Cue: AI-Powered Learning Companion for Product Designers
Cue: AI-Powered Learning Companion for Product Designers
Cue: AI-Powered Learning Companion for Product Designers
An AI-powered learning companion that turns scattered design resource saves into skills product designers can learn and apply at work
An AI-powered learning companion that turns scattered design resource saves into skills product designers can learn and apply at work
An AI-powered learning companion that turns scattered design resource saves into skills product designers can learn and apply at work
Role
Founder & Solo Product Designer
Role
Founder & Solo Product Designer
Role
Product Designer
Duration
1 week (and ongoing)
Duration
1 week (and ongoing)
Duration
1 week (and ongoing)
Team
Me, myself, and I
Team
Me, myself, and I
Team
Me, myself, and I
Tools
Figma・Claude・Mobbin・Vercel
Tools
Figma・Claude・Mobbin・Vercel
Tools
Figma・Claude・Mobbin・Vercel
Platform
iOS mobile app
Platform
iOS mobile app
Platform
iOS mobile app
Contribution
Product Strategy・Conversational AI UX・AI-Assisted Design Workflow・Information Architecture & Taxonomy
Contribution
Product Strategy・Conversational AI UX・AI-Assisted Design Workflow・Information Architecture & Taxonomy
Contribution
Product Strategy・Conversational AI UX・AI-Assisted Design Workflow・Information Architecture & Taxonomy
OVERVIEW
OVERVIEW
OVERVIEW
The idea of Cue came from my own habit. Every day I scroll through hundreds of designer influencers' feeds on Instagram and try to save so many reels and keep track of every DMs I sent for the resource links they sent me. But more often than not, the valuable resources were lost or hard to trace back when I finally need them or have the time to go through them.
Outside of Instagram, I also subscribe to multiple design/tech newsletters, browse articles or blogs, and watch youtube videos. These resources are scattered and messy, and I wish there’s a more organized way for designers to access and manage them more easily.
I realized product designers are constantly exposed to new tools, techniques, and inspiration, but these resources are scattered across social media, newsletters, websites, DMs, and bookmarks. Saving is easy, but turning these fragmented discoveries into an organized, accessible body of knowledge is not.
The idea of Cue came from my own habit. Every day I scroll through hundreds of designer influencers' feeds on Instagram and try to save so many reels and keep track of every DMs I sent for the resource links they sent me. But more often than not, the valuable resources were lost or hard to trace back when I finally need them or have the time to go through them.
Outside of Instagram, I also subscribe to multiple design/tech newsletters, browse articles or blogs, and watch youtube videos. These resources are scattered and messy, and I wish there’s a more organized way for designers to access and manage them more easily.
I realized product designers are constantly exposed to new tools, techniques, and inspiration, but these resources are scattered across social media, newsletters, websites, DMs, and bookmarks. Saving is easy, but turning these fragmented discoveries into an organized, accessible body of knowledge is not.
The idea of Cue came from my own habit. Every day I scroll through hundreds of designer influencers' feeds on Instagram and try to save so many reels and keep track of every DMs I sent for the resource links they sent me. But more often than not, the valuable resources were lost or hard to trace back when I finally need them or have the time to go through them.
Outside of Instagram, I also subscribe to multiple design/tech newsletters, browse articles or blogs, and watch youtube videos. These resources are scattered and messy, and I wish there’s a more organized way for designers to access and manage them more easily.
I realized product designers are constantly exposed to new tools, techniques, and inspiration, but these resources are scattered across social media, newsletters, websites, DMs, and bookmarks. Saving is easy, but turning these fragmented discoveries into an organized, accessible body of knowledge is not.
PROBLEM
PROBLEM
PROBLEM
How might we help product designers turn the scattered resources they discover every day into an organized, evolving knowledge system they can continuously learn from and apply to their work?
How might we help product designers turn the scattered resources they discover every day into an organized, evolving knowledge system they can continuously learn from and apply to their work?
How can Hyundai vehicles leverage AI to help drivers navigate unfamiliar routes safely in severe weather conditions when driving alone?
SOLUTION
SOLUTION
SOLUTION
Cue is a mobile app that turns what you save into what you know in 6 steps:
Save from anywhere. Share a reel, video, newsletter or article to Cue in one tap, without leaving the app you're in.
Cue turns it into a resource card. AI writes a summary and key takeaways, and pulls out every tool and link mentioned.
Your library organizes itself. Every save is sorted by skill, format and source, so you can find it again however you remember it.
Learn a little every day. Swipe through that day's saves as bite-sized cards, or get the gist in one AI summary when you're short on time.
See your growth and your taste. Track what you've learned and applied, and discover your Design DNA: the skills, styles and creators that define you as a designer.
Ask Cue when you need it. Describe what you're working on, and Cue finds the saves that apply, so what you saved gets used in real projects.
Cue is a mobile app that turns what you save into what you know in 6 steps:
Save from anywhere. Share a reel, video, newsletter or article to Cue in one tap, without leaving the app you're in.
Cue turns it into a resource card. AI writes a summary and key takeaways, and pulls out every tool and link mentioned.
Your library organizes itself. Every save is sorted by skill, format and source, so you can find it again however you remember it.
Learn a little every day. Swipe through that day's saves as bite-sized cards, or get the gist in one AI summary when you're short on time.
See your growth and your taste. Track what you've learned and applied, and discover your Design DNA: the skills, styles and creators that define you as a designer.
Ask Cue when you need it. Describe what you're working on, and Cue finds the saves that apply, so what you saved gets used in real projects.
Cue is a mobile app that turns what you save into what you know in 6 steps:
Save from anywhere. Share a reel, video, newsletter or article to Cue in one tap, without leaving the app you're in.
Cue turns it into a resource card. AI writes a summary and key takeaways, and pulls out every tool and link mentioned.
Your library organizes itself. Every save is sorted by skill, format and source, so you can find it again however you remember it.
Learn a little every day. Swipe through that day's saves as bite-sized cards, or get the gist in one AI summary when you're short on time.
See your growth and your taste. Track what you've learned and applied, and discover your Design DNA: the skills, styles and creators that define you as a designer.
Ask Cue when you need it. Describe what you're working on, and Cue finds the saves that apply, so what you saved gets used in real projects.
KEY DECISIONS
KEY DECISIONS
KEY DECISIONS
Saves are organized by how designers search, not by folders
Most save tools leave users to create and maintain their own folders, like Instagram Collections or Google Maps lists. But designers may remember what a resource was about, what format it was in, or where they saw it, rather than which folder they saved it to. So CUE automatically organizes every save across three dimensions: skill (Interaction, AI, Visual/UI), format (tutorials, tool recommendations…), and source (Instagram, LinkedIn, web).
Trade-off: Three category tabs and more filter chips create a busier interface than a single folder list.
Why it’s worth it: A little more interface complexity means less manual organization and more ways to find a resource without remembering where it was saved.
Saves are organized by how designers search, not by folders
Most save tools leave users to create and maintain their own folders, like Instagram Collections or Google Maps lists. But designers may remember what a resource was about, what format it was in, or where they saw it, rather than which folder they saved it to. So CUE automatically organizes every save across three dimensions: skill (Interaction, AI, Visual/UI), format (tutorials, tool recommendations…), and source (Instagram, LinkedIn, web).
Trade-off: Three category tabs and more filter chips create a busier interface than a single folder list.
Why it’s worth it: A little more interface complexity means less manual organization and more ways to find a resource without remembering where it was saved.
Saves are organized by how designers search, not by folders
Most save tools leave users to create and maintain their own folders, like Instagram Collections or Google Maps lists. But designers may remember what a resource was about, what format it was in, or where they saw it, rather than which folder they saved it to. So CUE automatically organizes every save across three dimensions: skill (Interaction, AI, Visual/UI), format (tutorials, tool recommendations…), and source (Instagram, LinkedIn, web).
Trade-off: Three category tabs and more filter chips create a busier interface than a single folder list.
Why it’s worth it: A little more interface complexity means less manual organization and more ways to find a resource without remembering where it was saved.
One card at a time over one long feed
A backlog of saved resources can feel like homework, making it easy to put off exploring them. Today presents one resource at a time, with a TL;DR upfront so designers can quickly decide whether to dive deeper. Full takeaways remain available when they're ready.
Trade-off: Swiping through individual cards takes more interactions than scanning a single feed.
Why it’s worth it: Making each resource easier to approach lowers the barrier to getting started. I'd rather help designers engage with a few resources meaningfully than overwhelm them with everything at once.
Transparent AI over a cleaner screen
Designers need to evaluate the advice they bring into their work, especially when AI is involved. Ask CUE surfaces its progress as it works (e.g., searching the Library, reading saved resources), while AI-generated content carries a “Written by CUE AI” label, distinct from the original source.
Trade-off: Progress updates and attribution labels take up screen space and add visual detail.
Why it’s worth it: Separating AI-generated explanations from original sources helps designers understand where information comes from and assess the advice for themselves. I kept these cues visually subtle so transparency doesn't come at the expense of readability.
A main path plus a shortcut,
over one fixed path
Designers don't always have the same amount of time to explore their saved resources. Today makes swiping through individual cards the primary experience, while “Get the Gist” offers a faster, consolidated view of the day's content.
Trade-off: Two entry points add interface elements and slightly reduce visual simplicity.
Why it’s worth it: The two modes support different browsing needs: focused exploration when time allows, and quick scanning when it doesn't. I kept “Get the Gist” secondary so the shortcut stays accessible without competing with the primary experience.
One card at a time over one long feed
A backlog of saved resources can feel like homework, making it easy to put off exploring them. Today presents one resource at a time, with a TL;DR upfront so designers can quickly decide whether to dive deeper. Full takeaways remain available when they're ready.
Trade-off: Swiping through individual cards takes more interactions than scanning a single feed.
Why it’s worth it: Making each resource easier to approach lowers the barrier to getting started. I'd rather help designers engage with a few resources meaningfully than overwhelm them with everything at once.
One card at a time over one long feed
A backlog of saved resources can feel like homework, making it easy to put off exploring them. Today presents one resource at a time, with a TL;DR upfront so designers can quickly decide whether to dive deeper. Full takeaways remain available when they're ready.
Trade-off: Swiping through individual cards takes more interactions than scanning a single feed.
Why it’s worth it: Making each resource easier to approach lowers the barrier to getting started. I'd rather help designers engage with a few resources meaningfully than overwhelm them with everything at once.
Transparent AI over a cleaner screen
Designers need to evaluate the advice they bring into their work, especially when AI is involved. Ask CUE surfaces its progress as it works (e.g., searching the Library, reading saved resources), while AI-generated content carries a “Written by CUE AI” label, distinct from the original source.
Trade-off: Progress updates and attribution labels take up screen space and add visual detail.
Why it’s worth it: Separating AI-generated explanations from original sources helps designers understand where information comes from and assess the advice for themselves. I kept these cues visually subtle so transparency doesn't come at the expense of readability.
Transparent AI over a cleaner screen
Designers need to evaluate the advice they bring into their work, especially when AI is involved. Ask CUE surfaces its progress as it works (e.g., searching the Library, reading saved resources), while AI-generated content carries a “Written by CUE AI” label, distinct from the original source.
Trade-off: Progress updates and attribution labels take up screen space and add visual detail.
Why it’s worth it: Separating AI-generated explanations from original sources helps designers understand where information comes from and assess the advice for themselves. I kept these cues visually subtle so transparency doesn't come at the expense of readability.
A main path plus a shortcut,
over one fixed path
Designers don't always have the same amount of time to explore their saved resources. Today makes swiping through individual cards the primary experience, while “Get the Gist” offers a faster, consolidated view of the day's content.
Trade-off: Two entry points add interface elements and slightly reduce visual simplicity.
Why it’s worth it: The two modes support different browsing needs: focused exploration when time allows, and quick scanning when it doesn't. I kept “Get the Gist” secondary so the shortcut stays accessible without competing with the primary experience.
A main path plus a shortcut,
over one fixed path
Designers don't always have the same amount of time to explore their saved resources. Today makes swiping through individual cards the primary experience, while “Get the Gist” offers a faster, consolidated view of the day's content.
Trade-off: Two entry points add interface elements and slightly reduce visual simplicity.
Why it’s worth it: The two modes support different browsing needs: focused exploration when time allows, and quick scanning when it doesn't. I kept “Get the Gist” secondary so the shortcut stays accessible without competing with the primary experience.
PROCESS: My AI-Native Design Workflow
PROCESS: My AI-Native Design Workflow
PROCESS: My AI-Native Design Workflow
I paired Claude + Mobbin MCP to move from raw ideas to research-informed UI,
and deployed the prototype with Vercel.
1. Work with AI to define a clear product scope
I used Claude to pressure-test my initial ideas, iterate on the problem statement, and shape a focused PRD with clear product scope.
2. Research before designing
I sketched multiple concepts for key screens, then connected Claude to Mobbin via MCP to research dozens of existing products, flows, and UI patterns. I evaluated the options and their trade-offs before developing my own wireframes.
3. Keep design decisions intentional
To avoid generic AI-generated interfaces, I combined my own sketches with real-world design references, explored multiple directions, and used precise design vocabulary to guide visual execution. AI accelerated exploration, but the final decisions remained mine.
I paired Claude + Mobbin MCP to move from raw ideas to research-informed UI,
and deployed the prototype with Vercel.
1. Work with AI to define a clear product scope
I used Claude to pressure-test my initial ideas, iterate on the problem statement, and shape a focused PRD with clear product scope.
2. Research before designing
I sketched multiple concepts for key screens, then connected Claude to Mobbin via MCP to research dozens of existing products, flows, and UI patterns. I evaluated the options and their trade-offs before developing my own wireframes.
3. Keep design decisions intentional
To avoid generic AI-generated interfaces, I combined my own sketches with real-world design references, explored multiple directions, and used precise design vocabulary to guide visual execution. AI accelerated exploration, but the final decisions remained mine.
I paired Claude + Mobbin MCP to move from raw ideas to research-informed UI,
and deployed the prototype with Vercel.
1. Work with AI to define a clear product scope
I used Claude to pressure-test my initial ideas, iterate on the problem statement, and shape a focused PRD with clear product scope.
2. Research before designing
I sketched multiple concepts for key screens, then connected Claude to Mobbin via MCP to research dozens of existing products, flows, and UI patterns. I evaluated the options and their trade-offs before developing my own wireframes.
3. Keep design decisions intentional
To avoid generic AI-generated interfaces, I combined my own sketches with real-world design references, explored multiple directions, and used precise design vocabulary to guide visual execution. AI accelerated exploration, but the final decisions remained mine.
Clear PRD
Clear PRD
Clear PRD
Mobbin research
Mobbin research
Mobbin research
10+ rounds of iteration
10+ rounds of iteration
10+ rounds of iteration
Evaluate trade-offs to build wireframes
Evaluate trade-offs to build wireframes
Evaluate trade-offs to build wireframes
Accurate visual style vocabulary
Accurate visual style vocabulary
Accurate visual style vocabulary
LEARNINGS
LEARNINGS
LEARNINGS
Research tools give patterns, not judgement and answers.
Mobbin showed how real apps solve a problem in minutes, but the research suggested status badges on cards and I removed them once I saw them on real content. My own UX judgement still decides what fits this product.
Using accurate design vocabulary in prompts affects the quality of AI-generated design immensely and avoids generic design.
Naming exact styles (Memphis, Mid-Century, Utilitarian, etc.) from 50 Design Styles Every Designer Should Know produced distinct directions instead of ten versions of "modern and clean." The better I could describe what I wanted, the better the output.
Research tools give patterns, not judgement and answers.
Mobbin showed how real apps solve a problem in minutes, but the research suggested status badges on cards and I removed them once I saw them on real content. My own UX judgement still decides what fits this product.
Using accurate design vocabulary in prompts affects the quality of AI-generated design immensely and avoids generic design.
Naming exact styles (Memphis, Mid-Century, Utilitarian, etc.) from 50 Design Styles Every Designer Should Know produced distinct directions instead of ten versions of "modern and clean." The better I could describe what I wanted, the better the output.
Research tools give patterns, not judgement and answers.
Mobbin showed how real apps solve a problem in minutes, but the research suggested status badges on cards and I removed them once I saw them on real content. My own UX judgement still decides what fits this product.
Using accurate design vocabulary in prompts affects the quality of AI-generated design immensely and avoids generic design.
Naming exact styles (Memphis, Mid-Century, Utilitarian, etc.) from 50 Design Styles Every Designer Should Know produced distinct directions instead of ten versions of "modern and clean." The better I could describe what I wanted, the better the output.
NEXT STEPS
NEXT STEPS
NEXT STEPS
Validate the knowledge model.
Hand-tag 100–150 of my saves, score the AI against them, and refine the taxonomy until it reaches 85% agreement.
Build the MVP and ship a beta.
Share-sheet capture, the background AI pipeline, the Library, Today and Ask Cue, then a TestFlight beta. This is also where I grow as a design engineer: building the APIs and data model myself, not just the screens.
Test with real users and iterate.
Test with 10–30 designers who save a lot of design content. First questions: do people understand the three axes without explanation, and does Today get saves revisited? Success means a 30% revisit rate and 25% of people still using it after four weeks.
Design a Chrome Extension.
Although this product is mainly designed for the use case of mobile scrolling, there are still times where designers discover valuable resources when they are on desktop. Designing a Chrome Extension is a next phase feature to make the product more complete.
Validate the knowledge model.
Hand-tag 100–150 of my saves, score the AI against them, and refine the taxonomy until it reaches 85% agreement.
Build the MVP and ship a beta.
Share-sheet capture, the background AI pipeline, the Library, Today and Ask Cue, then a TestFlight beta. This is also where I grow as a design engineer: building the APIs and data model myself, not just the screens.
Test with real users and iterate.
Test with 10–30 designers who save a lot of design content. First questions: do people understand the three axes without explanation, and does Today get saves revisited? Success means a 30% revisit rate and 25% of people still using it after four weeks.
Design a Chrome Extension.
Although this product is mainly designed for the use case of mobile scrolling, there are still times where designers discover valuable resources when they are on desktop. Designing a Chrome Extension is a next phase feature to make the product more complete.
Validate the knowledge model.
Hand-tag 100–150 of my saves, score the AI against them, and refine the taxonomy until it reaches 85% agreement.
Build the MVP and ship a beta.
Share-sheet capture, the background AI pipeline, the Library, Today and Ask Cue, then a TestFlight beta. This is also where I grow as a design engineer: building the APIs and data model myself, not just the screens.
Test with real users and iterate.
Test with 10–30 designers who save a lot of design content. First questions: do people understand the three axes without explanation, and does Today get saves revisited? Success means a 30% revisit rate and 25% of people still using it after four weeks.
Design a Chrome Extension.
Although this product is mainly designed for the use case of mobile scrolling, there are still times where designers discover valuable resources when they are on desktop. Designing a Chrome Extension is a next phase feature to make the product more complete.
Brooklyn, NY
Brooklyn, NY
Brooklyn, NY