Rightmove AI Property Search 🏠 — search UK property in plain English

🎉 Launch offer: completely FREE until 18 July 2026 — including AI scoring, no API key needed. Try it now while every run is on the house.

Search UK properties on Rightmove the way humans think — describe your ideal home in plain English and get AI-scored, ranked property matches with explanations of exactly why each one fits (or doesn't).

Input:  "Victorian flat in Hackney with south-facing balcony,
         near Overground, max £550k"

Output: Ranked properties with match scores:
  1. Brooke Road, E5 — 95/100
     ✓ Period conversion ✓ Balcony ✓ 0.2mi from Clapton Overground ✓ £550k
  2. Chatsworth Road, E5 — 82/100
     ✓ Victorian ✓ Near Homerton Overground ✗ No balcony

No API keys, no configuration, no Rightmove URLs to build. Type what you want, press Start.

What it does

1. Understands — your natural language query is parsed by AI into structured search intent: locations, bedrooms, budget, must-have features, style preferences, things to avoid
2. Searches — locations are resolved via Rightmove's own location lookup, then searched with your hard filters applied
3. Extracts — full listing details are pulled for the best candidates: description, key features, nearest stations, tenure, agent, images
4. Scores — AI evaluates every property against your wishlist, citing specific evidence from each listing
5. Ranks — results are sorted by match score, each with a one-line verdict and detailed ✓/✗ breakdown

Use cases

• House hunters — stop re-running the same filtered searches; describe the home once and let AI shortlist
• Buying agents & relocation services — turn a client brief directly into a ranked shortlist with justifications
• Property investors — screen for specifics filters can't express ("period conversion", "potential to extend", "not ex-council")
• AI agents & LLM workflows — pay-per-event pricing is fully MCP-compatible; the natural-language input makes this Actor trivially easy for agents to call

Use from AI agents & LLMs (MCP) 🤖

This Actor is built for the agent era. Because the entire input is one natural-language string, an LLM can call it without any parameter mapping — no location IDs, no filter syntax, no URL building. Whatever the user says about their dream home is the API call.

Connect via MCP — add the Apify MCP server to Claude, ChatGPT, Cursor, or any MCP-compatible agent and expose stueyboy/rightmove-natural-language-property-search as a tool. Your agent can then answer prompts like "find me a 3-bed house with a garden near good schools in Sheffield under £350k" with live, AI-ranked Rightmove listings.

Or call the API directly:

curl -X POST "https://api.apify.com/v2/acts/stueyboy~rightmove-natural-language-property-search/run-sync-get-dataset-items?token=YOUR_APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"query": "2-bed flat in Hackney with balcony, near Overground, max £550k"}'

Returns ranked JSON matches, ready to feed straight back into your agent's context. Pay-per-event pricing means agents are billed only for the results they actually generate — no subscriptions, no idle costs.

Pricing

Free until 18 July 2026 — all features, including AI scoring, at no charge during the launch period.

From 18 July 2026, pay per event — you only pay for what you use:

Event	Price	When
AI-scored property	$0.025	Each property scored and ranked by AI
Search result	$0.002	Each basic (unscored) property returned
Actor start	~$0.00001	Once per run

A typical run (search + 10 AI-scored properties) costs about $0.25 plus platform usage. Rightmove blocks datacenter IPs, so residential proxy usage (Apify proxy, billed as platform usage) is enabled by default.

Bring your own LLM key and AI scoring is free — set llmApiKey (DeepSeek, OpenAI, or any OpenAI-compatible endpoint) and the $0.025 per-property charge is waived; you pay your provider directly instead.

Input

Field	Type	Default	Description
query	string	(required)	Natural language search query
channel	enum	BUY	BUY or RENT
maxProperties	int	20	Max properties to collect from search
maxDetailPages	int	10	Max properties to fetch details for and AI-score
minScore	int	50	Only output properties scoring above this (0–100)
useLlmScoring	bool	true	AI scoring on/off (off = basic search, no scoring charges)
llmApiKey	string	—	Optional: your own LLM key (waives scoring charges)
llmProvider	enum	deepseek	deepseek, openai, or custom
llmModel	string	—	Optional model override
llmBaseUrl	string	—	OpenAI-compatible base URL (for custom provider)
proxyConfiguration	object	Apify residential (GB)	Recommended — Rightmove blocks datacenter IPs

Output

One dataset item per property, ranked by match score:

{
  "rank": 1,
  "score": 95,
  "property_id": 89657583,
  "url": "https://www.rightmove.co.uk/properties/89657583",
  "address": "Brooke Road, London, E5",
  "bedrooms": 2,
  "price": "£550,000",
  "type": "Flat",
  "match_summary": "First-floor period conversion with 2 bedrooms, within budget at £550k, 0.2 miles from Clapton Overground.",
  "match_details": [
    "✓ Period conversion (style preference met)",
    "✓ 0.2mi from Clapton Overground (near Overground met)",
    "✗ No south-facing balcony"
  ],
  "must_have_met": ["near Overground"],
  "must_have_missed": ["south-facing balcony"],
  "nice_to_have_met": [],
  "key_features": ["Period features", "Two double bedrooms"],
  "nearest_stations": ["Clapton (0.2mi)", "Rectory Road (0.6mi)"],
  "images": ["https://media.rightmove.co.uk/..."],
  "agent": "Example Estates"
}

The run summary (parsed intent, totals) is saved to the key-value store as OUTPUT:

{
  "query": "Victorian flat in Hackney...",
  "intent": {
    "locations": ["Hackney"],
    "min_bedrooms": 2,
    "max_bedrooms": 2,
    "max_price": 550000,
    "must_have": ["south-facing balcony", "near Overground"],
    "style_preference": "Victorian"
  },
  "total_found": 20,
  "detailed": 10,
  "scored": 10,
  "results_above_min_score": 7
}

Query tips

Write your query the way you'd brief an estate agent. The AI understands:

• Locations — boroughs, towns, postcodes, multiple areas ("Hackney or Islington")
• Budget — "max £550k", "under £2,000 pcm", "£300k–£400k"
• Must-haves vs nice-to-haves — "must have a garden, ideally with a garage"
• Style — "Victorian conversion", "new build", "period features"
• Exclusions — "not ground floor", "avoid busy roads", "no ex-council"
• Renting — say "to rent" and the search switches to rental listings

FAQ

Do I need an API key? No. AI scoring works out of the box — free until 18 July 2026, then charged per scored property. If you'd rather pay an LLM provider directly, supply your own key and the scoring charge is waived.

Why residential proxies? Rightmove blocks datacenter IPs. The default proxy configuration (Apify residential, GB) makes runs reliable; the cost is part of platform usage.

Can I use it without AI? Yes — set useLlmScoring to false for basic structured search results (locations, beds, and budget still parsed from your query).

Is it legal? The Actor reads only publicly available listing data, the same pages any browser downloads, with polite rate limits.

What makes this different

Every other Rightmove scraper is a "dumb pipe" — you build a filtered URL, it returns raw JSON, and you still have to read 200 listings yourself. This is the only Actor that understands what you actually want and tells you why each property matches. It replaces the hour of scrolling, not just the copy-paste.