top of page

HOME.ai

Simplifying Real Estate with AI

Hackzon AI Project | Role: UX Designer | 2024

Home.ai

Overview

Project Overview

Homebuyers can face hundreds of pages of disclosure documents before making one of the biggest financial decisions of their lives. Most don't read them. HomeAI was designed to change that.

My Role

  • UX Designer

  • Led research, information architecture, interaction & UI design, and AI interaction design

BACKGOURND

Frame 2092_edited_edited.png

Context

We conducted rapid interviews with real estate agents during the 48-hour hackathon, focusing on where buyers most commonly struggle. Three core problems emerged:

Information overload

Buyers receive disclosure packets they lack the legal or technical knowledge to interpret, often skimming or skipping entirely.

Hidden risk

Critical issues buried in document language go unnoticed until after closing, leading to costly surprises.

Decision paralysis

Without confidence in what they've read, buyers either delay offers or rely entirely on their agent — creating bottlenecks on both sides.

DESIGN PROCESS

My Role & Scope

5-person team · 48-hour Hackathon · 2024

As the sole UX designer, I owned the end-to-end design process — from research and information architecture to interaction design and AI UX patterns. I worked closely with the front-end engineer to ensure design feasibility, and aligned with the PM on scope and priorities under tight time constraints.

Research

Given the 48-hour constraint, I conducted rapid interviews with real estate agents — chosen over buyers because agents have direct visibility into where buyers consistently lose confidence.

Three insights shaped the design:

  • Most buyers skip disclosure documents entirely — legal language creates an immediate comprehension barrier

  • Buyers rely on agents to interpret documents, but agents represent the transaction, not the buyer

  • Without understanding what they've signed, buyers can't make confident offers

The through-line: buyers need a neutral, plain-language source they can actually trust.

From the research, the initial scope focused on one problem: making disclosure documents readable. But during definition, we identified a critical gap — comprehension without action is incomplete. A buyer who understands the risks still needs to translate that understanding into a competitive offer.

We made the call to expand scope to include offer making, connecting the two flows deliberately: what you learn from the disclosures should directly inform how you structure your offer.

This shaped two clear design priorities:

  • Reduce comprehension barrier — surface risks in plain language, linked back to source text

  • Bridge understanding to action — guide buyers from "what does this mean" to "what should I offer"

Define

Design

AI-assisted offer form

The offer form was designed to auto-fill based on what the AI learned from the disclosure review — turning comprehension directly into action. Buyers can override any field, keeping them in control while reducing the blank-page paralysis of starting an offer from scratch.

2

The 1-page disclosure as a bridge

The most deliberate design choice was the "Download 1-page disclosure" CTA on each property card — a compressed, plain-language summary sitting between browsing and committing. It lowers the barrier to understanding without forcing buyers through the full review flow.

1

Progressive disclosure over full document dump

Rather than presenting the entire disclosure document at once, I designed a layered review flow — AI surfaces the highest-risk items first, with every summary linked back to the original source text. Buyers can go deeper if they want, but the default view gives them what they actually need.

3

PROTOTYPE

Prototype & Test

We moved from low-fidelity wireframes to a fully interactive prototype within the 48-hour window. The hi-fi prototype covered the complete end-to-end flow: conversational property search, disclosure review, and offer submission.

Key design decisions validated during testing:

Conversational UI as the primary navigation

Rather than a traditional search interface, we used a persistent conversation panel to guide users through the entire flow — keeping the AI presence consistent without making it feel intrusive.

Celebration moment at offer submission

We added a "Congratulations" screen at the end of the offer flow — a deliberate emotional design choice to acknowledge that submitting an offer is a significant, stressful moment for buyers.

Given the Hackathon timeframe, testing was rapid and focused. Feedback from judges and peers pointed to the conversational flow as the strongest differentiator.

Outcome

HomeAI was awarded the Llama Parse Award at the LlamaIndex RAG Hackathon (Feb 2–4, 2024, Santa Clara, CA) — recognized for best use of document parsing and AI-powered UX in a real estate context.

Within 48 hours, the team shipped a fully interactive prototype covering end-to-end buyer flow: from conversational property search, through AI-assisted disclosure review, to offer submission.

REFLECTION

The offer flow was simplified given the 48-hour constraint — in a real product, it would need to handle far more complexity: contingencies, negotiation history, and state-specific requirements.

The bigger gap was validation. Without real homebuyers testing the product, the core question remains open: does the AI-assisted flow actually build trust, or just feel like it does?

bottom of page