OWNRS UAE

The product

OWNRS is part of the RHK Properties family, built for property owners in the UAE to manage their properties from their phone. The app keeps track of things like overdue rent, maintenance and the value of the owner's portfolio.

The part I'm proudest of is the AI assistant. An owner asks a question in plain words and gets an answer based on their own records, not a generic guess.

Why a plain chatbot wasn't enough

A language model on its own knows nothing about one owner's tenants, leases or payments. Ask it which rent is overdue and the best it can do is guess, confidently. For an app people use to manage real money, a guess is worse than no answer.

So the assistant had to work from the owner's actual data, and be able to show where an answer came from.

How the assistant works

I built it on the Gemini API, with a NestJS backend behind the React Native app. Three pieces work together.

• Retrieval (RAG): before the model answers, the pipeline pulls the owner's relevant records and hands them over with the question, so every answer is grounded in their data and cites it.
• Tool and function calling: for data that changes, the model calls functions that read live property data at the moment it's asked.
• An agentic workflow: for each question, the workflow decides which tools to call before an answer is written.

What it shows about my AI work

This is the pattern I build for clients: retrieval over your own data, tools the model can actually call, and a workflow that keeps it inside the job it was given. Shipped in a production app, not a demo.