Slice

The first version of the brief was emotionally compelling but too broad: build an AI companion for people who feel alone. WHO research makes an important distinction between social isolation and loneliness, and neither should be treated as a condition a consumer device can simply solve.

I reframed the project around a narrower, designable breakdown. A fixed home companion loses the experiences that happen outside. A wearable can retain context, but offers little physical presence when the person returns. Family access may help, but can also remove the older adult's agency if it becomes passive monitoring.

Three scenario lenses guided the concept: an older adult living independently, a person whose mobility changes how they access screens, and a family member who wants reassurance without continuous surveillance. These are design hypotheses from desk research and scenario critique, not substitute user interviews.

I reviewed four adjacent patterns rather than treating every AI product as a direct competitor. Limitless focuses on searchable recall. Friend explores a portable social presence. ElliQ creates proactive companionship in the home. Replika demonstrates a persistent conversational identity without a dedicated physical body.

The useful gap was not a longer feature list. It was the relationship between portability and presence. Slice would need to move with the person, become more physically expressive at home, and make the transition between those modes understandable.

The concept became a service relationship with two bodies. Outside, the wearable needs to be light, direct, and explicit about capture. At home, the larger companion can use richer sound and physical expression, but it should receive only context the user has chosen to keep.

That reframing changed the design brief from 'make a friendly AI pendant' to three concrete responsibilities: preserve continuity, expose state, and keep the primary user in control of what reaches the home companion or family.

The architecture separates sensing from memory and memory from sharing. The wearable can support immediate questions with a short local buffer. The private memory layer holds only user-approved items. Slice Home receives a limited working context when the wearable is docked.

The dock is therefore not a passive charger. It is the visible boundary between mobile and home states. A transfer sequence can be paused, reviewed, or cancelled; docking disables ambient capture on the wearable and makes the active body unambiguous.

The orange metaphor turns an abstract ecosystem into a physical story. One segment travels with the user; the whole form stays at home. The asymmetry makes docking legible without arrows, screens, or setup instructions.

The wearable is thin, rounded, and deliberately visible. A reversible magnetic clip supports a lapel, bag strap, or pocket. Edge perforations distribute audio while a continuous light ring communicates state from more than one viewing angle.

Always-listening products create value and risk through the same mechanism. Slice separates off, temporary buffer, saved memory, and shared memory. Each state has a distinct light and sound behavior, and every retained item can be reviewed later.

Bystander awareness is treated as an interaction rather than a policy page. Explicit capture begins with a visible and audible cue. Sensitive locations can disable ambient capture, while direct commands can still be processed as one-off requests without entering long-term memory.

The next stage is a functional appearance model paired with a Wizard-of-Oz conversation service. The first study should test attachment comfort, state comprehension, interruption timing, docking clarity, and whether proactive prompts feel supportive or intrusive.

A second study should include family members and bystanders. Success is not conversation length. It is whether users can explain when Slice is listening, find and delete a memory, control what reaches family, and recover from a misunderstood prompt without anxiety.

The project now resolves the product relationship at three levels: a portable object for immediate access, a deliberate docking ritual for state change, and a private memory layer that makes retention and sharing inspectable.

What remains open is more important than another render. I have not yet demonstrated that people want a companion to move between bodies, that the light language is understood under stress, or that proactive interaction improves rather than interrupts daily life. Those questions belong in prototypes and longitudinal field research.