A caseworker helping a family apply for government benefits is doing three jobs at once: identifying the family’s needs, finding appropriate programs and support, and helping the family fill out a potentially burdensome application. The family needs support that works. The caseworker needs to spend their time on the person in front of them, not the paperwork around them.
Both are working inside a system under real strain. Funding is getting tighter, states face financial penalties for payment error rates, and families and staff need to navigate new Medicaid and SNAP work requirements. Meanwhile, old challenges remain: long applications, documents to track down, and waits on the phone for support from a human.
These are the kinds of problems Nava has been working to address for years. Now, we’re doubling down on our commitment to identifying innovative solutions with our new Technologist-in-Residence program. Through this program, our team will explore the potential of AI and other technologies to support staff and families.
Nava Labs uses philanthropic funding to prototype innovations to the social safety net that government agencies could benefit from but can’t fund directly. The Technologist-in-Residence program gives us the space to build and ship public prototypes. This lets the team test emerging approaches to government service delivery in small, safe ways, instead of risking big rollouts to the public or staff.
The team consists of a designer/researcher, an engineer, and a product manager. That’s enough to take a problem from field research to a working prototype, and small enough to stay close to the people the work is for. We’ll draw on Nava’s broader work across state and federal benefit programs, and we’ll learn from evaluation partners like the Better Government Lab. This will enable us to test with rigor and ground prototypes in how these systems actually run.
Building on proven experimentation
Nava Labs has spent several years testing whether and how AI can help caseworkers do their jobs. These tools — brought together as our Caseworker Empowerment Toolkit — are the foundation that our Technologists-in-Residence will build upon. We’ll prototype using findings we can trust, failures we’ve already learned from, and steady feedback from the people doing the job.
The most developed tool is our Assistive Chatbot, which answers caseworkers’ questions about benefit eligibility in real time. A randomized controlled trial with 125 caseworkers found it improved accuracy by an average of 40%, with the biggest gains on the hardest questions and among newer staff. Our real-world pilot was messier because uptake varied from site to site. Nevertheless, we gained valuable insights from this pilot, both where it didn’t take hold and where it did.
We’re also piloting an agentic Form-Filling Assistant, a Referral Generator for local resources and programs, and a document analyzer that flags problems before they turn into a delay or a denial.
Each new experiment our Technologists-in-Residence conduct will build on our past experience, starting from a problem we’ve already seen in the field.
Working in the open
We’ll publish demos, write about what we learn, release open source components, and talk about the misses alongside the wins. We believe the field benefits when teams show their reasoning out loud instead of only publishing the polished result.
If you build technology for the public sector, our findings and open source components are yours to use. If you fund this kind of work, our experiments are a window into what’s worth backing before a grant cycle forces the call. If you research service delivery, our evaluations and methods are out in the open. And if you’re a caseworker or run a program, you’re the reason these experiments exist, and your problems are the ones we want to hear about.
So reach out to tell us what challenges or opportunities you face. The most useful prototypes start with a real pain point or goal from the people using and delivering services. Email us at labs@navapbc.com.
The team
Ryan Hansz
Ryan Hansz is a senior designer/researcher at Nava. He has nearly a decade of experience across the public and private sectors, including work with Bloom Works, the New York City Mayor’s Office of Economic Opportunity, and Kaiser Permanente.
Cory Trimm
Cory Trimm is a principal engineering manager at Nava. Nearly a decade at startups and four years embedded at the U.S. Department of Veterans Affairs with the U.S. Digital Service led him here, and he is currently experimenting at the intersection of applied AI and government.
Diana Griffin
Diana Griffin is a principal product manager at Nava. She has spent fifteen years working in social impact and civic tech, including startups focused on access to benefits and education, as well as public sector projects in state and federal benefits delivery.
