Designing a Smarter Way to Restock Smart Vending Machines
Role: Sr. UX/UI Designer | Company: Stockwell Ai | Project Scope: L
25%
Improvement to service operation time
36%
Reduction in servicing errors
25%
Improvement to service operation time
36%
Reduction in servicing errors
25%
Improvement to service operation time
36%
Reduction in servicing errors
Context
Background
Stockwell AI was a startup based in Oakland, CA that developed autonomous, AI-powered vending machines designed for modern convenience. These machines used computer vision and smart sensors to deliver a seamless shopping experience without traditional checkouts or staff. Stockwell didn’t just make the machines—they operated the entire business end-to-end: from product inventory and warehousing to managing the vehicle fleet and hiring restocking staff.
With full ownership of the ecosystem, there was a huge opportunity to optimize how we maintained and restocked our machines—especially as we scaled.
Problem
While the technology powering Stockwell's machines was advanced, the process for restocking them was not. Our vertically integrated system meant that we handled every piece—from supply chain to last-mile delivery. But the last mile was broken.
Restockers used spreadsheets, printed lists, and verbal instructions. Routes weren’t optimized. Errors were common. Machines were stocked inconsistently, and inventory accuracy suffered.
We needed a digital tool—a single app—that would:
Guide restockers through optimized routes
Display what needed restocking (and where)
Improve speed, reduce errors, and keep inventory data accurate
Process
Discovery
Discovery
Discovery
Defining Requirements + Early Design
Defining Requirements + Early Design
Defining Requirements + Early Design
Prototyping + Testing
Prototyping + Testing
Prototyping + Testing
Final designs + Implementation
Final designs + Implementation
Final designs + Implementation
Results
We launched the app with a small group of restockers to gather feedback before rolling it out company-wide. Early results showed measurable improvements in accuracy and time per restock.
25% reduction in average service time (from ~20 minutes per stop to ~16 minutes)
36% reduction of servicing errors
Designing a Smarter Way to Restock Smart Vending Machines
Role: Sr. UX/UI Designer | Company: Stockwell Ai | Project Scope: L
25%
Improvement to service operation time
36%
Reduction in servicing errors
25%
Improvement to service operation time
36%
Reduction in servicing errors
25%
Improvement to service operation time
36%
Reduction in servicing errors
Context
Background
Stockwell AI was a startup based in Oakland, CA that developed autonomous, AI-powered vending machines designed for modern convenience. These machines used computer vision and smart sensors to deliver a seamless shopping experience without traditional checkouts or staff. Stockwell didn’t just make the machines—they operated the entire business end-to-end: from product inventory and warehousing to managing the vehicle fleet and hiring restocking staff.
With full ownership of the ecosystem, there was a huge opportunity to optimize how we maintained and restocked our machines—especially as we scaled.
Problem
While the technology powering Stockwell's machines was advanced, the process for restocking them was not. Our vertically integrated system meant that we handled every piece—from supply chain to last-mile delivery. But the last mile was broken.
Restockers used spreadsheets, printed lists, and verbal instructions. Routes weren’t optimized. Errors were common. Machines were stocked inconsistently, and inventory accuracy suffered.
We needed a digital tool—a single app—that would:
Guide restockers through optimized routes
Display what needed restocking (and where)
Improve speed, reduce errors, and keep inventory data accurate
Process
Discovery
Defining Requirements + Early Design
Prototyping + Testing
Final designs + Implementation
Results
We launched the app with a small group of restockers to gather feedback before rolling it out company-wide. Early results showed measurable improvements in accuracy and time per restock.
25% reduction in average service time (from ~20 minutes per stop to ~16 minutes)
36% reduction of servicing errors