AWS Public Sector Blog

Transforming Resident Outreach using next-generation AI capabilities

AWS Branded Background with text "Transforming Resident Outreach using next-generation AI capabilities

“I need to submit documents to support my application, but I can’t take time off work to visit an office.”

A common scenario highlights a critical challenge facing state agencies today: how to make government services more accessible while reducing administrative burden. Citizens commonly need to submit documents to a state or local agency, but they can’t take time off to visit an office in person. Citizens increasingly expect experiences comparable to private sector organizations: digital-first and mobile-first.

Deloitte’s Resident Outreach solution (available in AWS Marketplace) enables state agencies to adapt their resident communications through the power of generative AI on Amazon Web Services (AWS).

In this post, we discuss how Deloitte empowers the four capabilities of Resident Outreach:

  • Multi-Modal Completion of Applications & Renewals
  • Application Nudges
  • Correspondences: Personalized, context-rich communication
  • Campaigns: Enhanced citizen messaging

Multi-Modal Completion of Applications & Renewals

Applicants frequently encounter obstacles in fully completing applications, stemming from challenges such as complex application procedures or language barriers. Incomplete applications necessitate follow-up communication, sapping an agency’s limited resources and resulting in eligible applicants not receiving their benefits or false negatives in application processing.

Generative AI multi-modal capabilities refer to the ability of AI systems to:

  • Ingest, process, and generate content
  • Enable omnichannel (i.e. mobile app, instant messaging, web, phone, and paper)
  • Interact through multiple formats or modalities (such as text, voice, images, and video)

Using AI models to understand and process content allows us to provide enhanced virtual agent capabilities. AI-based virtual agents orchestrate outreach to the population to verify eligibility or other key details. They do so through personalized, natural conversations.

Omnichannel, mobile-first (but not mobile only) design improves user experience by making it straightforward to switch between web browsers, apps, SMS, email, or voice-based assistants across different devices so users can interact via typing, speaking, or scanning documents.

Application Nudges

“We noticed you haven’t uploaded your income proof. Need help? Just click here to get assistance…”

Applicants sometimes submit incomplete applications leading to missed deadlines for crucial benefits. This can result in delayed or denied assistance for those who would otherwise be eligible.

Generative AI models, like those provided by Amazon Bedrock or Amazon SageMaker, can resolve this problem by creating highly personalized nudges based on individual user behavior, past interactions, and application history. Instead of generic reminders, users receive tailored messages that resonate with their specific needs. For example:

“Hi Jane, it looks like you’re halfway through your benefits renewal. Just two more steps to go!”

Application Nudges builds upon the multi-modal design philosophy we’ve described with the system initiating nudges to citizens for:

  • Application completion
  • Renewal reminders
  • Application processing and benefits status
  • Opt-in and appointment scheduling

Generative AI, combined with Amazon Translate and Amazon Polly, generates nudges in multiple languages and formats to cater to diverse user bases, providing accessibility for non-English speakers, visually impaired users, and those with other accessibility needs.

Correspondences: Personalized, context-rich communication

Agencies can prevent confused beneficiaries from calling or creating unnecessary workload by communicating clearly in an accessible manner.

Constituents are less likely to require customer service support when correspondence establishes:

  • A clear understanding of where constituents are in the process
  • Expected outcome
  • Expected timeline
  • Next steps
  • How constituents can take action on their own
  • (Opt-in) Notifications of progress or delays
  • Proactive Application Nudges (refer to the preceding section)

Deloitte is using AWS to streamline and enhance correspondence in the following ways:

  • Custom, simplified explanations using QR codes: QR codes have become a norm for accessing more information. QR codes are generated specific to a resident’s case, such that generative AI references the correspondence type and creates summaries customized to an individual’s unique circumstances. Summaries can even be videos in their preferred language, with specific instructions for each household member, and a call to action on next steps.
  • Audio-enabled correspondence for visually impaired residents: Generative AI combined with text-to-speech capabilities such as Amazon Polly convert written correspondences into audio format, which gives visually impaired individuals or residents who prefer auditory information the choice to listen to their messages. Audio isn’t limited to web interfaces or mobile apps. QR codes on printed correspondences link directly to an audio version, providing greater accessibility. For example, “For an audio version of this notice, scan here to listen.”
  • Summarized correspondences for older adults: Who doesn’t struggle with legal jargon or large amounts of information in a single letter? Concise generated summaries of complex documents simplify bureaucratic language into digestible formats. This is especially useful for older adults or those that care for them. For example, providing additional instructions for caregivers or agents of how to act on behalf of an elder.

“This notice explains changes in your healthcare plan. Here’s what you need to know and the next steps to take…”

Campaigns: Enhanced citizen messaging

Campaigns are one-way or two-way interactive (multilingual) conversations. Two-way campaigns provide an intuitive self-service solution that consist of messaging, voice calls, and other enhanced communication across agency programs.

Personalized messaging at scale

Generative AI crafts personalized messages tailored to individual recipients based on their demographics, behavior, preferences, and interaction history. Citizens receive highly relevant information. For example, you can enable an agency to connect only with households within a certain radius of a nearby office, sharing specific instructions relevant only to a local population:

“Hi Jane, this is a reminder to prepare for the upcoming hurricane season. You can pick up an emergency kit at the nearest distribution center located at…”

As part of an omnichannel approach, mass personalization includes voice interactions. First generate dynamic, conversational voice scripts tailored to specific recipients. Then use text-to-speech tools such as Amazon Polly to create engaging natural-sounding voice messages.

Intuitive analytics

Generative AI processes large datasets to produce near real-time reports on key metrics like open rates and overall impact. It predicts future campaign success by analyzing past data and resident behavior patterns. Using natural language processing, it gauges public sentiment from various feedback sources, enabling agencies to adjust strategies accordingly. Generative AI transforms complex data into visual reports, which campaign managers can use to identify trends, address issues, and communicate results to stakeholders. This comprehensive approach enhances decision-making and improves the efficiency of outreach efforts.

Enabling AWS services

Capability AWS service
Access LLM foundation models Amazon Bedrock
API management Amazon API Gateway
Marketing communications Amazon Pinpoint
Contact center Amazon Connect
Serverless compute AWS Lambda
Speech recognition Amazon Transcribe
AI voice generator Amazon Polly
Chat-based assistant (conversational interfaces) Amazon Lex
Text recognition or optical character recognition (OCR) Amazon Textract
Natural language processing Amazon Comprehend

Conclusion

Deloitte’s Resident Outreach is a set of solutions designed for government agencies to improve customer engagement, provide proactive nudges to users, and tailor personalized communication content and effective campaigns. These technologies help agencies overcome challenges in multilingual support, automated application processing, and citizen engagement.

State agencies must embrace innovative technologies to meet evolving resident expectations. Deloitte’s Resident Outreach uses AWS services to provide a solution that:

  • Enhances accessibility and inclusivity in communications
  • Streamlines application and renewal processes
  • Provides timely, personalized information
  • Improves operational efficiency
  • Helps gain insights to continually improve service delivery

The future of resident outreach lies in the intelligent adoption of AI technologies to build stronger relationships with residents, enabling better outcomes with efficient use of public resources.

Contact Deloitte Resident Outreach to start your journey to improve resident outreach through AI-powered solutions. To learn more about generative AI at AWS, contact your AWS account team or the AWS Public Sector team.

Jack Zhou

Jack Zhou

Jack is a senior solutions architect at AWS in Worldwide Public Sector, empowering global consulting partners to build long-term, profitable businesses with AWS. Previously, he was a senior technical account manager managing high-profile enterprise support accounts.

Robin Geddes

Robin Geddes

Robin is a senior solutions architect at AWS in Worldwide Public Sector, focusing on Oracle technology and applications. Previously he was a migrations specialist. He has a passion for highly-automated hands-off environment management to support large projects and high-throughput systems.

Sam Loewner

Sam Loewner

Sam is a senior manager in Deloitte’s Government and Public Services (GPS) sector and leads business process outsourcing and contact center solutions for state and local agencies. He specializes in improving customer experience through cloud-based technologies and service optimization.

Tony Pagnusat

Tony Pagnusat

Tony is a senior partner account manager at AWS, specializing in solutions and strategic growth opportunities. In his current role, Tony collaborates with global system integrators to develop and implement scalable solutions for public sector customers worldwide. His expertise lies in creating repeatable frameworks and driving global strategic initiatives.