Speculative AI in iMessages
ROLE
UI Design
Prototyping
TEAM
3 Designers
TIMEFRAME
May 2026
2 weeks
TOOLS
Figma
Protopie
OVERVIEW
A concept exploration of AI-native features for iMessages
Three features that help you keep up with plans when life doesn't slow down
This was a two-week school project exploring how Apple Intelligence could extend iMessage beyond message delivery, into understanding what conversations actually mean. The concept focuses on a single persona: someone who context-switches constantly, reads texts in passing, and loses plans in the noise.
PROBLEM
iMessage treats every message the same
A meme and time-sensitive plan look identical in your thread list
The existing system only tracks whether you've read something, not whether you've dealt with it. Once you open a message, iOS marks it as read.
But reading and responding are different things, and confirmed plans buried in a thread have no easy way of becoming something more permanent.
PERSONA
Jane—busy, well-intentioned, constantly context-switching
A 24-year-old graduate student balancing a demanding academic workload, a part-time campus job, and an active social life. Constantly moving between classes, study sessions, and outings, her attention is endlessly divided among immediate, competing priorities.
→
Busy and constantly context-switching
Reads texts in passing and loses the thread before she can act on it
→
Well-intentioned but inconsistent follow-through
Makes plans enthusiastically over text, never adds them to a calendar, assumes she'll remember
→
Reactive, not proactive
Responds when it's convenient, which sometimes means never
THE CONCEPT
Three features and a unifying direction
Follow up detection notices; semantic search finds; live events capture
Plans made over iMessage rarely happen in one thread, one day, or one straight line. For our persona Jane, they drift across conversations, get half-confirmed, and live in the back of her mind until Saturday arrives and she forgets if she actually sorted it out.
Follow-up detection notices when a time-sensitive plan/an open loop you read and meant to close goes unanswered, and keeps it quietly visible
Semantic search finds the context that's half-remembered, not by keyword but by meaning
Live events catches the moment a plan gets confirmed and turns it into a calendar entry without leaving the conversation
FULL FLOW
The full system in one conversation
Ananya texts on Wednesday asking if they're still on for Saturday. Alex reads it, forgets to reply, and can't remember which plan she even means. Follow-up detection catches the open loop. Semantic search finds the context. Live events captures the confirmation.
REFLECTION
Next steps and what I learned
Evaluate the tension between AI convenience and privacy: would this concept be utilized by users and fit in with Apple's design principles?
The biggest question is whether people would trust AI that reads their conversations, even when the outcome is useful. While this project was focused on high-fidelity prototyping, if I were to do this project over again, I'd focus on doing rapid concept evaluation with users.