NemoTheFinder

• 

  ui modified to remove pii

Inspiration

Moving around a lot for job requires me to haggle with brokers and landlords all the time. Often I have to allot dedicated hours from my day to negotiate rent, housing restrictions, and schedule apartment tours. It is a significant hassle for a lot of young people in their 20s and 30s.

What it does

NemoTheFinder is your apartment hunting buddy. You provide to it your 'must haves' and 'negotiables' in an apartment, a list of apartments you are interested in, and it phone screens the apartments for you. The app uses CallE voice agent to converse with property owners on your behalf and negotiate the stuff that generic property listing websites can't. It consolidates the findings from the calls into a dashboard in three groups, "Matched", "Negotiable", "Rejected". It also checks your availability(based off your google calender or provided timings) and negotiates a common time for apartment viewings. It is built to be lean in functionality to make sure you get your dream apt with minimal friction :)

How we built it

The app is built entirely on Python, with Streamlit for the display dashboard; It displays interactive flip cards for each screened property, isolating the (non) negotiables for the property owners It uses Sqlite for local data persistence. It uses CallE for agent conversations with property owners and integrates with Google Apis for checking availability and scheduling tours.

Challenges we ran into

At first we wanted to keep the app as cognitively minimal as possible by preloading popular rental requirements. However, the agent conversations turned non consequential without proper criterias. That's when we introduced the 'must haves' and 'negotiables', mimicking actual renter asks. Additionally, in India people often switch between languages during conversations and the agent had to be coded to understand and negotiate in both.

Accomplishments that we're proud of

My goal while building it was whether I would use the app or not. There is an overwhelming repo of information on the internet about renter housings and it is hard to manage this experience without mentally taxing the renter. My goal was a minimal interface that would ensure this experience is as hassle free as possible and NemoTheFinder does that excellently :)

What we learned

Used Streamlit for the first time, worked with CallE for the first time as well. It was fascinating to learn about all the 'awesome' :D stuff that people have built around CallE and made me realise how big agentic conversations can actually be in the future. I also learnt responsibility isolation; Have the application do one thing and do it to perfection.

What's next for NemoTheFinder

Need to host the app so that it requires minimal setup and can be used by the broader public. Would then make it public on twitter, facebook and reddit communities where people can try it out. Based on feedback will tweak features, but currently the immediate next step would be so that the app is able to scrape listings matching requirements of renter aswell. In the ideal flow, NemoTheFinder would take your requirements and schedule calls based off of distance, rent, etc to find the best match with minimal friction.

Built With

• calle
• google-calendar
• python
• sqlite
• streamlit