Assembly line inefficiency, scattered data and infrastructure,
broken record-keeping, inaccurate instructions, shift handoff,
inventory tracking and ordering, QA, communication.
Solution:
Build a single-point solution on iPad, new data structure,
real-time communication system.
Timeline:
1 yr, 5 months
Team:
1 lead product designer, 2 UI designers, 1 PM, 2 data architects, 4 development teams, 1 customer success manager
My Role and Ownership
Research
Product Roadmap
Prototyping
Visual Design
Design QA
Research:
Assessment included 5 days of shadowing 8 assembly line workers across 3 shifts; interviewed
3 supervisors and 2 inventory workers. Spoke with data architects to get an understanding
of limitations and possibilities.
Workflow Design:
Developed feature outlines, estimating feature design/build cycles and timelines
Prototyping
Built clickable iPad prototypes and tested them with workers on the plant floor (see How we validated it).
Visual Design:
Created the design and look and feel: styling and aesthetics.
Design QA:
Reviewed builds on iPad device and coordinated with developers on visual
and functionality changes.
“Wish I had everything all in one place,
and be able to keep it with me
on the assembly line”
Ken, assembly
Problems, Research Findings
2-4
hr
avg
Spent searching for information
90
min
avg
Spent waiting on communication
Stationary Computers
Workers are constantly leaving their specific work area to look something up.
Scattered Records
No single place to keep documents, no naming conventions, and lengthy searches to find anything in the broken system.
No Real Time Data Syncing
System data needed to be organized and accessible; the system was a mix of patches of data sets and needed rebuilding.
Solutions
Dashboards
User-type dashboards for quick summaries,
task outlines, risk management
Document Space
Document annotations, camera input, personalized
instructions or notes to add to steps or processes
Messaging, Calls
Direct communication with workers,
supervisors, file sharing
Workflow Breakdown
Workflow steps that included a timer, instructions, tools,
documents, communication, inventory tracking
Revised Search
New database structure allows for unified
deep search by tags and keywords
Account and PIN sign-in
New database structure allows for unique PIN sign-in, can
assign training, track efficiency, time off, and adjust work
schedules by manager.
Access to training documents, OSHA standards, safety protocols
Shift Handoff
Allows information to be recorded and passed on
easily to the next shift, automated instead of spoken
and lost in translation
Part Tracking
Inventory tracking, including status, part numbers,
order numbers, safety tracking, tied into work forecasting
and assembly dependency
At a Glance
Outcome
Information lookups dropped from hours to minutes. QA approvals moved to the iPad so work kept moving, and the central command mainframe became a backup system.
Decisions I pushed for
Step-by-step tracking at every station.
I pushed to break each assembly section into visible steps. Workers could see what came next and mark progress as they went. Supervisors got visibility into every station and part across the line, and the plant got step-level completion data it had never had, which made efficiency measurable.
Annotated photos that move up the chain.
The client only asked to view images. I proposed letting workers photograph a part or step, mark up the photo, and attach notes on what they ran into. That annotated photo could go to a supervisor for feedback and get escalated if needed, so problems on the floor got recorded and reached the right person instead of getting lost between shifts.
Led two UI designers
I set the interaction patterns, ran daily standups and design reviews, managed workloads and timelines, and reviewed every screen before handoff.
How we validated it
We started with assumptions and tested them on the plant floor. I built clickable prototypes in Sketch, loaded them on an iPad, and had assembly workers, supervisors and inventory staff use them at their stations while I watched, took notes and asked questions. Did the workflow make sense? Was anything missing?
I took those findings to the PM and lead developer. Together we worked out how each feature would be built and what data layer it depended on, and we sequenced the work so the data structure came first.
After launch, we ran an on-site pilot with 10 workers across two shifts, covering assembly, supervision and inventory, to confirm what worked and what was missing.
Results
Lookups dropped from hours to minutes.
Before, workers spent 2–4 hours a shift tracking down part numbers, process notes and instructions. That meant a walk or cart ride of up to half a mile to central command, and when they got there the computers were often all in use. They’d wait, go back empty-handed, or hunt for someone who knew. With the iPad, the same information was at their station in minutes, organized by area and station.
Approvals stopped holding up the line.
Workers sent completed tasks to QA from the iPad and kept working instead of waiting for a sign-off in person.
Central command became the backup.
With everything at their station, workers no longer needed to make the trip. The mainframe stayed in place only in case the iPads or the network went down.
Feedback was overwhelmingly positive.
Assembly workers, supervisors and inventory staff all pointed to the same things: finding what they needed without leaving the line, information grouped by their station, and faster approvals.
Built to scale beyond one plant.
The app was scoped for the F-35 line. The data structure and app framework were designed so Lockheed could adapt them to other assembly plants, each with its own data reorganized and site-specific changes, if the F-35 rollout succeeded.
Assembly Line Worker Dashboard
IA
Feature Set
Thought Process
I wanted the screen to give as much valuable information to the viewer about their job as possible. This typically goes against the premise that you give the user only the information needed at the time. I broke away from that, as this screen is their dashboard. Dashboards are an overview for seeing operations as a whole.
In this case, I wanted the worker to see what steps are taken to complete the assembly section. What needs inspection to move forward. Inside the steps are tasks, and a user can break down any part of their work into detail. There may be documents or specific instructions related to each task or step.
I wanted the user to not have to guess what they need to complete the task, what parts are there or not there, tools, documents, the ability to ask someone questions, and to send a completed task to QA for approval. The ability to take pictures and add them to the task or step.
Supervisor Dashboard
Thought Process
I wanted the screen to give the supervisor as much information as possible about the day-to-day projects that are under his umbrella. Supervisors can see workstations and then click on a tile to see the details. They can see their approval tasks and stations that are on hold.
Document Editing
Thought Process
The customer only wanted the ability to see images, possibly zoom in and see details if there were any. They also wanted to be able to see all of a batch of images or documents that were attached together.
Workers could add notes to any image and toggle them on or off. They could also take new photos and attach them to an image, document or process step.
A worker could have the ability to see any newly added notes or images. Who added them and when. This ability was for accountability and tracking.
What didn't make the cut
I also proposed the ability to see part numbers and stock as outlined in the image. This idea was too complex with the hundreds of thousands of parts and numbers. We shelved this idea, but everyone thought this could be useful in the future.
Images of parts. This idea was lofty and would require either someone to take photos of each part or suppliers to provide them. That was just not possible or valuable.
With the help of AI, I think that these ideas could be easier to implement in today’s product building.