Listings licensed at the source, published as one open standard, served to every AI assistant through one open, neutral rail — and the qualified lead comes back to you.
AI finds and filters. A human confirms. You keep the client.
The Placy pipeline: agencies and networks opt in listings via CRM, feed, API, or CSV; Placy ingests and licenses, normalizes to one open standard, deduplicates, enriches, keeps status live, serves via MCP, and governs; AI platforms — ChatGPT, Gemini, Claude, Perplexity — query one rail; the qualified lead returns to the agency, attributed.
Step 1
01
Source it agent-direct
Placy licenses listings directly from the agency that holds the mandate — never scraped from a portal. The agency has the right to authorize redistribution, so every listing is licensed at the source: authorized from the moment it exists, never dependent on a portal's database right.
CRM / feed connectors
REST/GraphQL sync API + webhooks
CSV
Every listing is opt-in
Step 2
02
One open standard
Every property maps to one clean, machine-readable contract — built on schema.org RealEstateListing plus Residence, Apartment, and House types, GeoCoordinates, Offer for price, floorSize, numberOfRooms, bedrooms, bathrooms, and energy rating. The vocabulary the industry never had. Placy defines it and publishes it as one open standard.
3 of 881portals worldwide use the proper listing schema
There is no common contract for a European property. Now there is.
@type
RealEstateListing
about
Apartment | House
geo
GeoCoordinates
offers.price
Offer
floorSize
m²
numberOfRooms
rooms · beds · baths
energyRating
A–G
status
live
A machine-readable field list for one property: listing type, residence type, coordinates, price offer, floor size, room counts, energy rating, and live status.
¹ Portals using the schema.org RealEstateListing type. Audited by Placy, 2026.
Step 3
03
Five listings. One property.
The same flat is often listed by five agencies across five portals. Placy mints a canonical property ID — fuzzy-matched on address, geocode, unit, floor area, and agency reference — and clusters the duplicates into one entity, with each agency exposed as an offer. Objectively cleaner than anything scraped.
duplicate listing
duplicate listing
duplicate listing
duplicate listing
duplicate listing
canonical ID
1 property · 5 offers
Five overlapping duplicate listing chips collapse into a single canonical listing card holding one property ID and five agency offers.
Step 4
04
Enrich with AI
Specialized models clean, categorize, and extract structured attributes from messy multilingual agent prose — and from brochure images and PDFs, because vision and OCR are mandatory in real estate. Lifestyle tags and latent attributes are inferred, not hand-typed. Structured, enriched, and deduped is what makes the data usable.
family home near schools
move-in ready
Step 5
05
Keep it live
A real-estate status state-machine replaces retail's stock count. Price and status changes push as first-class incremental events — webhooks, at a target ≤15-minute cadence. Showing a sold listing destroys trust instantly, so a fast live status pipeline is non-negotiable — and itself a selling point against stale portals.
available
under_offer
sold_stc
sold
reversible
price_reduced
withdrawn
The listing status state-machine: available, under offer, sold subject to contract, sold — reversible — with secondary states price reduced and withdrawn.
Step 6
06
Serve it through one rail
An MCP server and registry is the endpoint AI platforms call — not a website we hope they crawl. Query-able, not crawl-able. One integration lights up every current and future AI surface. Placy also serves /agents.md and /llms.txt at the marketplace level.
search
listing detail
qualified-lead handoff
Step 7
07
Govern it
Lawful-basis hygiene, verified provenance on every listing, and verified-agent identity via Web Bot Auth — the open standard supported by Cloudflare, AWS, and DataDome — to admit good agents and block scrapers. Governance is the compliance story and the precondition agencies require before handing over data. Provenance is a feature, not overhead. Credit-scoring use is firewalled — the one real EU AI Act high-risk trigger.
Both sides
08
What each side experiences
If you hold the listings
Connect once. CRM, feed, sync API, or CSV — nothing per-platform. Opt-in, per listing; you decide what's syndicated.
Leads come home. AI-attributed leads land in your dashboard, tagged with which assistant drove them.
You keep the relationship and the data. Rails that send you leads — never a portal that owns your buyer.
Hand off cleanly to a qualified-lead action, attributed to the agency.
Questions
09
Frequently asked questions
What is Placy, in one sentence?
The neutral, licensed supply rail that makes European real-estate listings readable to AI assistants — sourced directly from the agencies that hold them, served to every AI platform through one open standard.
Is Placy a property portal or a search site for buyers?
No. We don't compete with agencies for the buyer. We make listings findable inside the AI assistants people already use, and we route the qualified lead back to the agency. The agency keeps the client and the deal.
Does Placy scrape listings?
No. Every listing is licensed and opt-in, sourced directly from the agency. That's the whole point — structured, licensed, and live beats scraped, stale, and unauthorized.
Can I buy a house through an AI chat with Placy?
No — and deliberately so. Property is a high-value, regulated, once-in-a-decade decision. Placy is scoped to discovery and qualified-lead handoff. AI finds and filters. A human confirms. That's what people actually want: 44%2 would pay extra for a human to verify an AI housing decision.
² Cotality, 2026. Four-country survey including the UK — the leading indicator, not a Europe figure.
The pipeline is being built.Plug in from either side.