EstateX: AI-Valuation-Anchored Real Estate on the XRP Ledger

Sample dashboard

Inspiration

Real-world asset (RWA) tokenisation is often framed as “put assets on-chain and let the market decide the price.”
In reality, real estate cannot work like that.

Property markets rely on:

defensible valuation
explainability
compliance
trust from regulators and institutions

We were inspired by the gap between XRPL’s powerful settlement primitives and the lack of valuation-anchored, regulator-aware RWA platforms. With more experience with ML/AI than blockchain, we knew that we had to find a way to infuse it into our creation.

EstateX was born from a simple question:

What if XRPL handled settlement, but AI defined fair value first, something that is very often missing with RWAs?

Instead of speculative NFTs, we wanted to demonstrate how tokenised real estate could work responsibly.

What it does

EstateX is an AI-powered XRPL real estate trading prototype that:

Uses an AI Valuation Model (AVM) to estimate fair market value for properties
Explains why a property is priced the way it is using explainable AI
Tokenizes properties into fractional issued assets on XRPL (IOU model)
Simulates XRPL trustlines and settlement flows
Provides a chat-based assistant to explain valuations and guide decisions
Presents everything through a professional, institutional-grade UI

The key idea:
AI sets the price floor. XRPL settles ownership.

How we built it

Backend

FastAPI for a clean, modular API layer
XGBoost-based AI valuation model trained on official datasets structured real-estate features
Feature engineering & preprocessing pipeline for real-world messy data
Explainability layer using SHAP (with graceful fallback)
SQLAlchemy + SQLite/Postgres for persistence of data in database
XRPL transaction scaffolding using xrpl-py

Frontend

Vanilla HTML, CSS, and JavaScript for speed and clarity
SPA-style dashboard with properties, portfolio, trading, and chat
Real API calls for valuation & explanations
Mock inventory to focus on UX and core logic

XRPL Integration

Issued-currency model for fractional property tokens
Trustline-based access control
Designed to plug into XRPL Testnet/Mainnet without architectural changes

Challenges we ran into

Problem-Solution fit: We needed to make sure that our target problem was something real and actually worth solving, as well as that our solution was capable of solving said problem
Valuation realism: Real estate data is noisy, sparse, and non-stationary
Explainability vs performance tradeoffs with tree-based models
Designing XRPL flows that make sense for RWAs (trustlines, issuance, settlement)
Avoiding “crypto-hype UX” while still showcasing blockchain value

Accomplishments that we're proud of

Built a real valuation pipeline, not a mock price generator, based on REAL data
Integrated explainable AI in a user-facing way, with easy to understand visualisations
Designed an XRPL-native RWA architecture aligned with real financial constraints
Creating a Intuitive and easy-to-use UI that greatly enhances the user experience
Delivered a cohesive end-to-end prototype under hackathon time pressure

What we learned

RWAs demand far more rigour than typical Web3 apps
Explainability is not optional when money, assets, and regulation are involved
XRPL is exceptionally well-suited for settlement-centric, non-speculative use cases
Users trust systems more when AI decisions are transparent
A clean architecture matters more than flashy features, as our target audience is broad