ShelterFirst

Making it easier to access the most basic human right: shelter. No digital services exist that find the capacity and eligibility criteria of emergency shelters in cities. We changed that

Comment

Inspiration

We are currently living through one of the largest housing crises in human history. As a result, more Canadians than ever before are seeking emergency shelter to stay off the streets and find a safe place to recuperate. However, finding a shelter is still a challenging, manual process, where no digital service exists that lets individuals compare shelters by eligibility criteria, find the nearest one they are eligible for, and verify that the shelter has room in real-time. Calling shelters in a city with hundreds of different programs and places to go is a frustrating burden to place on someone who is in need of safety and healing. Further, we want to raise the bar: people shouldn't be placed in just any shelter, they should go to the shelter best for them based on their identity and lifestyle preferences.

70% of homeless individuals have cellphones, compared to 85% of the rest of the population; homeless individuals are digitally connected more than ever before, especially through low-bandwidth mediums like voice and SMS. We recognized an opportunity to innovate for homeless individuals and make the process for finding a shelter simpler; as a result, we could improve public health, social sustainability, and safety for the thousands of Canadians in need of emergency housing.

What it does

Users connect with the ShelterFirst service via SMS to enter a matching system that 1) identifies the shelters they are eligible for, 2) prioritizes shelters based on the user's unique preferences, 3) matches individuals to a shelter based on realtime availability (which was never available before) and the calculated priority and 4) provides step-by-step navigation to get to the shelter safely.

Shelter managers can add their shelter and update the current availability of their shelter on a quick, easy to use front-end. Many shelter managers are collecting this information using a simple counter app due to COVID-19 regulations. Our counter serves the same purpose, but also updates our database to provide timely information to those who need it. As a result, fewer individuals will be turned away from shelters that didn't have room to take them to begin with.

How we built it

We used the Twilio SMS API and webhooks written in express and Node.js to facilitate communication with our users via SMS. These webhooks also connected with other server endpoints that contain our decisioning logic, which are also written in express and Node.js.

We used Firebase to store our data in real time.

We used Google Cloud Platform's Directions API to calculate which shelters were the closest and prioritize those for matching and provide users step by step directions to the nearest shelter. We were able to capture users' locations through natural language, so it's simple to communicate where you currently are despite not having access to location services

Lastly, we built a simple web system for shelter managers using HTML, SASS, JavaScript, and Node.js that updated our data in real time and allowed for new shelters to be entered into the system.

Challenges we ran into

One major challenge was with the logic of the SMS communication. We had four different outgoing message categories (statements, prompting questions, demographic questions, and preference questions), and shifting between these depending on user input was initially difficult to conceptualize and implement. Another challenge was collecting the distance information for each of the shelters and sorting between the distances, since the response from the Directions API was initially confusing. Lastly, building the custom decisioning logic that matched users to the best shelter for them was an interesting challenge.

Accomplishments that we're proud of

We were able to build a database of potential shelters in one consolidated place, which is something the city of London doesn't even have readily available. That itself would be a win, but we were able to build on this dataset by allowing shelter administrators to update their availability with just a few clicks of a button. This information saves lives, as it prevents homeless individuals from wasting their time going to a shelter that was never going to let them in due to capacity constraints, which often forced homeless individuals to miss the cutoff for other shelters and sleep on the streets. Being able to use this information in a custom matching system via SMS was a really cool thing for our team to see - we immediately realized its potential impact and how it could save lives, which is something we're proud of.

What we learned

We learned how to use Twilio SMS APIs and webhooks to facilitate communications and connect to our business logic, sending out different messages depending on the user's responses. In addition, we taught ourselves how to integrate the webhooks to our Firebase database to communicate valuable information to the users.

This experience taught us how to use multiple Google Maps APIs to get directions and distance data for the shelters in our application. We also learned how to handle several interesting edge cases with our database since this system uses data that is modified and used by many different systems at the same time.

What's next for ShelterFirst

One addition to make could be to integrate locations for other basic services like public washrooms, showers, and food banks to connect users to human rights resources. Another feature that we would like to add is a social aspect with tags and user ratings for each shelter to give users a sense of what their experience may be like at a shelter based on the first-hand experiences of others. We would also like to leverage the Twilio Voice API to make this system accessible via a toll free number, which can be called for free at any payphone, reaching the entire homeless demographic.

We would also like to use Raspberry Pis and/or Arduinos with turnstiles to create a cheap system for shelter managers to automatically collect live availability data. This would ensure the occupancy data in our database is up to date and seamless to collect from otherwise busy shelter managers. Lastly, we would like to integrate into municipalities "smart cities" initiatives to gather more robust data and make this system more accessible and well known.

Share this project:

Updates