Case study

Property agent platform

An independent property consultant wanted to represent buyers properly, without drowning in listings, spreadsheets and email. Lutenant designed, built and runs the platform that does it (client name withheld ahead of public launch).

85
versions live
since 2 Nov 2026
1,080
automated checks
on every release
6
languages, with
translated listings
Client platform: how the system fits together
Client platform system map: the app at the centre, connected to the buyer's agent, buyers, listing agents, decision makers, partner agencies' listings, the database and the email service.
System architecture walkthrough (client name removed).

The problem

International buyers relocating to a new market need someone on their side. Doing that well meant juggling homes spread across dozens of agency websites, each in its own format; the same house listed by several agencies at different prices; and each buyer's wishes, shortlist and viewings kept by hand.

The work that earns the fee, finding the right home and negotiating for the buyer, was being squeezed by the admin around it.

What was built

  • One place for every buyer: their search, shortlist, viewings and introductions, with website enquiries and voice-dictated briefs turned into buyer records in one step.
  • Listings from partner agencies: read from partner agencies' own websites each night with full photo galleries, sold homes filtered out, and the same home listed twice shown once.
  • Introductions that are tracked: from the first introduction, fee sign-off and prior-contact check through to the signed promissory contract and commission received.
  • A monthly report to each partner agency, partner agreement links, and a dashboard showing how the business is doing.
  • Six languages (English, Portuguese, French, German, Dutch and Spanish), with on-demand translation of property titles and descriptions.

How it's built

The same way I built platforms at Amazon and MongoDB, scaled to a small business:

  • Runs on Google Cloud with a managed MongoDB Atlas database, so there are no servers to look after.
  • Each agent can only ever see and change their own records.
  • 1,080 automated checks run before every release.
  • Released in small, frequent versions: 85 versions live since 2 November 2026.

Where it stands

The platform went live in November 2026 and is in its first phase: onboarding partner agencies and taking its first buyers through to a sale. This page will be updated with results as they come in.

← Back to all case studies

Have work like this in your business?

A 30-minute call, no pitch. Tell me where the hours go and I'll tell you honestly whether AI can help.