DeedScan

DeedScan is a commission-free FSBO marketplace that helps Canadians sell or buy houses directly, while simultaneously giving buyers neighborhood insights, verification and fraud checks.

Comment

What inspired us

We built DeedScan because one thing kept bothering us: selling a home in Canada is already expensive, and then on top of that, sellers often lose a huge amount in commission fees. For one of the biggest financial decisions in someone’s life, that just feels unfair.

We wanted to create something that gives more control back to regular people. The idea was to make a commission-free FSBO marketplace where buyers and sellers can connect directly, while still making the experience feel trustworthy and structured. A lot of for-sale-by-owner options can feel sketchy or incomplete, so we did not want this to be just another listing board. We wanted it to actually help people make decisions with more confidence.

We were also thinking a lot about how people do not just buy a house, they buy into an area and a lifestyle. That is why we cared about giving context around the listing too, not just the property itself.

Another thing that inspired us was thinking about how people interact with real estate in the real world. When someone walks past a house with a "For Sale" sign, they are often curious but have no easy way to instantly learn more about it. That moment of curiosity usually leads to searching online later or contacting an agent. We realized that the sign itself could become the entry point into the marketplace. By adding a QR code that launches a lightweight App Clip experience, a buyer can instantly view the listing, neighborhood details and contact the seller directly without downloading an app.

What it does

DeedScan is a commission-free marketplace for people selling their homes without a realtor.

Sellers can create listings with their property details, photos, price and description. Buyers can browse homes, look at detailed listing pages and get a better sense of the surrounding area. The goal is to make direct home buying and selling feel simpler, clearer and more affordable.

A big part of our project is also trust and fraud prevention. Since this is an FSBO platform, we knew it would only work if users felt safe using it. Because of that, we included:

  • fraud detection checks
  • verification checks for both buyers and sellers
  • extra admin verification for additional review and moderation

We also used Gemini to support confidence scores, fraud checks, verification flows, AI analysis, NLP-based search and a chatbot experience. To make the chatbot more reliable, we used prompt engineering and limited it to the actual listings on our platform so it could answer questions with less risk of hallucination.

That was really important to us because removing commission only matters if the platform still feels credible and safe.

DeedScan also introduces a new way to discover listings in the real world. Sellers can generate a QR code for their yard signs, and when someone scans that code it launches a lightweight mobile experience built with Reactiv ClipKit that simulates Apple App Clips.

This allows buyers walking past a property to instantly open the listing without installing an app. The clip shows key property details, photos, neighborhood context and an AI trust score. From there, the buyer can immediately tap "Message Seller" or "View Full Listing", which connects directly to the DeedScan web platform.

This removes friction at the very first step in the home search process.

How we built it

We built DeedScan as both a web platform and an instant mobile experience, focusing on the full marketplace flow from both the buyer and seller perspectives.

On the web platform, we worked on:

-seller listing creation -detailed property pages -buyer browsing and property discovery -neighborhood context around listings -verification and fraud-prevention flows -admin-side checks for added trust and moderation -QR code generation for physical yard signs -verified messaging service for buyers and sellers

A key part of our project was bridging the physical world and the digital marketplace.

Sellers can generate a QR code for their yard sign, and when someone scans it with their phone it launches a lightweight native iOS App Clip experience.

This mobile experience was built using Reactiv ClipKit, a platform that allows developers to create App Clip-style mobile experiences that launch instantly from QR codes without requiring a full app download. Using Reactiv, we built a lightweight property viewer that shows listing details, photos, neighborhood information and a direct seller messaging entry point.

The App Clip acts as an instant entry point into the platform.

When a buyer scans the sign they immediately see:

-the property listing -photos and price -property specs -a neighborhood snapshot -an AI-powered trust badge

All of this appears **without **downloading an app or creating an account.

From there, users can tap:

-Message Seller -View Full Listing

Both options deep-link directly into the DeedScan web platform, where the buyer can authenticate and continue the interaction.

Authentication and Access Control:

We used Auth0 for authentication and role-based access control (RBAC) so buyers, sellers and admins each had the correct permissions across the platform.

Once a user is authenticated and verified, Auth0 issues a JWT access token which allows us to:

-securely identify the user -determine their permissions -protect routes and APIs -enforce role-specific experiences

This helped make identity management and security structured and reliable across the application.

Data and Backend Architecture - We used Prisma to model and enforce our application data.

A large part of our work involved making sure our Auth0 roles and JWT-based permissions aligned correctly with our Prisma schema, ensuring that buyers, sellers and admins only had access to the data and actions they were supposed to.

AI Features with Gemini - We integrated Google Gemini across several core features of the platform, including:

-AI-powered fraud detection signals -price estimate suggestions -natural language listing search -our conversational assistant “Deedy”

Deedy helps buyers ask questions about listings and explore homes more easily.

To improve reliability, we used prompt engineering and constrained the chatbot's context to listings within our platform so it could respond accurately without hallucinating unrelated information.

Development Workflow - We also used Google Antigravity throughout the weekend as part of our development workflow.

It helped us with:

-feature development -automated testing -debugging complex UI issues

It was particularly useful for diagnosing a rendering issue related to displaying uploaded documents correctly within the interface.

Product Design:

A lot of our effort went into making DeedScan feel like a real product rather than a simple prototype. We focused on making both the buyer experience and seller workflow feel intentional and intuitive. Most importantly, we designed the platform so that trust is built directly into the experience, rather than treated as an afterthought.

Challenges we ran into

One of the biggest challenges was scope. Real estate is such a huge space, and once we started building, it was easy to think of ten more things we wanted to add. We had to keep asking ourselves what was actually necessary for the MVP and what could wait.

Another big challenge was access control. Since we have different user types on the platform, we had to make sure people only had access to what they were actually supposed to. That meant working through RBAC with Auth0 and making sure it matched properly with our Prisma schema and database logic. It sounds simple in theory, but in practice it took a lot of thought because we had to make sure buyers, sellers and admins all had the right permissions without exposing actions or data they should not be able to touch. We did not want a situation where someone could access admin-only functionality or modify data that was not theirs.

We also used Gemini across core trust and search features, including AI analysis, NLP-based search and a chatbot experience. For the chatbot, we used prompt engineering and constrained its context to the listings on our platform so it could answer listing-related questions more reliably and with less hallucination. We ran into quite a few hallucination issues and prompt specification problems here.

That challenge also tied directly into trust. Since this is a direct buyer-seller platform, people will naturally worry about scams, fake listings and whether the other side is legitimate. So we could not just focus on nice UI, we also had to think seriously about fraud detection, verification and admin review. Building those ideas into the platform while still moving fast during a hackathon was definitely challenging.

We also had to design the App Clip experience to work smoothly alongside the web platform, making sure the QR scan flow, deep linking and listing data all connected correctly between the mobile clip and the main application. We ran into the usual integration and setup issues that happen when building quickly as a team. There were moments where environment configuration, authentication flow problems and merging work slowed us down, and we had to debug those while still trying to finish the product.

What we learned

We learned that for a marketplace like this, trust is just as important as functionality. It is not enough to let people post and browse listings. You need to think about how the platform makes users feel safe, informed and confident enough to actually use it.

We also learned a lot about product scoping. At the start, it is tempting to build every feature possible, but we realized it is better to do the core flow well than to spread effort too thin. That helped us stay focused on the main value of DeedScan: lowering costs, reducing friction and making direct home selling feel more realistic.

On top of that, we learned how much goes into turning an idea into something that feels real. It is one thing to have a concept, but another thing to shape it into a usable product with actual buyer and seller considerations behind it.

We also learned that authentication is only one part of security. The harder part is making sure permissions are enforced consistently across the app, database and admin flows, especially when multiple user roles are involved.

In addition, we learned that using AI well is not just about adding a model to the product. It takes careful prompt design, constrained context and repeated testing to make the output actually reliable in a high-trust setting like housing.

We also learned how powerful reducing friction at the very first interaction can be. The App Clip approach showed us that when someone sees a house in the real world, the fastest way to engage them is to remove every barrier between curiosity and information.

What's next for DeedScan

There is still a lot we would want to improve. We want to make the fraud and verification system even stronger, expand the admin review process and continue improving the buyer-seller experience overall.

A big next step for DeedScan would be adding payments and more complete transaction support so the platform can handle more of the end-to-end home selling journey. We also want to improve the quality and customization of listings, giving sellers more ways to present their homes in a professional and flexible way.

Another feature we would love to build is customized virtual tours, so buyers can explore properties in a more immersive and personalized way instead of relying only on static photos. Overall, we want to keep expanding DeedScan into a more complete, trustworthy and user-friendly platform for Canadians who want to buy and sell homes more directly and more affordably.

Built With

Share this project:

Updates