As the human services agency for the State of Arizona, the Department of Economic Security (DES), led by Director Michael Trailor, offers more than 40 programs and services, which include child support services; child care assistance; employment services; unemployment insurance, nutrition, cash and medical assistance; shelter services; and supports for individuals with developmental disabilities.
DES’s first chatbot project focused on the Supplemental Nutrition Assistance Program (SNAP) and the associated policy manual that helps employees determine eligibility. The challenge was streamlining the eligibility process. The current process for determining eligibility for Nutrition and Cash Assistance can be complicated and create delays in several areas. In some cases, applicants wait in lines to meet with a DES employee to review his or her application, and the employee may need to look up the relevant policies in the online eligibility manual before making a determination. Because the eligibility process involves document verification and a series of questions, the added task of looking up policy online can make a long process even longer.
Andy Pitman, Director of State and Local Government Health and Human Services Solutions, recently sat down with Sanjiv Rastogi, CIO of Arizona DES, to discuss its chatbot progress.
Sanjiv, please tell us a bit about the mission of DES.
DES is the human services agency for the State of Arizona. We serve approximately 2.9 million citizens with services across the spectrum – from unemployment insurance to nutrition assistance to developmental disabilities to child support, for example. DES invests approximately $4 billion to serve those in need. Our mission is to make Arizona stronger by helping Arizonans reach their potential through temporary assistance for those in need, and care for the vulnerable.
So, part of that mission is assisting Arizona citizens in their enrollment for programs like SNAP. I understand that can be challenging. Please explain why.
The programs we administer are supported through a combination of state and federal government funding. This funding requires eligibility determination before offering services. Due to the manual nature of the eligibility process and lengthy manuals, it is a complex process. Finishing these interviews in a timely manner with 100% accuracy has been nearly impossible until now, and some of our applicants must come back a second time and may have the decision for benefits impacted.
You felt that a chatbot could improve discovery of that information so that the citizens’ enrollment experience could be improved. Please tell us about that.
As we looked at areas of improvement to the quality of service to citizens, we met with the Division of Benefits and Medical Eligibility (DBME) and determined that the benefits eligibility process was the most significant friction point. Our employees required 16 weeks of training before they were qualified to interview applicants. Our evaluation uncovered three possible solutions:
- Easy access to Frequently Asked Questions (FAQs)
- More training
- Modernizing our technology
In the spirit of lean transformation through the Arizona Management System (AMS), we landed on implementing modern technology to truly address the problem, which led us to the idea of a chatbot. Thus, SEAN (Service Evaluation Answers Now) was born.
The SEAN bot enables employees to access FAQs and the exact section of the manual needed to move the interview process forward more accurately and quickly. By shortening the length of the interview and providing the correct determination of eligibility the first time, rather than requiring multiple visits. As a result, wait times are expected to reduce dramatically.
How did you build the chatbot, and why did you choose the Azure platform?
We already use Microsoft solutions and wanted to build on the systems we already had. It had to be a flexible, scalable, reliable platform, so we went to Microsoft partners to ask what suggestions they had. MAQ Software helped us decide on and implement a solution built on Azure with cognitive services, which gives us the grow-as-you-go infrastructure, platform, SaaS, and AI integration that DES needs. The level of confidence we have in this solution allows us to build, not just for today, but as an evergreen platform that will bring DES into the future.
What’s next for AZ DES in terms of using AI?
The majority of services provided by DES require eligibility evaluations and determinations, which inherently require the use of policies and manuals and regulations. We want to scale this solution to other services where the same kind of capabilities are needed.
Our goal is to expand the SEAN bot we built for SNAP across all divisions at DES, and eventually, create a citizen-facing bot, so they can access information in the policy manual and be more prepared when they come in for eligibility interviews. Technology has the unique ability today to help us leapfrog how we deliver services to our citizens.
We are looking at different possibilities to do that without having to change the underlying legacy systems immediately. Innovation with tech is our path forward.
Be sure to listen to our latest podcast and learn more about AI in social services.