Making message retrieval seamless with AI
Apple Intelligence
Search+
ROLE
Product Designer
TIMELINE
48 hours
COLLABORATORS
Nydia Kushta
Nydia Kushta
Shaina Desroches
Shaina Desroches
Yuwei Huang
Yuwei Huang
Adela Milkes
Adela Milkes
Colin Mendoza
Colin Mendoza
SKILLS
Artificial Intelligence
Artificial Intelligence
User Research
User Research
Interaction Design
Interaction Design
Visual Design
Visual Design
Design Systems
Design Systems
THE CONTEXT
THE CONTEXT
Searching messages should feel natural.
Apple’s current message search requires users to remember exact words, making it difficult to find old texts based on memory or topic. We set out to design an AI-powered search experience that understands intent, allowing users to find conversations, topics, and attachments with simple, natural language. By making search smarter and more intuitive, messaging becomes more fluid, personal, and effortless.
Apple’s current message search requires users to remember exact words, making it difficult to find old texts based on memory or topic. We set out to design an AI-powered search experience that understands intent, allowing users to find conversations, topics, and attachments with simple, natural language. By making search smarter and more intuitive, messaging becomes more fluid, personal, and effortless.
THE CHALLENGE
THE CHALLENGE
Bridging the gap between human memory and machine search.
Bridging the gap between human memory and machine search.
The current iOS Messages search tool depends on exact keywords, which doesn’t match how people naturally recall conversations. We often remember moments by feeling, topic, or context, not by specific phrases. This mismatch makes searching old messages time-consuming and frustrating. To address this, our team set out to design an AI-powered search feature that interprets natural language queries while maintaining Apple’s standard of privacy and simplicity. Our challenge was to make the experience feel effortless, trustworthy, and consistent with iOS design principles.
USER RESEARCH
USER RESEARCH
Understanding how people actually search for memories.
Understanding how people actually search for memories.
Before designing solutions, we wanted to understand how users currently search their messages and where frustration begins. We surveyed 20 iPhone users aged 18 to 50 to learn how they find old messages and manage information. We then identified key themes that shaped our design direction.
Before designing solutions, we wanted to understand how users currently search their messages and where frustration begins. We surveyed 20 iPhone users aged 18 to 50 to learn how they find old messages and manage information. We then identified key themes that shaped our design direction.
![](data:image/svg+xml,<svg display="block" role="presentation" viewBox="0 0 20 20" xmlns="http://www.w3.org/2000/svg"><g d="M 8 5 C 9.657 5 11 6.343 11 8 C 11 9.657 9.657 11 8 11 C 6.343 11 5 9.657 5 8 C 5 6.343 6.343 5 8 5 Z M 8 0 C 12.418 0 16 3.582 16 8 C 16 12.418 12.418 16 8 16 C 3.582 16 0 12.418 0 8 C 0 3.582 3.582 0 8 0 Z" fill="transparent" height="16px" id="iM0VeY9zZ" transform="translate(2 2)" width="16px"><path d="M 3 0 C 4.657 0 6 1.343 6 3 C 6 4.657 4.657 6 3 6 C 1.343 6 0 4.657 0 3 C 0 1.343 1.343 0 3 0 Z" fill="transparent" height="6px" id="Xq85pCCxz" stroke-dasharray="0" stroke-linecap="butt" stroke-linejoin="miter" stroke-miterlimit="4" stroke-width="2" stroke="rgb(182, 181, 185)" transform="translate(5 5)" width="6px"/><path d="M 8 0 C 12.418 0 16 3.582 16 8 C 16 12.418 12.418 16 8 16 C 3.582 16 0 12.418 0 8 C 0 3.582 3.582 0 8 0 Z" fill="transparent" height="16px" id="sXmfua_s8" stroke-dasharray="3.5" stroke-linecap="round" stroke-linejoin="round" stroke-width="2" stroke="rgb(182, 181, 185)" width="16px"/></g></svg>)
🌴
🤒
🗓️
👟
🚌
check out some user feedback we received
🌴
casual language
🍽️
memories
💡
privacy
🌴
🤒
🗓️
👟
🚌
check out some user feedback we received
🌴
casual language
🍽️
memories
💡
privacy
People who have given up trying to find a message after multiple failed search attempts
0%
0%
People who have given up trying to find a message after multiple failed search attempts
0%
0%
People who screenshot or pin texts as a backup system
0%
0%
People who screenshot or pin texts as a backup system
0%
0%
Omari (22, International Student)
The Meme Archivist
Core needs
Search that handles code-switching, slang, and mixed language phrases.
Visual search results: quick previews of memes, photos, GIFs.
Wants a clean, fast experience, less typing, more results.
Frustrations
Endless scrolling breaks up conversation flow.
Has to screenshot messages so they don't get lost.
USER PERSONAS
USER PERSONAS
Designing for different kinds of communicators.
Designing for different kinds of communicators.
We created user personas to understand the diverse ways people interact with their messages. Each persona represented unique goals, frustrations, and behaviors that shaped how we approached search, memory, and personalization. Our personas revealed that users vary widely in how they recall and organize information. Some value structure and precision, while others rely on visual cues or conversational memory. Recognizing these patterns allowed us to design a search experience that flexes to different mental models rather than forcing users to adapt to the system.
Olivia (37, Marketing Manager)
The Over-Organizer
Core needs
Wants AI to group messages by topics (e.g. “wedding planning,” “travel ideas”).
Needs quick access to photos, receipts, and links shared months ago.
Values smart filters that surface relevant details automatically.
Frustrations
Wastes time scrolling through mixed media and unrelated results.
Finds Apple’s current search too rigid for her multitasking workflow.
KEY FEATURES
KEY FEATURES
Building a smarter, more transparent search experience.
Building a smarter, more transparent search experience.
Our design focuses on making AI-powered search feel natural, transparent, and distinctly Apple. We used Apple Intelligence branding and native design components to build familiarity and trust. A visible feedback loop shows how the AI interprets each query, turning abstract language into clear, searchable concepts. By understanding context instead of keywords, the feature recognizes slang, abbreviations, and tone. Guided prompts like “describe your conversation…” encourage users to explore and discover with confidence.
Our design focuses on making AI-powered search feel natural, transparent, and distinctly Apple. We used Apple Intelligence branding and native design components to build familiarity and trust. A visible feedback loop shows how the AI interprets each query, turning abstract language into clear, searchable concepts. By understanding context instead of keywords, the feature recognizes slang, abbreviations, and tone. Guided prompts like “describe your conversation…” encourage users to explore and discover with confidence.
IDEATION
Drafting a smarter, context-aware iMessage search.
Drafting a smarter, context-aware iMessage search.
To make searching iMessage more intuitive, we imagined a flow where users could type naturally and let the system understand their intent. By parsing context, tone, slang, and language variations, the AI can connect the query to related messages, photos, documents, and links. The goal was to design a feature that organizes results clearly and reduces the mental effort of searching, making the process feel effortless while keeping the interface clean and aligned with Apple’s design principles.
To make searching iMessage more intuitive, we imagined a flow where users could type naturally and let the system understand their intent. By parsing context, tone, slang, and language variations, the AI can connect the query to related messages, photos, documents, and links. The goal was to design a feature that organizes results clearly and reduces the mental effort of searching, making the process feel effortless while keeping the interface clean and aligned with Apple’s design principles.
FINAL DESIGNS
FINAL DESIGNS
Bringing Apple Intelligence to life.
Bringing Apple Intelligence to life.
The hi-fi prototype highlights the polished interactions and visual design of Apple Intelligence. Users can see how natural language queries are handled, results are organized, and the interface responds seamlessly, demonstrating how our design principles translate into an intuitive, engaging experience.
Here we have Emma, who can't remember the name of that Mexican restaurant she went to last week, but knows its on her phone somewhere!
METRICS
METRICS
Evaluating Apple Intelligence through user impact.
Evaluating Apple Intelligence through user impact.
We identified metrics that could measure the impact of the contextual search, including query success rate, time to find messages, and feature adoption. Tracking these would help us understand how well the AI interprets natural language, how efficiently users can locate messages, and how readily the new feature is embraced. These metrics guide design decisions by highlighting areas for improvement and ensuring the search experience remains intuitive and valuable for users.
We identified metrics that could measure the impact of the contextual search, including query success rate, time to find messages, and feature adoption. Tracking these would help us understand how well the AI interprets natural language, how efficiently users can locate messages, and how readily the new feature is embraced. These metrics guide design decisions by highlighting areas for improvement and ensuring the search experience remains intuitive and valuable for users.
Query Success Rate
Percentage of natural language queries that result in a user clicking on a suggested conversation.
Time to Find
Average time taken to find a specific message using the new feature vs. the old keyword-only search.
Feature Adoption
Percentage of users who use the new contextual search field over the standard search
REFLECTIONS
REFLECTIONS
What I learned.
What I learned.
Throughout this process, I was constantly reframing my perspective from how I approached design to what I know about AI. Being so limited in our time (and not so limited on pizza), efficiency was key to making a clean product. During our design sprint, we relied on the iOS design system, and I was surprised to find how much easier designing became. I always hear about innovation, and while it is important, I finally realized that building on the tried and true can be really powerful. A major source of innovation these days is AI as well, and a lot of new AI products seem to revolutionize the market and create something that didn't exist before. I learned that AI doesn't always have to be about the big moments, it can be about the small interactions that help people enhance their daily lives.
Apple
Intelligence
Search+
Making message retrieval seamless with AI
ROLE
Product Designer
COLLABORATORS
Nydia Kushta
Shaina Desroches
Yuwei Huang
Adela Milkes
Colin Mendoza
SKILLS
Artificial Intelligence
User Research
Interaction Design
Visual Design
Design Systems
TIMELINE
48 hours
THE CONTEXT
Searching messages should feel natural.
Apple’s current message search requires users to remember exact words, making it difficult to find old texts based on memory or topic. We set out to design an AI-powered search experience that understands intent, allowing users to find conversations, topics, and attachments with simple, natural language. By making search smarter and more intuitive, messaging becomes more fluid, personal, and effortless.
THE CHALLENGE
Bridging the gap between human memory and machine search.
The current iOS Messages search tool depends on exact keywords, which doesn’t match how people naturally recall conversations. We often remember moments by feeling, topic, or context, not by specific phrases. This mismatch makes searching old messages time-consuming and frustrating. To address this, our team set out to design an AI-powered search feature that interprets natural language queries while maintaining Apple’s standard of privacy and simplicity. Our challenge was to make the experience feel effortless, trustworthy, and consistent with iOS design principles.
USER RESEARCH
Understanding how people actually search for memories.
Before designing solutions, we wanted to understand how users currently search their messages and where frustration begins. We surveyed 20 iPhone users aged 18 to 50 to learn how they find old messages and manage information. We then identified key themes that shaped our design direction.
🌴
🤒
🗓️
👟
🚌
check out some user feedback we received
🌴
casual language
🍽️
memories
💡
privacy
People who have given up trying to find a message after multiple failed search attempts
0%
0%
People who screenshot or pin texts as a backup system
0%
0%
USER PERSONAS
Designing for different kinds of communicators.
We created user personas to understand the diverse ways people interact with their messages. Each persona represented unique goals, frustrations, and behaviors that shaped how we approached search, memory, and personalization. Our personas revealed that users vary widely in how they recall and organize information. Some value structure and precision, while others rely on visual cues or conversational memory. Recognizing these patterns allowed us to design a search experience that flexes to different mental models rather than forcing users to adapt to the system.
Olivia (37, Marketing Manager)
The Over-Organizer
Core needs
Wants AI to group messages by topics (e.g. “wedding planning,” “travel ideas”).
Needs quick access to photos, receipts, and links shared months ago.
Values smart filters that surface relevant details automatically.
Frustrations
Wastes time scrolling through mixed media and unrelated results.
Finds Apple’s current search too rigid for her multitasking workflow.
Omari (22, International Student)
The Meme Archivist
Core needs
Search that handles code-switching, slang, and mixed language phrases.
Visual search results: quick previews of memes, photos, GIFs.
Wants a clean, fast experience, less typing, more results.
Frustrations
Endless scrolling breaks up conversation flow.
Has to screenshot messages so they don't get lost.
KEY FEATURES
Building a smarter, more transparent search experience.
Our design focuses on making AI-powered search feel natural, transparent, and distinctly Apple. We used Apple Intelligence branding and native design components to build familiarity and trust. A visible feedback loop shows how the AI interprets each query, turning abstract language into clear, searchable concepts. By understanding context instead of keywords, the feature recognizes slang, abbreviations, and tone. Guided prompts like “describe your conversation…” encourage users to explore and discover with confidence.
@prianca
@nandi
feedback loop
guided prompt
IDEATION
Drafting a smarter, context-aware iMessage search.
To make searching iMessage more intuitive, we imagined a flow where users could type naturally and let the system understand their intent. By parsing context, tone, slang, and language variations, the AI can connect the query to related messages, photos, documents, and links. The goal was to design a feature that organizes results clearly and reduces the mental effort of searching, making the process feel effortless while keeping the interface clean and aligned with Apple’s design principles.
FINAL DESIGNS
Bringing Apple Intelligence to life.
The hi-fi prototype highlights the polished interactions and visual design of Apple Intelligence. Users can see how natural language queries are handled, results are organized, and the interface responds seamlessly, demonstrating how our design principles translate into an intuitive, engaging experience.
Here we have Emma, who can't remember the name of that Mexican restaurant she went to last week, but knows its on her phone somewhere!
METRICS
Evaluating Apple Intelligence through user impact.
We identified metrics that could measure the impact of the contextual search, including query success rate, time to find messages, and feature adoption. Tracking these would help us understand how well the AI interprets natural language, how efficiently users can locate messages, and how readily the new feature is embraced. These metrics guide design decisions by highlighting areas for improvement and ensuring the search experience remains intuitive and valuable for users.
Query Success Rate
Percentage of natural language queries that result in a user clicking on a suggested conversation.
Time to Find
Average time taken to find a specific message using the new feature vs. the old keyword-only search.
Feature Adoption
Percentage of users who use the new contextual search field over the standard search
REFLECTIONS
What I learned.
Throughout this process, I was constantly reframing my perspective from how I approached design to what I know about AI. Being so limited in our time (and not so limited on pizza), efficiency was key to making a clean product. During our design sprint, we relied on the iOS design system, and I was surprised to find how much easier designing became. I always hear about innovation, and while it is important, I finally realized that building on the tried and true can be really powerful. A major source of innovation these days is AI as well, and a lot of new AI products seem to revolutionize the market and create something that didn't exist before. I learned that AI doesn't always have to be about the big moments, it can be about the small interactions that help people enhance their daily lives.
