Plugin + Website
Semi-Autonomous Vehicles and Weft @ Crown: An AI Opportunity Evaluation and Management Software
roles
Product Designer
UX Researcher
Interaction Designer
tools
Figma
Claude Code
team
Me
overview
Interaction design for semi-autonomous vehicles and concepting an AI workflow aid.
In my 14 weeks at Crown as an Interaction Design intern, I worked with each part of an adaptive design process on forklifts: creating user journey maps in Figjam, designing screens in Figma, researching interaction patterns for semi-/autonomous vehicles (AGVs), user testing for design systems, developing UX for multimodal communication, and anticipating error states.
Due to Crown's NDA policies on ongoing projects, I provide a brief overview of my work below.
I was also mentored to create a personal project on any subject: Weft.
Business Context at Crown
There are no existing models of what autonomy looks like quite yet in design. How do I design for new interactions and an operator who has no attention to spare?
The intent of UX in this role flipped my typical design goals-- I didn't design to hold attention, but to deliver information succinctly, for minimal attention and operator safety.
1
Crown is a global manufacturer of warehouse forklifts, based in 80+ countries.
2
Primary clients are Amazon, Walmart, Target, Pepsico, Costco, + more!
3
Forklifts have double the fatality rate cars do.
Unlike cars, forklift users rely on operator signaling with the vehicle's lights and signals.
So how can we design for a shift to autonomous systems without relying on this human operator?
Primary Work Summary
Figma-ing for autonomy, design systems, and new products!
A brief summary of what I did as an interaction and UX design intern.
1
Designed screens for autonomy in vehicles
2
User journey mapped for edge cases
3
Design system component library audits + restructuring, increasing efficiency by 15% and reducing components by 37%
4
Animated UI for multimodal communication and error states,
5
Researched AI methods, developing several Figma AI plugins for documentation and editing
If you'd like to learn more about my work, reach out to me at joanaliuj@gmail.com to talk! :)
Hear From People I Worked With
Some Kind Words!
Special thanks to my team + manager, Justin! As I worked on several projects, I partnered with Industrial designers, Software Engineers, other Interaction Designers, and had several opportunities to share findings with upper management.
"Joana proved to have a thorough understanding of each project we gave her - the technology, the business objectives, the most suitable tools for the tasks, skillful prioritization of features and information architecture, a seemingly innate ability to understand the users’ needs and contexts, along with her enthusiastic, charismatic, and fact-backed way of communicating from peers to upper leadership. Can’t say enough good things about her. I would hire her back in a heartbeat!"
Justin Montan
Interaction Designer (My Manager)
“She immersed herself and adapted to designing for automation remarkably quickly. She showed a clear depth of understanding of the context she was designing for and communicated design intent very effectively. She exceeded expectations. Her work was fantastic; her process showed a strong understanding of appropriateness and rigor. She has a fantastic work ethic as well.”
Paul Magee
Head of Advanced & Emerging Technologies Department
From main work -> personal
Personal Project
The most impactful issue almost didn't get said. This tool increases our chances of hearing it.
"One day, a teammate and I brought up an issue for a few seconds in a shareout; we didn't find it important due to assumed constraints. However, this topic ended up being of high interest to the stakeholder.
We almost didn't present it; this cost was almost never realized. And this experience isn't unique to me, designers, or my company as a whole. It feels like a weird combination of expertise and luck."
Ideas aren't being translated effectively in current systems. Some of the most valuable insights sit within businesses' own teams while millions are spent on external strategic planning.
Your insight as a technician may move through physical conversations, meetings, and end there. Your insight as a designer may rest across Slack chats, notes, meetings, PowerPoint presentations, and more.
Either way, each handoff (across time) strips away the context a frontline employee understands that determines an idea's value.
An idea may reach someone with the authority to act at a high level.
But at that stage, strategy leaders may understand a different category of context, or only have enough context to support the ideas that require the least interpretation:
= it feels logical to select immediately actionable ideas, with clear statistics attached in a slideshow. The rest are forgotten or never heard, although these are often the most context-informed, with potential for future impact.
The Solution: Weft
Weft captures your intelligence before it disappears, turning organizational insight into action.
Developing, connecting, and prioritizing opportunities autonomously…
Easy access
Notice an opportunity? An issue with potential you don't have time to make a solution for? Not feasible, but valuable?
Upload an idea in seconds using text, voice, screenshots, drawings…
Building context
You submit an opportunity from the widget. Weft automatically builds support into an Idea Card.
Diverse communications cause splintered context. Each idea card's features are based on user needs I identified that pull that back together: in control and transparency in human-AI interaction.
Analyze through similarity
Weft auto-merges, groups, and can define topics by similarity.
Make sense of a team's varied input, insights, and different areas of expertise. Each card tracks similar and merged ideas. Similar cards are closer.
Dig deeper
Cross-analyze select ideas for more specific exploration.
Highlight categorized content to see what matches, or note an odd overlap of two ideas.
Priotize
See rankings, at any topic or idea level.
Ranking is based off of human + AI input. Not one or the other.
Rank
=
your urgency rating
+
merge/similarity count
+
team votes
+
growth speed
+
cross-team prominence
Iteration, not perfection
Privacy and Control is Yours first. Others only see what you submit to the Weave.
Have ideas on accelerating your own process or want to build a thought privately?
summary
Right now, we lose opportunities twice: for every employee who doesn't bring up a core solution, and every manager who can't weigh what's never shared, now or in the future.
Some of the most powerful intelligence lies fragmented across the experts in your workforce.
Weft highlights that intelligence, enriches it with AI, identifies relationships across the organization, and transforms fragmented ideas into a prioritized opportunity portfolio.
Before the Design...
How did I get there?
User Research
How does information and communication flow in a corporate environment?
Due to time constraints, I recruited and interviewed 4 full-time employees.
1:. Interviews Process
_
4
Interviews with full-time employees
1
Week
100 to 100k
Employees per company
1 to 20+
Years of experience
2:. Interviews Insight
+
Market Analysis
B2B market gap: there are no products that adapt and rely equally on human and algorithmic insights for ideas.
I researched companies in 1) the "idea" creating or decision-making space, 2) across B2B and B2C spaces, and 3) what level of AI/algorithm/human insight was used.
Ideation and Flows
User Flow to Information Architecture
I began with potential use cases for an idea creation + decision aid program. How would workers want to use this?
Users I interviewed cared about their control over AI interaction + and ability to analyze information on ideas.
Based on my outlined use cases + user flows, I designed the overarching product's information architecture, marking where levels of control, privacy, and AI should or should not be incorporated.
Weft is not designed to replace external strategizing or specialized strategy departments. It augments and refines the process with each company's existing natural experts.
Wireframing
An Example: How can a user cross-analyze ideas efficiently?
A sample of some of my screens, thinking about what users needed to know from Weft, from low to high-fidelity. Optimizing for less clutter and more clarity.
1:. Low-fidelity…
_
2:. Increasing fidelity
+
Pivots
Idea Analysis is valued over Exploration
I initially designed to encourage idea exploration. But feedback show analysis > exploration for any corporate worker, so I restructured.
1:. An idea's strength should be easy to see.
+
2:. The first tab was an exploratory newsfeed. How do I make analysis come first?
+
Scalability
How Items Connect at any Scale
One idea's sizes need to fit together 1) visually, 2) content/structurally, 3) at any scale, and 4) as a user was in motion, while panning or zooming in/out.
I found it challenging + exciting to define parameters for how these cards collapsed: how can scale, color, distance, and content draw connection?
Prototyping and AI
Figma's Limits, Prototyping, and Deployment
Figma can't prototype node-based interactions like you see in Weft. What's next, and what can? Claude Code.
It was difficult, even with Claude, to get granular control: the idea cards didn't shift sizes smoothly and responsively, often jumping, lagging, or becoming the wrong shape.
I know responsiveness is a common item to design for. I researched and ended up using a variety of Claude Skills from Github (and developed one for Figma -> Code conversions along the way) to automate this process in the future.
Reflection
On Interactions and designing for ultra-responsivity
Weft does not have a backend. I'd like to build that out next, so I can begin harnessing it for my own design + art ideation and research processes :)
