核心决策来自一个洞察：纪念不应只发生在“特定日子”的强仪式里，而应融入日常可感知的轻交互——从“静态供奉”转向“多感官、可触发、可记录”的当代纪念。据此定下三个目标：让陪伴更持久、让纪念可被触摸、让 AI 更具情感。最关键的取舍在于 AI 的边界：用户既渴望“它还在”，又害怕 AI 的诡异（模拟说话、缺乏灵魂）。所以我们刻意不让 AI 模拟宠物说话，而是只还原它的样子与动作、加上可触摸的实体回应——用克制换真实，避开恐怖谷。
The insight: remembrance shouldn't live only in ritual days, but in everyday, senseable micro-interactions — moving from static offering to a multi-sensory, triggerable form. Three goals: lasting companionship, touchable memory, emotional-not-uncanny AI. The key trade-off was AI's boundary: people longed for presence but feared the uncanny — so we deliberately kept AI from faking speech, restoring only the look and motion plus tangible response.
