Using Conversational AI to engage your employees and provide support during the COVID-19 pandemic

The COVID-19 pandemic has significantly impacted our lives, with no end in sight, researchers are tirelessly trying to develop a vaccine or cure for this deadly virus. Reduced in-person interactions and an influx of misinformation is resulting in many of our customers struggling to manage communications and ensure employees have the support they need during this challenging time. 

The APJ Innovation Office was actively searching for solutions within the IT community and started seeing dashboards, simple WhatsApp chatbots and information to support consumers. There was no holistic vision for supporting employees or engaging them. During a conversation with my manager Anil Bhaskaran, he asked if the APJ Innovation Office could deliver a solution to support employees during the pandemic.

I collaborated with my peer Ashvini Kadaambi and following a call with our team, we decided to develop a solution that provides visualisation information as well as a platform that encompasses data and capability to support employees with questions such as:

• Where do I receive the latest information on COVID-19?
• What are the latest policies surrounding health?
• Can I still travel?
• I think I have symptoms?
• What are the latest cases?
• What are the new cases?
• Is the office open?
• Can I take leave?
• A customer has asked to meet face to face, should I go?
• Will the company cover my electricity costs?

The first step of the project was to design the user experience (UX). Sujit Adichikat created the initial screen designs to work towards (Figure 1). Our application would need to operate on both mobile and desktop/laptop devices and the controls and usability had to be seamless and consistent across all platforms.

Figure 1: First designs of Bridge-IT.

Sai Prakash took the design to concept stage, delivering within hours the landing page was deployed on SAP Cloud Platform (SCP) in early February. SAP Cloud Platform was selected as it supported all the services required to develop the platform including application services, mobile services, conversational AI, HANA , identity management and data intelligence. I designed a platform architecture for Bridge-IT, with the flexibility to use any number of services depending on what is requested from our customers.

Figure 2: The Bridge-IT architecture using both SAP and public services

Since a great amount of information already existed, we performed an analysis of the data that was being provided publicly. We started with the information provided by WHO (World Health Organisation) since at the time (early February) it was the most relevant and up-to-date. This posed a challenge as the data was in PDF format and was only released daily. We needed this process to be automated since manually entering data is tedious, so Anton Efremov applied his technical expertise by scraping the WHO web page and extracting the PDF for processing (Code Snippet 1). Anton used the Go language to perform this process as has a rich and simple capability to query web content.

func whoScrape(in, out chan interface{}) {

       for input := range in {
              url, ok := input.(string)

              if !ok {	

              ...
              }


	

              defer res.Body.Close()


	
              ...
              
	

              // Load the HTML document

              doc, err := goquery.NewDocumentFromReader(res.Body)

              if err != nil {

                      log.Fatal(err)
              }
	

              // Find the report items

              doc.Find(..)
      }

}

Code Snippet 1: Bridget page parsing

This provided us with the initial data, however, the format of the PDF’s kept changing and we identified that supporting data was being made available through other sites including Johns Hopkins and World o meters. Johns Hopkins was using ESRI ArcGIS and they provided a public hub where their data is publicly available. At this stage, Joni Liu leveraged his network tracing skills to support structuring the API calls to ensure the right data was extracted. 

The ESRI data provided useful information, however, we were able to determine that certain countries published more up-to-date data, the local data would be updated first and then synchronised with the global data. We emailed our global innovation colleagues to understand what country-specific data they had seen. The team supplied their COVID-19 specific links so we can target SAP relevant employee information.

With the relevant data captured, we had to store this information in a database. Since we were using SCP, HANA with its high performance, was a logical choice.  Jay Adnure and Changki Lee created the table structures and an API so that Bridge-IT and other applications could utilise our data. The data was exposed as a REST API, as this would allow our Conversational AI tool to be able to extract the information and include it as part of a conversation. Abinaya Sreenivasan and Sungyoul Baik were instrumental in the development of the chatbot and presentation of statistics (Figure 3).

Figure 3: Bridge-IT conversations for employees

Although the solution was answering questions and providing information on COVID-19, we decided we needed to enrich its capability by understanding what the SAP Global Health and HR team would like employees to access. Anil connected us with the right team, they had started developing a COVID-19 guidance tool). Annika Horn and Melanie Gomell from Global Health Management had collaborated with Nicolas Chrétien on developing the guidance tool using Qualtrics (Figure 4). The Qualtrics team have released free templates to support customers with COVID-19, these templates can be applied to Bridge-IT based on customers request.

Figure 4: The COVID-Guidance tool developed by the Global Health team

With the tool in place, we had to ensure all the HR related queries covered, this is where the global HR team involving Goerg Arnold and Rocco Guilliano provided us with the resources and the predefined FAQ and material for us to train Bridge-IT. This was the most challenging since each question (there were over 80) could have numerous utterances and we needed to train the bot to answer with the most appropriate answer (Figure 5).

Figure 5: FAQ questions that can be answered

Bridge-IT was now ready, the final feature we needed to complete was the visualisation of all the data. This is where Tetsuya Haneishi and Vriddhi Shetty supported us with some great visualisations using SAP Analytics Cloud so our customers can visually understand the effects of COVID-19 and add their data to visualise the impacts on operations as COVID-19 expands (Figure 6). In our figure, we mapped how COVID-19 was impacting the stock market. Now that we had completed all features of Bridge-IT, we started showing customers, our first customer had Bridge-IT installed within 2 days and we have a backlog of customers wanting to implement Bridge-IT to provide employees with their latest messaging and services. We're busy working with the global team to provide our Bridge-IT as open-source and we are preparing the deployment information.

Figure 6: Bridge-IT Visualisations using SAC

The crux of Bridge-IT only took 2 weeks to develop and we’ve continuously added more features as feedback from employees and customers. We decided to open-source the project to any previous SAP Cloud Platform Customer or any customer evaluating the SAP Cloud Platform. We're currently working with a few partners on a deployment package to assist with global distribution. The first partner that accepted to deploy Bridge-IT was Bourne Digital since they could deploy with their COVID-19 enterprise tracing app. We continue to listen to our colleagues, partners and customers to enhance Bridge-IT and support our global community. Vriddhi created a video that covers most of the features and functions of Bridge-IT, happy to receive input and feedback on what is important to you or your customers (Video 1).

Video 1: A video overview of Bridge-IT

This project is truly a collaboration of our APJ Innovation Office, SAP Global Health and Qualtrics team. There are so many people to thank including all those mentioned so far in this article (Anil Bhaskaran, Ashvini Kaadambi, Sujit Adichikat, Abinaya Seenivasan, Joni Liu, Sai Prakash, Jai Andure, Sungyoul Baik, Anton Efremov, Tetsuya Haneishi, Vriddhi Shetty, Changki Lee, Annika Horn, Melanie Gomell , Goerg Arnold, Rocco Giuliano and Nicolas Chrétien). Sinead Kaya and Claris Liow from the APJ Innovation Office for their contribution to messaging and sharing with our global colleagues. Sharon Om and Dorcus Yeung global innovation team, John Gibson for his support from the global centre of excellence, Martin Burger from the Global Health IBU and Andre Russel for his support in MENA and EMEA. Sai Methukupally for his guidance on the conversational AI product. Jeanette Tan, Sangita Jeyapathy and Fred Azis-Laranjo for their contribution with global communication and guidance around messaging. Selim Ahmed for his support on being the first partner (Bourne Digital) to offer Bridge-IT as part of their COVID-19 product offering. Last but not least, the rest of APJ Innovation Office for providing their time to train Bridge-IT including Brian Orloff, Lorraine Longato, Hyuk Jin Joh, William Tan, Vanessa Lim, Tanvir Muhammad, Peggy Renders, Mohammed Youssef, Shalini Agrowal, Ee Leen Gan and Nivedita Valluru Lakshmi. If you'd like to find out more, we have launched our public web site, please contact us and we'd be happy to work with our partners on how we can best support you and your business.