Summary
The American Arbitration Association® (AAA®), the National Center for State Courts (NCSC), and Lancaster County Court of Common Pleas partnered to develop an AI-powered court diversion eligibility screener and case management system to support the Court’s high-volume consumer credit diversion program. This initiative aimed to streamline and support the administration of the Court’s diversion program – one that moves hundreds of credit card debt cases off the Court’s docket each year to a conciliation program that facilitates settlement between the parties – thereby expanding parties’ access to justice through practical, scalable technology.
Harnessing traditional and AI technologies, the system automates key administrative tasks—such as data entry and form generation— and applies an AI tool designed to mimic the judgment of a seasoned court clerk to screen volumes of documents for eligibility for the program and aid court staff in these processes. Built through close collaboration with court staff, the tool reflects a user-centered design approach grounded in real-world workflows.
The AAA is making a white-labeled version of the tool, complete with user guides, available for free through the NCSC to help other courts manage similar challenges. Courts interested in exploring how this technology may benefit their high-volume docket should reach out to NCSC via their website.
Key Takeaways
1. Smart, Targeted Technology Makes a Big Difference
Rather than overhauling entire systems, the project focused on specific, manual pain points—like data entry, eligibility review, and document generation. That clarity and focus enabled the AAA to design and build a solution responsive to the court’s needs and supportive of court staff.
2. People-First Design Drives Effective Innovation
The system was shaped by directly observing and collaborating with Lancaster County's court staff, especially the program coordinator. This human-centered approach ensured the technology truly addressed the day-to-day challenges faced by users, enhancing both adoption and effectiveness.
3. Replication and Access Are Built In
Recognizing the scalability of this model, NCSC is prepared to extend a white-labeled version of the tool to more jurisdictions beyond Lancaster County. Interested courts can reach out to NCSC to determine eligibility.
Final Thoughts
This project is a powerful example of how intentional, collaborative innovation—anchored in real-world needs—can elevate access to justice. By balancing automation with human judgment and offering a free, adaptable toolkit, AAA and its partners are paving the way for scalable, sustainable court modernization.
Transcript
Zach Sarno: Hey there. Welcome to another Tiny Chat. I'm Zach Sarno. That's Danielle Hirsch. We're from the National Center for State Courts, and today we're thrilled to be joined by some friends from the American Arbitration Association. We'll speak to them in a moment. Just so you know who we're talking with, we have Walling Almonte, who's a Senior Project Analyst, Diana Didia, who's Senior Vice President, and CIO, Jen Barnett, who's Senior Counsel, and Yogesh Chaudhari, who is Vice President of Software Development and Operations, and we partnered with them to build something really exciting that you're going to learn about right now.
Danielle Hirsch: So Diana, tell us why AAA was interested in working with the National Center and the Lancaster County Court to assist with their Consumer Credit Card Debt Diversion Program.
Diana Didia: Sure, I think at the AAA, we're always looking for ways to leverage our ADR expertise and our technical expertise to expand access to justice in meaningful ways. It's really at the heart of our mission. And you know what? We're, we're really, we really believe in the expansion of ADR and applying it to access to justice. So when the National Center for the State Courts reached out to us to collaborate on this opportunity with Lancaster County, we really felt that it aligned very well with both of our capabilities and our values.
Also, it's interesting, like the more we learned about what was happening at Lancaster County that they had set up this credit card diversion program, and that it was achieving a lot of success in preventing a large number of credit card debt cases from reaching the court docket. But it was interesting to learn that really, that whole process was manual, and although the administrator was doing a fabulous job that really it, you know, applying a technological solution to what she was doing, you know, would really help with making it a lighter process, and then ultimately having what you know, it'd be more scalable, not just in Lancaster County, but may it may be applicable to other court jurisdictions. I also think that the timing was really, was really great, because for us, we were just starting to implement, we were on our own AI learning journey, and we were just starting to implement AI into our own workflows and processes. And so we felt like, you know, this opportunity that we could actually see how what we were learning could be put to good use in the real world, was exciting for us. And you're going to be speaking to some of my team members here, and we really had a great time going out and meeting the Lancaster County administrator and learning what she was doing, and then seeing how we could really help with, you know, apply our expertise to make her process easier for her to administer.
Zach Sarno: And you know, to that end, Yogesh, you and the team helped build this court diversion eligibility screener tool to address some of the issues Diana was mentioning. And maybe you can tell us a little bit more about the case management, the human review procedure, protections, how does it all work?
Yogesh Chaudhari: Yeah, when we met the administrator that Diana mentioned that we saw that they had the well-defined process and the rules established, but the most of the work was done manually, and they're keeping into the spreadsheet color coded statuses. And we thought that they, if you can just build a small case management system, it will help them to ease their work. So we started building this new system. First thing we saw that they need to bring the data from their main case administration system into the their work. So we incorporated the smart copy and Smart Paste so they don't have to type in every information. Then all the information is gathered. Then the next thing we thought about of using the AI in processing the documents, because all the evidences and the supplement documents supported, they had, as I mentioned, that they had the well-established rules, so it was easy for us to implement those and ask AI to do the work for us, rather than manual review of the documents. So that's one more thing we did. Third thing, we thought that they had to send lot of communication to the parties time to time based on the milestones, and that's where we incorporated the mail merge system into the system that helped them to just one click merging the letter for all their needs. So all this putting together the administrator feel that it's going to ease their work drastically.
Danielle Hirsch: Super. Well, a program like this probably taught you some lessons, and we're curious about some of those lessons related, both in terms of large language models, but also helping in a high-volume court docket. So Walling, could you share?
Walling Almonte: Sure. Yeah, there were a lot of takeaways. Some were technical. Some about the importance of not relying on only on technology. But from a technology perspective, we learned that large language models are only as good as their prompt engineering. So, for the eligibility screener, we had to strike a balance between specificity and flexibility. The documents we received were varied widely in format. So, we created a prompt that instructed the AI to behave very similar to a court clerk checking for the exact legal criteria, drawing from multiple documents and explaining its reasoning with citations. So that was an iterative process that taught us a lot about how to guide the LLMs for high-stakes use cases, but maybe more importantly, we were reminded that innovation only works if it starts with the people who use the technology and when we take the time to truly listen to them. So once our team went to Lancaster County in person, we watched how the coordinator was working, what she was clicking on, what information she had to enter from scratch every time she created a new document, and where she paused to double-check things. So that pretty much grounded everything we built. So, the system wasn't designed in a vacuum. It was shaped directly by her real-world pain points she was suffering. And we also saw that small focus interventions can have an outsized impact when dealing with high-volume caseloads. So, by targeting just a few friction points, data entry, document review, form generation, a lot of different things that she was doing, we were able to reduce a lot of manual labor without needing a huge it overhaul. So, we found that that was very encouraging for other courts that don't have large technology budgets or teams.
Zach Sarno: Good advice in there and Jen turning to you, I'm sure that there's some other lessons learned. Because while this partnership was designed to assist the Lancaster County Court, there's also interest in sharing this more broadly with the court community. So, for those courts that are interested, what other advice do you have in doing something similar?
Jen Barnett: Sure, I think I have three main takeaways. Some of them have already been touched on a little bit, and I'm going to save the most exciting one for last. But I would say the first piece of advice would be to just do it. Take your first step. You don't need to undertake a large, complex technology overhaul of an entire system to make an impact. Here we focused on a discrete project. Lancaster County had effectively one and a half people brilliantly managing hundreds of cases through a really innovative diversion program. But the program was very manual, and because of that, it presented for us an opportunity to take advantage of discrete and targeted technology. That clarity and focus really helped ensure that we designed a solution that would be practical, achievable and responsive to their needs, but it also helped us take it on a project that was doable within our staffing and time availability.
Next, I would say collaboration. And I know Walling touched on this. Collaboration was key. The system wasn't designed in a vacuum. Rather, you know, collaborating closely with the people who are actually doing the work at the courthouse was essential to designing a solution that spoke to the user's actual pain points. The design was inspired and shaped by being in the room at the courthouse, observing, listening to the program coordinator, and then involving her continuously throughout the design and testing phases. That level of engagement made it possible for us to build something that I think all of us really think is better than what we initially envisioned at our first meeting when we started talking about the project.
And then lastly, and something we're very excited about - we are going to be making a white-labeled version of this full system, including the eligibility screening tool and the case management system, along with user guides, available on the National Center for State Courts’ website, so that way any court can access the technology and adapt it to their specific needs free of charge.
Zach Sarno: Fantastic. Well, thank you all for all of your efforts on this one. It's been a team effort, and it's really great stuff. We're excited to be working with you, and excited to be working with the broader core community share what we've learned. Thanks for joining us, and thank you again for the partnership.
Diana, Walling, Jen, and Yogesh: Thank you.
Danielle Hirsch: Thanks again to the AAA for all the great work that they've done. As Jen mentioned, we are working with the AAA to make a white-label available of the tools that were discussed. If you're interested in working with us to figure out how to make your high-volume docket work better and possibly use this technology for free, please reach out to us and there will be details on our website. Thanks again and take care.