Inspiration

Our primary inspiration stems from the major concerns surrounding the recent Hurricane Relief efforts in Puerto Rico.

As businesses move towards the “Algorithmic Business Model”, it is imperative that we consider data driven decisions in every aspect of the business model. At the same time, we also appreciate Appian’s robust business process automation and data integration capabilities. Our inspiration came from the challenge of combining Appian’s strengths to create a business application that illustrates its data driven decisions capabilities, and apply it to Disaster Management - where lives truly matter.

What it does

Our application’s major capabilities include:

Disaster Recovery - At its core, the application is built around disaster recovery. Each component listed below is used to help distribute resources from the distribution centers to the various recovery centers set up to assist with each disaster.

Inventory Management - Volunteers using the application can request for inventory items to be delivered to their recovery centers. The inventory catalog is handled by the managers, while individual volunteers can request carts of items. Users can select the quantities and types of items leveraging Appian’s intuitive UI. Appian’s robust UI allows for a seamless user experience for disaster volunteers and managers.

Machine Learning Recommendations - As users add items to their cart, a Google serverless machine learning API recommends other items to add to the request based on a deep learning algorithm trained on request data. The recommendations update in real time as items are added to the cart. These recommendations help to ensure the proper goods are sent to each recovery center, and reduces human error.

Automated Reviews - Once requests are submitted, the application automatically reviews them based on their priority and the available inventory. This is done through a combinatorial auction, which is implemented through serverless Google cloud functions. The auction runs on a regular interval and ensures that each request can be accomodated given the available inventory. Should there be insufficient inventory for all of the requests, the algorithm finds the optimal combination of requests to approve given the priority of each. Given the criticality of disasters, automated reviews are essential to remove process bottlenecks.

Facial Recognition - When survivors arrive at a recovery center, they may not have proof of identity. To facilitate and track donations, users can upload a survivor's picture and link donations to that image. Any subsequent visits can now be tracked using the picture by leveraging the AWS Facial Rekognition Service. This also helps in reuniting survivors with their missing loved ones. By uploading an image of the missing person and comparing it with existing survivor images, the system can automatically return images of matching survivors and display information about when and where that survivor was last seen.

Drone Imaging - To facilitate the identification of areas in need, the application integrates with ArcGIS to allow users to tag aerial photographs based on the severity of the disaster taking place.

Multilingual Support - Disasters are a global phenomena, and so the application must function in a variety of languages. Using Google Cloud Translation API, the application can translate both system text and user entered data.

Twilio SMS - To notify users as quickly as possible, we have integrated with Twilio to send SMS messages. This allows disaster volunteers to be notified immediately while they're in the field.

Executive Reporting - Leveraging Appian’s rich UI, the application provides executive reports which highlight the key performance indicators of the request process and display the overall status of the recovery efforts.

How we built it

We built this application with a mobile first approach, as disaster volunteers will not always have computers on hand and will often be accessing the application via mobile devices. Thanks to Appian’s incredible mobile capabilities, this was easily accomplished.

Appian is the core of our application, and we leverage the rich variety of UI features to bring a powerful yet easy to use interface for volunteers, managers, and executives, both on web browsers and mobile applications.

Using Appian’s integration and web content components, we connected to a variety of external systems including Twilio SMS, Google Cloud Functions, Google Cloud Translation, AWS Lambda, AWS API Gateway, AWS Rekognition, and ArcGIS. This allowed us to achieve a high degree of automation and machine learning functionality, all orchestrated through Appian’s process models and SAIL interfaces.

Challenges we ran into

Leveraging facial recognition to uniquely identify and track survivor donations.

Combinatorial auctions are computationally intensive. To improve performance and streamline our design, the code for the auction is hosted outside of Appian, inside of a Google Cloud Function.

UI styling for optimal mobile experience - It was critical that the application have an excellent mobile UI experience. We achieved this by both leveraging the SAIL designer’s ability to preview mobile interfaces, as well as actually testing on mobile devices.

Limiting Google Cloud Translation calls - The Google Could Translation calls can be costly when multiple are needed on a single interface, which forced us to think about how to design our forms to minimize the impact of these calls.

Accomplishments that we're proud of

Leveraging the new Appian 19.4 UI/UX features - We are happy to have used many of the new Appian 19.4 features, such as the new colors and styles for progress bars and gauges.

Combinatorial Auctions - Combinatorial Auctions are notoriously difficult. We believe our solution of hosting them on a separate web service greatly improved the overall performance of our application.

Facial recognition - We were pleased with the facial recognition solution we implemented, particularly with the ability to search for users by image. We believe this can provide real value for a variety of use cases.

Recommendation Engine - We think the recommendation engine is a powerful addition to the Appian platform. Helping users with their daily tasks will save time and reduce errors for all parties involved.

What we learned

We learned the importance of mobile development, and how simple it is to build amazing mobile applications with Appian. We also learned how easy it is to integrate with a variety of external services for both UI and Process enhancements.

What's next for Data Driven Decisions using Appian (Disaster Management)

We want to integrate the application with DroneDeploy to achieve full real-time recovery control. This will allow the Appian application to not only facilitate the placement of orders but their execution and physical delivery as well. We believe with this additional functionality the application will be on its way towards automating the entire disaster recovery process.

Thank you!

Built With

Share this project:
×

Updates