For developers

Your units, in a feed where buyers never fill in a filter.

theedge.estate deals properties one at a time, like a deck. People swipe and save what resonates; the algorithm infers type, budget, bedrooms and area on its own and builds a shortlist. Your sales team gets a brief with a taste profile, not a bare “call me back”.

Connect your feed

How it works

1

A feed without filters

The first cards probe across the feed, cheap to expensive. After that every card is chosen from what the person saved and passed.

2

A taste model

A fast pass is a hard no, a long look is almost a yes. Once the model is confident it reveals the jackpot — the closest match in the feed — and a shortlist.

3

A brief for sales

Name and channel, the profile (format, areas, beds, budget, must-haves), the shortlist and feed behaviour — to Telegram, email, amoCRM or a webhook into your CRM.

Or just say it

“Villa with a pool under a million in Bang Tao” — by voice or text. The request becomes the starting profile and swipes refine it.

Lists travel in messengers

A property or a whole list goes out as a link to WhatsApp or Telegram, with a preview, photo and price. Whoever opens it lands in the same feed.

What the feed must carry

Required
  • projectdevelopment name
  • typevilla, apartment, serviced, townhouse
  • areadistrict as a string
  • price + currencylocal-currency price
  • bedrooms, sqmbeds and floor area
  • images[].url3–5 frames, 1600px+
  • images[].tagexterior, interior, view, pool, plan
Nice to have
  • statusready or handover year
  • yieldnet yield
  • view, amenityview and pool
  • lat, lngfor distances
  • tenureownership form

Price bands, exploration strata and the vs-market figure are computed from the feed itself.

Formats

Yandex Realty XML

The feed you already send to Yandex. We pull it by URL every night; identical layouts collapse into one “from” card.

CSV or Excel

A price list: project, area, type, beds, size, price and photo links. Headers in English or Russian.

JSON over the API

For a direct integration: your system posts the feed to our endpoint with a token.

Serving rules

  • No filters: type, budget, beds and area are inferred from your picks
  • First 6 cards probe across the feed, cheap to expensive
  • A feature's weight is how well it separates saved from passed
  • Neighbouring bedroom counts and price bands get partial credit
  • A fast pass is a hard no; a long look before passing is almost a yes
  • A surprising choice weighs double; old signals decay
  • Every third card tests the weakest feature
  • The lead frame is chosen for the profile; the deciding frame reinforces its feature
  • “Back” undoes the decision and its signal — the model rolls back
  • The jackpot drops at 58% confidence with two of four features settled

Talk about connecting

Tell us which projects and where. We'll send how to connect the feed and show your units in the feed.