commerce workflow
Transforming fragmented commerce workflows into an AI-assisted, automated workflow solution
00

problem
Within the commerce organization, legacy workflows were heavily fragmented across disconnected tools. The existing system was dictated by the limitations of a third-party SaaS product rather than user needs, resulting in an inefficient process heavily reliant on manual hand-off forms. This over-reliance on manual intervention severely impacted productivity and increased the margin for error. Furthermore, the engineering team was spending considerable time maintaining case intake forms, which had become bloated over time with fields that were rarely needed, relevant, or used. As the organization pivoted to overhaul its operations, the mandate for our team was clear: design a dynamic, automated workflow ecosystem that eliminated manual inefficiencies and provided real-time visibility across the board.
solution
We replaced the UI layer of the third-party product with a Single Page Application (SPA), aligned with the design tenets of the modern commerce platform and design system. By building the UI layer in Angular and using APIs to interact with the multitude of tools used by the commerce team, we essentially took back control of the user experience of workflow management ecosystem. The model utilized variable states and modes to keep the user experience seamless. The interface dynamically adapted to the user's specific role through the implementation of Role-Based Access Control (RBAC), surfacing only relevant data and actions. The UI prioritized immediate action over passive information gathering. Furthermore, we designed the UI to be served as a Micro-Frontend (MFE), allowing these modules to be plugged directly into any of the tools the commerce organization was already using.
When I joined the project, the team was caught in an infinite loop of iterations. They were struggling to reconcile the ambitious vision laid out by the design strategist with the ground reality of technical platform constraints. Engineering was waiting for hand-offs that never materialized.
My objective was to ground the design—honoring the strategic vision while remaining realistic about technical limitations. The primary constraint? I had exactly two sprints to hit the ground running, untangle the process, and deliver an engineering-ready design.

To get a lay of the land, I had several 1:1 with the business lead, design strategist and the design team to understand the vision, objective and the progress so far. It was clear that the team had missed a milestone and there wasn't a plan in place to show that we could get back in track to not derail the project timelines. We created and shared an action plan for the next few weeks.
Prior to my assignment, design, research and process architects were engaged by business to build an understanding of the current landscape and process. They had built a massive repo containing journey maps, process flows, and market research. Given the aggressive timeline, I had to make sense of this data fast to inform my design decisions. I used Copilot (the only AI vetted to be used at the org) to rapidly synthesize extensive research data to generate insights, and identify patterns to inform my design direction. Parallelly, I had long sessions with the product architect to understand the current ecosystem, its limitations and any potential blockers.
After rapid prototyping and iterating through multiple variations, we zeroed in on a Single-Page Application (SPA) solution. This would allow us to embed the workflow components (built as MFE's) in any commerce tool where workflow had to be integrated.
With the baseline UX defined and a solid design-to-engineering pipeline established, the next immediate challenge was scale. The deadline to sunset the legacy tool was rapidly approaching, turning this phase into a race against the clock.
My focus shifted to collaborating closely with design teams across the broader commerce and workflow organizations to deeply understand their specific process executions. The objective was to build a highly scalable solution capable of:
Integrating seamlessly into any existing commerce tool.
Picking up the process exactly where one team handed it off.
Guiding that flow seamlessly through our new system.
Navigating Technical Limitations and Manual Realities: At the time, the reality of the existing workflows was heavily manual and friction-heavy. Cross-team hand-offs relied on cumbersome data dumps, uploading massive Excel sheets, and attaching documents to intake forms. A case would be generated, assigned, worked on, and then passed to the next team using that exact same manual loop.
While our ultimate vision was full automation, we hit a hard technical constraint: the teams supporting the other commerce tools lacked the time and bandwidth to build autonomous data transfer pipelines before the legacy system's shutdown date.
I had to design a pragmatic solution. I optimized the UI for these unavoidable manual hand-offs—streamlining the manual triggers via intake forms to reduce friction—while ensuring the underlying architecture was fully prepared to support seamless, API-driven automation the moment those external pipelines were eventually built.
year
2025
timeframe
3 Months
tools
Figma, CoPilot
category
B2B Web App
impact
manual entry reduction
27%
UI templatization
66%
Reduced manual data entry by 27% by implementing smart defaults and drastically cut down decision-making time by filtering drop-down selections based on the user role, surfacing only relevant options. Standardized the underlying case intake architecture, creating reusable templates for 60-65% of the standard case fields. This allowed the engineering team to rapidly generate and deploy new case intake forms by only having to build the 5 to 6 unique fields required for specific workflows. The application is designed to be scalable, designed to support future "agentic workflows" where child tickets can be actioned autonomously by AI agents rather than human operators.
learnings
Pragmatism Over Perfection: A visionary design strategy is only valuable if it can be built. By anchoring the vision in the reality of technical and platform constraints, I was able to break the team out of an endless cycle of theoretical iterations and deliver an engineering-ready product. Accelerating Synthesis with AI: When time is the ultimate constraint, resourcefulness is key. Leveraging AI allowed me to bypass a redundant discovery phase, immediately identify patterns, and start solving the core problems within a strict two-sprint window. Prioritizing Action-Oriented UX: In bloated enterprise tools, cognitive overload is a massive barrier. Shifting the UI's focus from dense information display to context-driven action proved that the most effective design intervention is often simply getting out of the user's way.
01

02

03

see also


