trend. Source: Home sales (apartments, villas and townhouses) recorded in Dubai's property register. Period: Whole calendar years from 2011 to the year of the newest recorded sale; that year is labelled 'year to date' with its last day and is never a whole year. Years from 2007 to 2010 are the register's early years, given only when asked for and labelled so. Too few records: A cell with fewer than 30 sales in its year comes back as 'too few sales for a figure' with its count, never a number. These are typical prices of what sold, not an index: compare a cell only with the same cell in another whole year, never pool cells, and give no forecast or advice.
get_ajman_sales
ChatGPTThe figures /ajman shows, while Keyva shows them: sale transactions in Ajman's real estate development projects. With no arguments, the count and the total for each calendar year the file holds in full, and across every quarter it holds. With one unit type (apartment, studio, villa, office or shop) and one rooms count, the count and the total for that unit type and rooms count over the last four quarters in the file, and their average, given only as the total divided by the count. Call it for a question about Ajman property sales. Source: Ajman's sales file, open data published by the Department of Land and Real Estate Regulation, Ajman: one row per project, unit type, rooms count and quarter, each with a count of sale transactions and their total, and no sizes, no dates inside a quarter and no rents. Keyva adds the rows up across projects and quarters, and works out an average as the total divided by the count. Period: Calendar years, each with a total only where all four of its quarters are in the file; every quarter the file holds, from the first to the newest; and, for one unit type and rooms count, the last four quarters in the file. The answer names the years and quarters and the file's date, and says the newest quarter may be incomplete. Too few records: An average needs 5 or more sale transactions of one unit type and one rooms count in those quarters: with fewer, the answer gives the count and the total and no average, and a single sale is given with its price. A quarter the publisher's file does not hold is named as missing, never counted as zero, and a year missing one has no total. Where the file lists no sale transactions of a unit type and rooms count, the answer says only that, never that none took place; figures whose rows failed Keyva's checks are said to be withheld, never zero. The file covers sale transactions in Ajman's real estate development projects only, not every sale in Ajman: never give a median, a price per sq ft, a rent, a yield, a return or a percentage change from it, never put unit types or rooms counts together in one average, and never set these figures beside another emirate's. Never rank or compare projects, places or developers by these counts or totals. Name the publisher beside every figure, as the answer does, and give no advice.
get_area_detail
ChatGPTFull detail for one area: median price/sqft, price trend, rental yield where rent data exists, and the projects in it. Use after search_areas when the user asks about a specific area. Every figure includes its period and record count from the official property register.
get_area_rent
ChatGPTWhat homes in an area actually RENT for, from registered Ejari contracts: median annual rent, rent per sqft, the middle half of the market, how many contracts it rests on, and the share that are renewals. Use for any question about renting, what a rent should be, or whether a quoted rent is high. Sales tools cannot answer rent questions. Every figure includes its period and record count from the official property register.
get_building
ChatGPTLook up one Dubai tower or building by name and return what flats in it actually sold for, over the whole record and over the last twelve months. Use whenever a reader names a specific building ('Princess Tower', 'Marina Pinnacle') rather than an area or a project. Every figure includes its period and record count from the official property register.
get_developer
ChatGPTA developer's registered track record, summed over every company Keyva groups under its name: registered sales and the dates they span, completed vs still-selling projects, areas covered. Use when the user asks whether a developer is credible or how big they are. Every figure includes its period and record count from the official property register.
get_project
ChatGPTFull price breakdown for one specific development or project by name: median sale price and typical size for each bedroom count (studio, 1-bed, 2-bed, 3-bed), with ready (completed-home) and off-plan sales kept separate and never pooled, over the 12 months to the register's newest sale. Use after search_projects when the user asks what a specific bedroom size costs in a named development. Every figure includes its period and record count from the official property register.
get_ras_al_khaimah_figures
ChatGPTRas Al Khaimah's published property figures, as /ras-al-khaimah shows them: the monthly registered sales value in AED (last 24 months, with whole quarters summed), the share of each quarter's sales value by property category, building permits per month, and the latest census population, with its year. Call it for any question about Ras Al Khaimah (RAK). Give every figure with its period and credit the Ras Al Khaimah Statistics Center. It has no sale counts, sizes, prices per sq ft, areas, projects or rents. Every figure includes its period and record count from the official property register.
mortgage_figures
ChatGPTHow many mortgage registrations there are for every 100 completed-home sales, for one area or the whole city, as the area page and /cash-vs-mortgage show it, and for an area its mortgage registrations of the last 12 months, by type where the area page shows that split to the reader. Pass an area, or leave it out for the city. Source: Mortgage registrations and completed-home sales recorded in Dubai's property register; the figure per 100 is Keyva's arithmetic on the two counts. Period: An area's figure per 100 covers every year the register holds, and the answer names those years; the city's is given for all years and for the 12 months to the newest completed-home sale or mortgage registration, with that day. An area's registrations by type cover the 12 months to the newest mortgage registration city-wide, with that day. Too few records: An area needs 100 completed-home sales for a figure per 100, and 20 mortgage registrations in its 12 months for a count; the city needs 1,000 completed-home sales in each of its two periods. With fewer, the answer says Keyva gives no figure and names the floor: never a number, never zero. These count registrations against sales: they are not the share of buyers who took a mortgage or paid cash, since a mortgage is not linked to a sale, a remortgage registers with no sale beside it, and an area can pass 100 per 100. A modification or a transfer between lenders is not a new purchase. Give no advice on whether to take a mortgage.
price_by_bedrooms
ChatGPTWhat each bedroom count actually sold for in one area — studio, 1-bed, 2-bed and so on, with median price, median size and the number of sales behind each. Use when someone asks what a specific size of home costs, rather than an area-wide average across every size. Every figure includes its period and record count from the official property register.
project_returns
ChatGPTWhat each bedroom size in one named project or tower sells for, rents for and yields, as the project page's Returns table shows them: the 12-month typical price, the ready-home price, the yearly rent and the gross yield. Pass the project's or the tower's name; a tower's answer covers the whole project it belongs to. Source: Home sales and new tenancy contracts recorded in Dubai's property register. The rent is from the project's own recorded new lets where enough name it; otherwise from homes of about that size in the area's rent sector, and the answer says so. Period: The 12 months to today for the price, the ready-home price and the rent, each with its count; a size is listed once the project has 5 sales of it over all years. Too few records: A size needs 5 recorded sales over all years to be listed, 10 sales in the last 12 months for a recent price, 10 ready sales for the ready-home price and 10 recorded new lets of that size for a rent. With fewer, the answer gives the count and no figure: never estimate, never an older price in its place, never zero. The yield is gross (before service charges, agency fees and empty months), worked out by Keyva from the rent over the 12-month ready-home price: quote it as given and never recompute it. Where the rent is the area's and not the project's, no yield is given and the area's rent must not be divided by the project's price. The 12-month price counts ready and off-plan sales together and says so. Give no verdict or advice.
rent_by_bedrooms
ChatGPTWhat a home of a given size RENTS for in one area, by bedroom count: the typical yearly rent (median) for a studio, 1-bed, 2-bed and so on, from registered Ejari new lets, with the number of contracts behind every figure and the months they cover. Call it when someone asks what a 2-bed (or any size) rents for in an area, or wants rents laid out by size. It needs an area: get the slug from search_areas first, and ask which area if the reader named none. Pass bedrooms (0 for a studio, 5 for five or more) to answer one size, or leave it out for every size. Apartments and villas are separate cells: pass propertyType (APARTMENT or VILLA) to answer one, and leave it out to get both, each in its own block. A size with too few contracts comes back as 'not enough contracts' with no figure: say that plainly and never estimate one. It reports what the register holds and does not judge a rent: do not tell the reader a rent is high, low or fair. For an area-wide rent across all sizes use get_area_rent; for what a size sold for use price_by_bedrooms. Every figure includes its period and record count from the official property register.
search_areas
ChatGPTSearch Dubai areas tracked on Keyva by name, or list the most active ones. Returns real median price/sqft, recent price movement, and transaction counts from the official property register data. Use this for any question about area prices, which areas are affordable, or where activity is concentrated. Every figure includes its period and record count from the official property register.
search_guides
ChatGPTSearch Keyva's published guides on Dubai property process, regulation and costs (buying as a foreigner, Golden Visa, mortgages, RERA/escrow, transaction costs, service charges). Use this for process/regulatory questions before answering from general knowledge.
search_projects
ChatGPTSearch specific developments/projects by name, optionally filtered to an area. Returns median price/sqft and developer where known. Every figure includes its period and record count from the official property register.
service_charge
ChatGPTThe service charge (the owners' association budget, in AED per sqft per year) on record with the official property register for an area, a project or a building. Use it for 'what is the service charge in X' or 'what are the fees in this tower'. Pass name: an area, project or building name or slug (a full name is most reliable); add scope ('area', 'project' or 'building') when the name could be more than one. Returns, for an area, the middle of its projects' filed budgets with how many projects and which budget years stand behind it, and only when enough projects filed: otherwise it says there are too few budgets and gives no number, so repeat that and do not estimate. For a project or building it returns that project's own filing and its budget year, described as one filing and not a statistic, with how many cluster or tower budgets it holds. Every figure is a budget recorded for a year in advance, and charges have risen since: always give the year, never present it as this year's bill or as a cost of owning. It does not say whether a charge is cheap, expensive or fair, does not compare buildings or areas, does not subtract the charge from a yield, and gives no advice: state the figure, its year and its count, and point to the owner's invoice or the official property register's Service Charge Index for a specific home. Every figure includes its period and record count from the official property register.
tenancy_gap
ChatGPTHow much less sitting tenants (tenancies on renewal contracts) pay per square foot than new lets, for one area or for all areas together. Call it for 'do sitting tenants pay less than new lets in JVC', 'what is the gap between renewal rents and new-let rents', 'how far behind the market are old rents'. Pass area as a name or slug; leave it out for all areas together. Returns the typical rent per sqft on each side with the number of renewals and the number of new lets behind it and the months they start in, and the gap as a per cent of the new-let rate. An area with too few renewals or new lets is returned as 'too few renewals or new lets to compare', never as a number: repeat that, do not estimate. It states the recorded gap only: it compares two groups of contracts, not the same home over time, gives no advice on a tenancy and cannot say what one tenant's rent will do. For a specific rent increase use check_rent_increase. Every figure includes its period and record count from the official property register.