Platform
Get started

Casafari is the AI agent-native real estate data intelligence platform. The most complete property index in Europe: a deduplicated, cleaned property graph of residential and commercial property, for sale and for rent, in 16 countries.The most complete property index in Europe: a deduplicated, cleaned property graph.

How the graph is built
Markdown Open in Claude Open in ChatGPT

High Potential Seller Leads · Use cases

Compliant seller-prospecting workflow for an AI agent

For Builders of AI agents for agencies. Let an AI agent call High Potential Seller Leads (no names, phones or emails), read the property record, and leave consent and channel checks to a human.

How do I build an assistant that proposes seller prospects without handling owner personal data?

Why keep the agent away from personal data?

Agencies want an assistant that finds homes likely to sell in a territory, but a workflow that touches owner names or numbers inherits GDPR obligations. Under GDPR, prospecting calls need a lawful basis for the data held and, separately, must comply with the ePrivacy Directive and national telemarketing law, and a privacy notice is owed within one month when data was not obtained from the person, according to Tomba. Seller Leads returns no owner names, phone numbers or emails, so the agent reasons about properties, not people.

How does it work step by step?

  1. Teach the tools. The agent reads /openapi.json, /llms.txt and the Markdown (.md) hub pages; both products are REST only, with no MCP tools.
  2. Tool: list leads. GET /v1/seller-leads with lat, lon, radius_m, min_price_cuts, min_price_cut_pct, since, types, price_min, price_max and operation, derived from a request such as "flats near our office cut more than 10% since September".
  3. Tool: read the property. GET /v1/seller-leads/{lead_id} or GET /v1/property-sourcing/properties/{property_id}?expand=listings,history for the dated cuts and the agencies advertising the home.
  4. Draft, don't contact. The agent writes a prospect summary and hands consent and channel checks to a human.
  5. Recover from errors. On 429 request_limit_reached, wait for the named window; on 402 payment_required, show the upgrade_url.
bash
curl "https://platform.casafari.com/v1/seller-leads?lat=48.8566&lon=2.3522&radius_m=300&min_price_cut_pct=10&since=2026-09-01" \
  -H "Authorization: Bearer $CASAFARI_API_KEY"

What do you get?

Grounded prospect lists keyed by property_id, with cut counts, percentages and dates from one deduplicated record per home, which the assistant can explain and a human can act on. The rules the human applies differ by country: Spain requires prior consent or another GDPR Article 6.1 basis for commercial calls since 29 June 2023, according to Maldita.es; Germany requires express consent for telephone advertising to consumers, according to marken.legal; France requires prior opt-in consent for estate agents' prospecting calls from 11 August 2026, with calls limited to 10:00-13:00 and 14:00-20:00 on weekdays, according to MaFormationImmo.

What are the limits?

Seller Leads covers Spain, France, Italy, Germany, Greece and Andorra, within 300 metres of a point. Authenticate with a cf_sk_… key or an OAuth access token (Platform API keys); guards are 10 or 50 distinct places a day and 10 or 20 requests a minute, by plan. The agent must never infer or fetch owner identities elsewhere and present them as Casafari data. See also Grounded property tool for an AI assistant.

Esc

Tip: add .md to any URL to read it as Markdown.