End-of-Life Service Design

Designing a bereavement service where AI quietly manages paperwork while people are given physical space to grieve.

Sensitive research

After a loved one dies, next-of-kin are handed two conflicting jobs at once: grieve, and simultaneously manage a maze of paperwork, accounts, and logistics they've likely never navigated before. Our team set out to design a service that could ease that second burden without displacing the first. This required intensive, sensitive research; we grounded the project in secondary research, semi-structured interviews, and a co-design session where participants worked through empathy mapping and creative exercises to talk through death, grief, and legacy.

Making sense of grief

Affinity diagramming across our data surfaced four core insights: the tension between emotional and administrative needs, a broader cultural lack of "death literacy," a desire to honor a life rather than just process its end, and a cautious opening for AI to absorb logistical burden without touching the human parts of grief. We mapped the full ecosystem of touchpoints, from the first 24 hours after a death through months of estate work, into a service blueprint and user journey map.

Service blueprint for grief center and AI assistant

Two ways to take care

That research led to To-Gather: a hybrid service pairing a physical Center for Life & Death with a companion app. I led design of the Center itself — a space meant to hold grief counseling, death planning, and paperwork processing under one roof, with dedicated areas for group therapy and community events, native greenery and outdoor space for reflection, and infrastructure built for staff who absorb much of the community's emotional load. The companion app, To-Gather, uses AI as a backstage tool — OCR-based document handling, task triage, a dynamic calendar for deadlines — deliberately kept administrative rather than emotional, with next-of-kin retained as the final approver on anything AI surfaces.

Screens for death-related administrative management

Challenges and shortcomings

AI guardrails. Prototype feedback clarified that any AI misstep within grief care could be read as cold and deter interaction. The AI's role had to stay legible, consensual, and strictly administrative, never a stand-in for human judgment or legal authority.

An emotional rollercoaster. We underestimated how disorienting the shift from emotional to logistical content would feel inside the product itself; our first narrative prototype moved users from a grief scenario into task management too quickly.

Split co-design. Our team's first co-design session, while valuable, was run separately by different facilitators. A single combined session with shared facilitation would have let the whole team observe participant reactions firsthand rather than relying on notes.