Inspiration

OurStreets built an intuitive, crowdsourcing platform to capture momentary issues that impact the most vulnerable among us. Data Society is a data science company that built a foundational platform to ingest and identify grocery inventory data. When COVID-19 hit, both companies were compelled to find a way to contribute.

People all over the US are having trouble finding basic necessities, such as toilet paper, hand sanitizer, eggs, and bread. Every shopping trip represents a risk of exposure, and many people increased their risk exponentially to go on multiple trips to get the supplies that they needed.

No platform exists to provide real-time information about store inventory or levels of crowding at grocery stores.This is where OurStreets and Data Society saw an opportunity. By building an app that collects current inventory and crowding at stores from users, we can direct people to less-crowded stores that have the supplies they need without risk.

Data Society recently joined forces with OurStreets after exploring utilization of an app built for World Central Kitchen that optimizes store location and items in store (during a disaster). We will collaborate with OurStreets to pull in real time data and also classify and apply advanced machine learning techniques to further help with the crowd sourcing effort.

What it does

OurStreets Supplies helps you find what you need without risk. Using the existing framework of the OurStreets mobile application, currently available for Android and iOS, OurStreets Supplies is an intuitive reporting tool that gives users the power to report and track essential supplies like toilet paper, hand sanitizer, and fresh fruit and vegetables at retailers across the country, making their shopping trips more efficient and helping prevent the spread of COVID-19.

The app is simple: an interface for shoppers and retailers to update stock levels, and a search feature for shoppers and regulators to better understand the supply level of essential goods in their communities. A public facing map on the OurStreets website will enable everyone to be a stakeholder in creating safe and healthy communities. Retailers and government organizations who partner with OurStreets are equipped with a dashboard to track user submissions and data insights. With OurStreets Supplies, retailers are able to encourage safe shopping behavior, creating safer conditions for front-line employees. Municipal partners gain insights into demand and shortages across communities, empowering legislators to make changes to policy and enforcement to make sure constituents have access to the basic goods they need to survive. OurStreets Supplies is partnering with Data Society, a data science education company, to understand inventory levels and supply chain patterns provided by retailers. Data Society’s natural language processing will allow OurStreets to understand which essentials customers need most, creating a positive feedback loop that can be used to update the app to be the most effective for users.

How we built it

OurStreets

OurStreets is a community-driven data platform to report and analyze real-world issues. Our mission is to amplify people-power to make communities safer and healthier. We realized our existing infrastructure and tech stack could be used to help people report issues like empty shelves, and more importantly, where the supplies their neighbors need most were, we immediately began developing these new features. Talking about how we were able to leverage our current infrastructure

  • We repurposed our shared mobility ticketing system to connect with retailers in flexible ways.
  • We adapted our existing Expo and React Native mobile application to have a new flow for submitting supply reports.
  • We adapted our backend with new database tables to handle report submissions outside of the Safe Streets reporter infrastructure.
  • We created end points and APIs to store new data as well as building out the infrastructure to make reports searchable.

Data Society

Data Society is a D.C. based data science training and solution company to bring more data-driven decisions in all parts of the society.

Last year, we worked with the World Central Kitchen (José Andrés’ non-profit organization) to support disaster relief responders in providing emergency food relief. To expedite the sourcing of nutritionally-fit food, our data scientists developed an application that provided responders with data on supplies and nutrition at nearby grocery stores. This type of streamlining can save precious hours, which translates into our citizens’ wellbeing and stability during uncertain times.

We repurposed the foundation of the algorithms to help with the hardships people are facing due to the COVID-19:

  • Using a linear optimization objective function in R, we are able to give recommendations on the optimal grocery stores for the users to go, based on the availability of their items in need, user location information from Google Places API, in order to minimize unnecessary trips

Challenges we ran into

  • Because of the scale of the COVID-19 pandemic and the immediacy of lock-downs and shelter-in-place orders, we knew there would need a wide range of features that would be useful. In order to meet the moment, we had to make tough decisions to create a product that would work well now and delay additional features and UI/UX improvements for a later release.
  • Because of the immediate need, we created disaster recovery plans, ensuring our tech stack could support the anticipated scale. In the event that our relational database cannot handle the high volume, we are ready to migrate to a NoSQL database to avoid any data loss or interruption of service to users making reports. We’ve also migrated certain data over to NoSQL and are building a pipeline for new reports to handle larger volumes of data.
  • In order to optimize speed for iOS, we reworked the location lookup by switching from a static lookup of location to watch location APIs.
  • Scaling quickly and building the plane while flying it. Our five person team had to be quickly augmented with volunteers and partner companies.

Accomplishments that we are proud of

  • Using our current infrastructure to help out in a time of crisis.
  • Augmenting our core Safe Streets product with an entirely new tool, OurStreets Supplies, in 10 days.
  • Updating our brand and content strategy to address wider issues of health and safety in communities across the country.
  • Increasing our active installs of the app 10x in less than a week
  • Securing our first two retail partners (Union Kitchen and Open Door Market).
  • All the media coverage we’ve received thus far

What we learned

  • We’ve accelerated our customer discovery process to understand how best retailers will be able to use this tool and have secured our first partnership with Union Kitchen.
  • Our flexible infrastructure and community-driven platform was built for safe streets and could be applied to many real-world issues.Our mission is about amplifying people-power to create safe and healthy communities, regardless of the issue.
  • Using NLP to continuously update categories(tags) both on front/back end based on crowd source feedback

What's next for OurStreets Supplies

  • Release OurStreets Supplies Search by end of this coming week, so folks can find the supplies they need without the unnecessary risk of exposure to COVID-19.
  • Use NLP to analyze report comments and feedback through our website to adjust the types of items that can be logged through the app.
  • Implement data validation infrastructure using image recognition for submitted photos to understand which items people are looking for most.
  • Analyze photo metadata and report locations to track and report major supply chain issues and other geographic information that can be used by policy makers to address underlying issues that exacerbate health and human safety for the most vulnerable populations in communities across the country.
  • Translate the app into Spanish to increase its effectiveness in areas with large Spanish-speaking populations.
  • Use search words to forecast demand for the products.
  • Build two recommendation engines. One that focuses on recommending stores that are close to the user’s current location and that currently has the products the user needs. The second focuses on recommending items within the current store the user is at that are similar to what the user is looking for. Both recommendation engines will be based off of text and image data collected from users.
  • Build a predictive model based on consumer requests to notify retailers of surges in product demand and availability.
  • Collect user feedback to make improvements to UI/UX.
  • Partner with retailers and municipalities at the state, regional and local levels.
  • Repurpose our existing email, Slack, ZenDesk and Hubspot integrations to meet retailers where they already are

Built With

Share this project:

Updates