Peer-run behavioral health organizations offer holistic wellness support by combining mental health services with housing, employment, and income assistance. However, staffing and expertise limitations hinder behavioral health organizations, making meeting all service user needs difficult. To address this challenge, we partner with a peer-run behavioral health organization to co-design a large language model (LLM)- based application to support peer providers. We co-design the application through comic boarding, a narrative-driven co-design method, to explore peer providers’ and service users’ perspectives on the opportunities and potential risks. We then propose PeerCoPilot, an LLM-powered assistant that helps peer providers create wellness plans, construct step-by-step goals, and find resources for these goals. We conducted human evaluations with 15 peer providers and 6 service users and found that both groups overwhelmingly supported using PeerCoPilot. PeerCoPilot is now used by a group of peer providers at CSPNJ, a large peer-run behavioral health organization serving over 10,000 service users.
Naveen Raman is a PhD student in the Machine Learning department at Carnegie Mellon University. His research focuses on improving decision-making in human environments, including topics such as human-AI collaboration, explainability, and sequential decision-making. Naveen is funded by an NSF GRFP Fellowship and he has deployed his research to domains including food sustainability and behavioral mental health. Previously, he obtained an MPhil from the University of Cambridge and a BS from the University of Maryland, and has received awards including the Churchill Scholarship and a finalist for the CRA Undergraduate Researcher of the Year.