Project Story

About the project

Inspiration

Renting property in the UK is notoriously stressful. From deciphering dense 30-page tenancy agreements written in legal jargon to tracking confusing utility bills, navigating energy ratings (EPC), and trying to hold landlords accountable to the Housing Act or the Renters' Rights Act, tenants are often left completely on their own.

We were inspired to build FlatMate because we wanted to bridge the gap between initial lease-signing and the final day of the tenancy. Too many renters miss critical clauses regarding guest policies, illegal financing, or unfair restrictions simply because agreements are designed to be incomprehensible. We wanted to create an AI-powered buddy that doesn't just read the lease upfront, but stays with you throughout the entire life of your tenancy.

What It Does

FlatMate is an end-to-end tenancy companion application designed to support renters at every stage of their renting journey:

Lease Demystifier: Uploads and analyzes tenancy agreement PDFs, instantly flagging crucial need-to-knows, hidden restrictions, and potential legal contradictions against UK legislation (such as housing acts).
Automated Landlord Communication: Automatically generates formal draft emails to landlords regarding any legal discrepancies or issues found within the agreement.
Ongoing Tenancy Buddy: Acts as an interactive chatbot helper from move-in day onwards, assisting with maintenance issues, tracking bills based on regional estimates, and handling meter readings.
Move-In Bookkeeping & Calendar Management: Features photo upload capabilities for move-in condition tracking and a smart calendar that coordinates bill payments and calculates notice periods (such as 2-month termination notice tracking).

How We Built It

We designed FlatMate with a robust modern tech stack tailored for flexibility, document processing, and AI intelligence:

AI & LLMs: Powered by the Gemini API to handle complex multimodal document analysis (processing lease agreements and move-in photos) and power the conversational chatbot buddy.
Backend & Deployment: Built to scale and deployed seamlessly using Vultr.
Data & Architecture: Utilizes Retrieval-Augmented Generation (RAG) coupled with a dedicated database of UK housing legislature to ensure all advice and flags remain strictly grounded in current legal frameworks.

Challenges We Faced

Complex Legal Parsing: Ensuring the AI could accurately distinguish between standard tenancy clauses and actual legal contradictions or illegal financing terms required fine-tuning prompts and establishing strict guardrails against UK housing law.
Multimodal Inputs: Balancing the processing of text-heavy PDF agreements alongside image uploads for property condition tracking required careful optimization of our LLM pipeline.
Feature Scope: Keeping the app focused on an end-to-end journey—from initial agreement review to ongoing bills and final move-out notices—meant carefully prioritizing core features for our hackathon MVP.

What We Learned

We gained a deep understanding of UK tenant rights, housing acts, and the complexities of building localized, legally-grounded RAG applications. We also learned how to effectively structure an AI assistant that transitions smoothly from a document parser into an active, day-to-day lifestyle companion for renters.