Give your AI assistant live UK property data. Search the whole market, pull sold prices, EPC, planning and comparables, work out yields — and update your deal pipeline — without leaving the chat.
https://propertyengine.co.uk/mcpIllustrative demo — the example prompts and results shown here are mocked up.
The PropertyEngine MCP server exposes the whole platform to your assistant — sourcing, appraisal, market data and CRM. You never name a tool; you just ask, and it picks.
Ask for what you actually want — 3-bed houses in Leeds under £200k that need work, flats outside Article 4 areas in Cardiff, anything back on the market and reduced twice. Your assistant turns that into a real search across every listing we index, then narrows it with the same investor filters you would use in the app.
search_areassearch_propertiesget_search_filterslookup_propertypresent_propertiesPaste a Rightmove or Zoopla link, or name an address. Back comes the full listing history and every price change, what it last sold for, EPC current and potential, planning applications on the address, and a comparables pack — sold, for sale, under offer and rentals — sorted nearest first.
get_propertyget_property_photosget_property_comparablesArea-level market intelligence in a single call: on-market counts and asking-price stats, sold-price percentiles, market turnover, rental yields by postcode sector with average sale and rental prices, and flip outcomes — what people bought, refurbished and resold for.
get_market_statssearch_areasAdd properties to a board, label them, leave notes and log calls — in one message, across several properties at once. Your own CRM records are searchable too, so “which properties did I note the vendor was motivated on?” and “what have I not called this week?” are questions with real answers.
list_pipelinesadd_properties_to_pipelineadd_property_labelsadd_property_commentslog_property_callsget_property_activitycreate_pipelineNo filter panels, no exports, no spreadsheet gymnastics. These are real questions the connector answers — the four at the top are the ones playing in the demo above.
“Find 3-bed houses in Leeds under £200k that look like they need refurbishment — which have the best margin?”
214 listings match in Leeds. Here are the two where the asking price sits furthest below what refurbished stock nearby is actually selling for.
“Here is a Rightmove link — what is it actually worth, and what rent would it achieve? rightmove.co.uk/properties/154820913”
Asking £249,950. Six genuinely comparable sales within 400m land it at £232k–£241k, so the asking price is ahead of the market — and it has been listed twice before at £265k.
“Which postcode sectors in Nottingham give the best rental yields on 2-bed flats right now?”
Yields for 2-bed flats by sector, using average achieved sale prices against current asking rents. NG7 3 leads, though stock there is older and service charges run higher.
“Add both to my Q3 Refurbs pipeline, label them needs-refurb, and log that I called the agent on Cardigan Road.”
Done — both are on the Q3 Refurbs board in the Sourcing stage, labelled needs-refurb, and the call is logged against Cardigan Road with today’s date.
“Show me everything in Birmingham that has been reduced more than 10% and is back on the market.”
search_properties“What did number 14 last sell for, and has anyone applied for planning on it?”
get_property“Find flats outside Article 4 areas in Cardiff where an HMO conversion still works.”
search_properties · get_search_filters“Which properties in my pipeline have I not called in the last two weeks?”
search_properties (CRM filters)“Compare the sold-price growth in Stockport and Oldham over the last three years.”
get_market_stats“Pull the photos for this listing and tell me how bad the kitchen really is.”
get_property_photos“Find company-owned houses in Sheffield with a lease under 80 years.”
search_properties · get_search_filters“Where did I note the vendor was motivated? Bring those back up with my comments.”
search_properties (CRM filters)When your assistant has properties to show you, it does not dump a wall of text. It renders an interactive list inside the conversation: thumbnail, address, price and the figures that matter, with its own analysis pinned to each row.
The assistant’s conclusions and the numbers it used sit on the row itself, not buried in the paragraph above.
The list keeps the original search behind it, so “next page” pulls the following matches straight from the live data.
Open any property in the full app to run the numbers, check the map or start a direct-to-vendor letter.
Rendered by hosts that support MCP Apps, such as Claude. On other clients the same results come back as text.
Back on the market twice and reduced £16k. Refurbished terraces on the same street sold for £238k–£249k.
Cheapest £/sqm here, and an EPC potential of B means a modest spec gets it lettable.
Under offer, but it fell through in January. Outside the Article 4 area, so an HMO conversion stays permitted development.
Illustrative example of the property results view.
Most real estate MCP servers are built on US listing feeds. This one is built on the datasets a UK investor actually underwrites with — and the same ones behind PropertyEngine's sourcing filters.
Everything currently on the market, with the full listing history behind it — every price change, every relisting, every sold STC and fall-through.
What the property last sold for and when, plus sold-price percentiles for the surrounding area and previously flipped properties with the uplift achieved.
The rating today and the rating it could reach after improvements — the difference between a property that is lettable now and one that needs spending on.
Applications on the address with status and full-text searchable descriptions, so approved-but-unbuilt consent does not get missed.
Whether an HMO conversion still falls under permitted development, and whether the property sits inside a named conservation area.
Licences by status, the managing agent named on the licence, and expiry dates — useful both for compliance and for finding tired stock.
Properties held by a company, from Land Registry proprietor records — the starting point for most off-market approaches.
Internal floor area, plot size, price per square metre and years remaining on the lease, so value comparisons hold up.
Ask a general-purpose model about UK property and it answers from memory. It cannot tell you what came on the market this morning, what the house two doors down sold for in 2019, or whether the flat you are looking at sits inside an Article 4 direction — and when a model does not know, it tends to produce something plausible rather than nothing at all.
The Model Context Protocol closes that gap. It is an open standard for connecting assistants to live tools and data, and a remote MCP server is simply that connection published on the web. Add the PropertyEngine server once and every answer stops being a guess: each figure comes from a real query against listing, Land Registry, EPC, planning and licensing records, and every property links back to the app.
What that changes in practice is the shape of the work. Sourcing stops being a session of clicking filter panels and becomes a sentence. Appraisal stops being ten browser tabs and becomes a paste. And because your own pipeline, labels, notes and call logs are exposed to the assistant too, the follow-up — “save these two, label them, remind me who I have not called” — happens in the same breath as the search.
There is nothing to install and no API key to manage. Paste one URL, sign in with a 6-digit code, and start asking.
https://propertyengine.co.uk/mcpWeb, desktop and mobile. Any plan — the free plan allows one custom connector.
Web and desktop, on a paid plan. Developer Mode is a ChatGPT beta feature.
It is a standard remote MCP server, so any client that supports remote servers — Claude Code and Cursor among them — connects to the same URL. The connector is included with the Ultimate plan.
How the connector works, what data it reaches, and how to set it up in Claude or ChatGPT.
A UK property data connector for AI assistants
It is a remote MCP (Model Context Protocol) server that gives an AI assistant like Claude or ChatGPT direct, live access to UK property data and your PropertyEngine workspace.
Once connected, the assistant can search listings for sale and to let across the UK, pull Land Registry sold prices, EPC ratings, planning applications and comparables, work out area rental yields, and read and update your own pipeline, labels, notes and call logs — all from a normal chat, with no exporting or copy-pasting.
The open standard for connecting AI models to tools and data
MCP is an open standard that lets AI assistants call external tools and read external data in a consistent way. Rather than every assistant needing its own bespoke integration, a service publishes one MCP server and any MCP-compatible client can use it.
A remote MCP server like ours runs on the web — you add it by URL and sign in, so nothing gets installed on your machine and the assistant acts as you, with your permissions.
Claude, ChatGPT, and other MCP-compatible clients
Claude on web, desktop and mobile — added under Settings › Connectors › Add custom connector. See the Claude setup guide.
ChatGPT on web and desktop with a paid plan — added through Developer Mode under Settings › Plugins. See the ChatGPT setup guide.
It is a standard remote MCP server, so any other client that supports remote MCP servers — Claude Code or Cursor, for example — can connect to the same URL.
No — you paste one URL and sign in
No. There is nothing to install, no API key to manage and no config file to edit. You paste https://propertyengine.co.uk/mcp into your assistant, click Connect, and sign in with the 6-digit code we email you — no password.
It usually takes under a minute from start to first question.
Not from the portals — but you can paste their links here
Neither portal publishes an MCP server of its own. PropertyEngine indexes UK listings from the major portals alongside Land Registry, EPC and planning data.
That means you can paste a Rightmove or Zoopla link straight into the chat and the assistant will pull the full record for that address — its listing history, previous sales, EPC and nearby comparables.
Listings, sold prices, EPC, planning, licensing and ownership
Yes — pipelines, labels, notes and call logs, when you ask
Yes — it can add properties to pipelines, apply labels, leave comments and log calls. It only does so when you ask.
The assistant acts as you, inside your workspace, and sees exactly what you would see in the app. ChatGPT asks you to confirm before it makes a change. You can revoke access at any time from your assistant's connector settings.
The Ultimate plan, plus an account with your AI assistant
The connector is included on the Ultimate plan — see pricing.
You will also need an account with your assistant. Any Claude plan works, including the free plan, which allows one custom connector. ChatGPT Developer Mode requires a paid plan (Plus, Pro, Business, Enterprise or Edu).
A one-time email code — no passwords shared
Your assistant sends you to PropertyEngine to sign in, we email you a 6-digit code, and you approve access — your assistant never sees a password.
Access is scoped to your own workspace, and you can revoke it at any time from the connector settings in Claude or ChatGPT.
Live records instead of recalled guesses
Without a connector, an assistant answers from its training data. It cannot see what is on the market today, what a house last sold for, or what is on your pipeline — and it may fill the gaps with plausible-sounding guesses.
With the PropertyEngine MCP server connected, every figure comes from a live query against real listing, Land Registry, EPC and planning records, and each property links straight through to PropertyEngine.
Yes in Claude; ChatGPT Developer Mode is web and desktop
In Claude, the connector works the same on web, desktop and mobile once it has been added — useful when you are stood outside a property.
ChatGPT Developer Mode is set up on web or desktop. Because it is still a beta feature, mobile availability depends on ChatGPT.
A clickable results list rendered inside the chat
When the assistant has properties to show you, it renders an interactive list inside the conversation instead of a wall of text — thumbnail, address, price and key figures per property, with its own analysis attached to each row.
You can page through the rest of the matches without asking again, and clicking a property opens it in PropertyEngine. Hosts that support MCP Apps, such as Claude, render this view; on other clients the same results come back as text.