Lessons from public-sector AI deployment
Practical considerations from our partnership with Yuseong-gu, Daejeon for deploying AI care services in public institutions.
Deploying AI services in public institutions is fundamentally different from serving individual consumers. Our partnership with Yuseong-gu, Daejeon taught us five critical lessons.
Staff Turnover Is Inevitable
Public-sector staff rotate frequently. Every piece of operational knowledge must be documented and dashboard access must be separable from individual accounts.
Onboarding Is Mediated
Elderly residents don't onboard themselves. Caregivers or social workers handle the initial setup. Your UX must support this mediated onboarding flow.
Reports Drive Budget Renewal
Monthly operation reports with concrete metrics (sessions completed, mood trends, medication adherence) provide the evidence that institutional decision-makers need to justify continued funding.
Privacy Forms Must Be Customized
Standard privacy consent templates don't satisfy institutional requirements. Each organization has its own forms, and your consent flow must accommodate them.
Start Small
A pilot of 20–30 residents is the sweet spot. Large enough for meaningful data, small enough for detailed feedback collection.