# Persistent Memory Turns Health Metrics Into Coaching Context ## Claim Persistent memory makes health metrics actionable by interpreting today's measurement against the person's recent history. <!-- atomicity: 1 idea ✓ --> ## Scope & conditions - This holds when the agent retains relevant events and preferences across sessions instead of treating each measurement as an isolated score. - Memory doesn't make the advice medically valid; it improves continuity, not clinical authority. - The source is a sponsored hobbyist build, so it demonstrates feasibility rather than health outcomes. ## Evidence [[3. Resources/Readwise/Articles/I Connected Apple Health to an AI Agent... Here's What Happened]] shows an agent using prior illness and repeatedly ignored workout suggestions to interpret current activity rather than penalizing every low count identically. ## Links - [[Long-Horizon AI Shifts the Bottleneck From Model Capability to Memory and Effort Scaffolding]] — applies-in: health coaching becomes longitudinal only when the harness preserves relevant state across sessions. - [[Retrieval Quality Is Set by the Unit of Knowledge Not the Retrieval Algorithm]] — because useful health interpretation depends on retaining events and preferences at a unit the agent can retrieve accurately. - [[AI Literacy Is Relational Not Just Technical]] — contrasts: continuity can strengthen reliance, which makes boundaries around medical authority and trust more important. - [[3. Resources/MOCs/AI & the Future of Knowledge Work MOC]] — applies-in: the MOC lists this note under Amplifier thesis & the harness.