Skip to main content
All case studies

Tahanan. Property platform

Built with AI in days. Sold by a licensed broker.

Tahanan is a Philippine property platform for buyers abroad, run with a licensed broker. It lists developer projects, tells each visitor what they may legally buy, and matches them to a shortlist in 9 questions.

Here AI did the building, not the talking. The product has no chatbot on purpose. The matching is a set of rules you can read, and every sale closes through the broker.

The Tahanan home page on a phone. Screen 1 of 4.

How it works

  1. InDeveloper emails and brochuresThe broker forwards developer materials. Prices, dates, photos and the rules on who may buy.
  2. AIAI builds and fillsThe site, the catalog and the admin were built from a written brief. The materials became listings, each with its source and date.
  3. PersonThe team publishesA person writes the honest note on each project, sets the source and date on every market figure, and approves what goes live.
  4. OutA call with the brokerThe visitor answers 9 questions, sees a shortlist of up to 6 projects, and books a call. The broker closes the sale.

Where the work happens

  1. The Tahanan browse page: category tabs, filters for region, type, status and price, and project cards.

    The catalog

    61 developer projects with the company's own notes, a plain answer to who can buy each one, and filters for region, type and budget.

  2. Step 1 of 9 of the Tahanan questionnaire: what is your citizenship status, with 5 answers.

    The 9 questions

    The first question is citizenship, because it decides what a person may legally buy. The shortlist is built from the answers by rules you can read, not by a chatbot.

  3. The Tahanan admin leads page: a filtered list of leads with source, status and date.

    Every lead, answered by a person

    Who answered the questionnaire, where they came from, who is assigned and when they were called. The broker works this list. No AI replies to a lead.

  4. The Tahanan admin project list: projects with developer, place, status, image count and visibility.

    The broker's own admin

    61 projects with their status, photos and notes. The team edits here and the change is live when saved. Nothing goes to the public site without a person.

What AI does

  • Built the whole product from a written brief: the catalog, the questionnaire, accounts, the admin, the market pages and the guides.
  • Turned forwarded developer emails and folders into catalog entries, each with its source and date.
  • Gave the developers' own renders 5 seconds of camera motion for the home page. Nothing in the pictures was invented.
  • Fetches the exchange rate every day and refuses a move over 5%, so prices in dollars stay honest.

What people do

  • Close every sale. The site's job is to earn the call.
  • Write the note on each project, including cautions.
  • Put a source and a date on every market figure before it is published.
  • Answer every lead in the buyer's preferred channel: email, WhatsApp, Viber or phone.

By the numbers

2
days from the first database to a live site, October 12 to 14, 2026
61
developer projects listed, October 2026
9
questions to a shortlist of up to 6 projects
4
citizenship cases, each with a plain answer on what you may buy
0
chatbots. On purpose. A broker answers
30
days of daily backups kept at all times

What is under the hood

Built from a written brief in days. The first database was set up on October 12, 2026, and the site was live on its own domain on October 14, 2026.

The honest rules are built in: no countdown timers, no fake scarcity, no paid placement, a neutral default sort, pesos first with dollars marked approximate and dated.

A test refuses any release where a visitor could read about AI, placeholders or drafts. Visitors see a property site, not a technology demo.

Every table is backed up daily and kept 30 days. Errors email the team. Admins sign in with 2 steps.

Want a system like this for your business?

Tell us what the work is. We reply with what we would build and what running it costs.