Hokie Homes

Inspiration

We took a lot of inspiration for cites like Zillow and Apartments.com but even with these resources we still wanted an easier way to navigate the process. After experiencing a rough time finding housing for ourselves during college, we decided to make it easier on us and all of our peers.

What it does

It takes a user input, does a google search and copies this data into gpt-4, from there we provide 3 houses/apartments that fit their needs, and if they are interested in applying we help them with that process by providing a to-do list with every consideration they can take. We can then track this for them and provide a schedule with deadlines.

How we built it

1. We started from the ground up teaching ourselves how to implement Open AI how to give context and provide a user input and get out a formatted Json file.
2. We then started building with React after attending a seminar and seeing the capabilities, we created a search box with 3 add-ins for Maximum Rent, Bedrooms, and Bathrooms needed.
3. The next hardest part was connecting this to our python search script we used flask to build a local webserver to pull the data inserted and give it to our search function. We then needed to send this response back to react to populate the data.
4. After getting a data output, we implemented this into our React and choose to do an all in one application and webpage, immediately populating the data below the search bar. It took a while to format correctly but AI helped a lot
5. Once our information was correctly displayed, we then began work on the to-do list area. Which would give users advice and steps they should take for successfully renting an apartment or house.

Challenges we ran into

1. Coming into our project, we did not know exactly what an agent is, but we knew that we were going to incorporate it into out project. So the first challenge we faced was choosing between building our own agent or using a publicly available model.
2. On Saturday, we decided to completely restart our project because of the complexities we faced while using HTML, JavaScript, Python, etc. which led into us having to spend much more time and maximize our time efficiency.
3. We were in a group of 2, and around half-way through our second attempt at the project, one of our laptops stopped charging. This led to some quick decision making and planning for the rest of the project.
4. Another major issue we faced was AI hallucinations, some of our prompts were two strict and the returns were just completely made up. We fixed this by recommending the AI include a street name in their prompt, we then listed every street in Blacksburg.
   

Accomplishments that we're proud of

• Successfully pulling data from the web and parsing it in a python script with Open AI
• Overcame AI shortcomings and hallucinations
• Being able to adapt to changes and many challenges we faced
• Finishing our first Hackathon!

What we learned

• How to use an API
• How to Use React
• What an Agent is and how they work
• UI/UX design
• Time management

What's next for Hokie Homes

In a week we would hope to make a more automated applying process including sending an email for our users and establish a high tier UI with React. In a month we would hope to automate further by adding events to calendars, scheduling meetings with property owners, and sending notifications to user when an update occurs. Lastly we would like to expand to more campuses than just Virginia Tech to help every college student with renting apartments and homes.

Built With

• flask
• javascript
• openapi
• python
• react
• serpapi