Meta Pixel and Google Analytics load for measurement only with your consent. The site works without them. Details

Skip to main contentPROJEADAMTR
TRChange area

For professionals · In research

An open, bounded AI toolkit for real estate work.

We are researching a bounded, standalone version of the highest-scoring agent in our internal council evaluation, for professionals to adapt to their own workflows. First we will validate the task and safety boundaries; only then would we share source files and evaluation methods.

Current stage: research and user validation

OPEN CONTRIBUTION · HUMAN REVIEW

Make a task easier.
Keep the decision human.

There is no download, GitHub repository or release date yet.

WHY WE ARE RESEARCHING THIS

There may be an opportunity. We need to prove the impact.

The data suggests that access to AI and getting useful results at work are different things. An open kit might help bridge that gap; for now, that is a hypothesis to test.

14.0%

of Turkish enterprises with 10+ employees reported using AI

TÜİK · 2026
72.3%

of non-using enterprises considering AI cited a lack of expertise

TÜİK · 2026
46%

of U.S. real estate professionals reported a neutral or unnoticeable AI impact

NAR · 2025

TÜİK covers enterprises with 10 or more employees; these are not real-estate-specific rates. NAR surveyed U.S. members, so its result cannot be generalized to Turkey. The sources use different methods and should not be compared directly. The agent score is from our internal council evaluation only; it is not an independent sector benchmark or a customer outcome.

See the validation roadmap

OUR AIM FOR THE SECTOR

More than sharing files: make good practice inspectable.

The kit should be judged by whether it makes a real professional task clearer, repeatable and easier to control—not by download counts.

01

A lower starting barrier

An example task, setup notes and synthetic output should help teams run their own first test.

02

A starting point teams can adapt

Professionals can edit their own copies in their own repositories; we will not assume one workflow fits everyone.

03

Changes people can review

Proposed changes to the shared version should include a reason and a test case; they will be accepted or declined after review.

ROADMAP

Validate first. Publish later.

An open-source release needs a license, source files, tests and a named maintenance responsibility. We will not present an unpublished product as ready or compatible with every platform.

  1. 01

    Choose a real task

    Validate one recurring workflow with advisors, brokerage owners or project sales teams.

  2. 02

    Set the boundaries

    Package the agent without real customer data or unapproved access to external actions.

  3. 03

    Test before sharing

    Evaluate it on synthetic cases and with professionals; publish the sources, version and limitations.

USE BOUNDARIES

AI drafts. A professional verifies.

The first kit will not publish listings, message customers, write to a CRM or guarantee a sales outcome on its own. An authorized person must verify prices, availability, project details and regulatory information before publication.

RESEARCH SOURCES

Check the evidence behind the claims.

These sources help frame a sector opportunity. They do not prove the impact or safety of a Proje Adam TR kit.

This is a sector experiment we intend to evaluate with evidence.

Back to the professional area