Memory Experience
Designing visibility and structure for AI-generated memory.
2024 | Kin Ai
Time frame: 1.5 month
My roles: UX, and UI designer
Contextualizing
As users interacted with Kin through conversations and journaling, the system continuously generated memories: capturing people, activities, interests, and key moments over time.
As this information accumulated, it became necessary to provide users with a clear way to understand what was being stored, when it was created, and how different memories related to each other.
Problem
AI-generated memories were created in the background with limited visibility to users, making it difficult for them to build a clear mental model of what Kin remembered about them. Reviewing memories as isolated items lacked sufficient context, and as the volume of stored information grew, understanding relationships and timelines became increasingly challenging.
Design Goal
Design a memory experience that:
-
Makes AI-generated memory visible and understandable
-
Organizes memories over time
-
Helps users build trust through transparency
-
Avoids overwhelming users with complexity or control too early
Early exploration focused on mapping how memories, entities, timelines, and relationships could work together as a cohesive system—prioritizing clarity, temporal context, and sense-making before introducing interaction or control.
Solution Overview
A dedicated Memory Experience was designed to organize and surface everything shared with Kin, combining temporal structure with relational context.
The experience is built around three complementary layers:
-
A day-based timeline, focused on when things happened
-
A graph view, focused on how memories and entities connect
-
An entity detail view, allowing users to tap into a specific node to see related information
From the timeline, users can move between days to review messages, journal entries, and newly created memories. The graph view provides a high-level understanding of relationships, while tapping on a node reveals additional context about that specific entity, such as related memories and mentions.
This layered approach helps users move from overview to detail, supporting sense-making without overwhelming them with control or complexity.
Role of AI
AI is responsible for detecting entities, generating memories, and establishing relationships based on conversation and journaling data.
The interface’s role is to translate this abstract system into something users can see and understand without interpretation, judgment, or rewriting of user content.
Design Considerations
Balancing transparency and cognitive load was essential. The experience needed to explain AI memory clearly without overwhelming users or requiring them to manage it prematurely.
Sequencing also played an important role: starting with visibility before introducing control helped users build trust and confidence in how memory works.
Key Considerations
Designing memory for a conversational AI required balancing transparency with cognitive load. Making AI-generated memory visible was essential to help users build trust and a clear mental model of how Kin works, especially as the system operates automatically in the background.
Structuring memory around time and relationships helped users understand not only what was remembered, but when it happened and how different pieces of information connected. This contextualization was critical to avoid memory feeling fragmented or opaque.
Another key consideration was sequencing. Prioritizing visibility and sense-making before introducing interaction or control allowed users to first understand the system, laying the groundwork for future iterations involving editing, correction, or deeper interaction with AI memory.
Learnings
Designing the Memory Experience helped me better understand how AI memory is structured and formed over time. Translating an abstract system (entities, relationships, and timelines) into a visual interface forced me to deeply reason about how information is captured, connected, and surfaced by the AI.
This project highlighted how much clarity can be gained through visualization. By designing this page, I was able to uncover gaps, assumptions, and opportunities in how memory was being generated, which wouldn’t have been as visible without putting the system on screen.
It also reinforced the importance of designing for transparency in AI-driven products. Making memory visible not only benefits users, but also helps designers and teams build a clearer shared understanding of how intelligent systems behave.