Linda Lee

CLIENT WORK / EXPERIENCE DESIGN

AI-Powered Supply Chain Automation

CONNECTED. AUTOMATED. HUMAN-CENTERED.

MY ROLE

Strategy & UX Designer

CONTRIBUTIONS

Design Thinking Facilitation, UX Strategy, Journey Mapping, Interaction Design, UI Design, Prototyping

DESIGN TOOLS

Figma, Mural, Adobe

DURATION

8 weeks

DURATION

8 weeks

01 · THE PROBLEM

Challenge

A large enterprise organization relied on manual processes and existing technology to manage high volumes of purchase orders and vendor transactions. Teams had to navigate multiple systems to verify information, while inconsistent data and variations in vendor file formats created processing delays, unreliable forecasts, and repeated invoice rejections.

02 · THE RESPONSE

Solution

Designed and prototyped an Agentic AI solution to streamline complex supply chain operations across purchase order automation and bid processing. By translating manual workflows into AI-enabled experiences, the pilot demonstrated how intelligent agents could handle high-volume transactions, reduce manual intervention, and improve operational efficiency.

01 · UNDERSTANDING THE OPPORTUNITY

Identifying the Opportunities

Through discovery workshops and workflow walkthroughs, we identified two priority areas where Agentic AI could reduce manual effort and improve operational efficiency

Governance shifts from a final checkpoint to a continuous layer across how AI is built, assessed, and monitored.

Purchase Order Automation

Approximately 60% of PO processing was manual, requiring employees to navigate multiple systems to verify information and complete transactions

KEY PAIN POINTS

Fragmented information: Freight terms, pricing, inventory, and vendor details spread across multiple sources

Unreliable data: Inaccurate shipment forecasts required manual vendor confirmation

Repetitive work: Individual updates and invoice discrepancies created additional back-and-forth

DESIGN OPPORTUNITY

Bring PO information together, simplify bulk updates, and guide employees through validation before release.

Bid Desk Processing

The existing process struggled with varying vendor file formats and required employees to manually coordinate bid requests and responses

KEY PAIN POINTS

Inconsistent formats: Vendor pricing files were not always recognized by existing automation

Manual coordination: Employees managed partner requests and tracked responses

Limited automation: Only 13% of transactions were automated, with an average turnaround time of approximately two hours

DESIGN OPPORTUNITY

Connect partner selection, pricing requests, response tracking, and bid validation in one streamlined workflow

THE RESULT

Final Designs

Automating Purchase Order Workflows

Designed a conversational AI workflow that centralizes PO information, streamlines bulk updates, and guides employees through validation before order release

Key Experience Highlights

* Initiate PO processing through a frequent action on the landing page

* Retrieve and review orders requiring updates using a Buyer ID

* Apply freight term changes across multiple orders in one action

* Validate shipping methods and order details before release

Streamlining Bid Desk Processing

Introduced an AI-driven bid management experience that automates partner coordination, handles pricing requests, tracks responses, and simplifies bid validation

Key Experience Highlights

* Create bid opportunities and select partners for pricing requests

* Review AI-generated partner emails before sending

* Automatically track responses and update opportunities

* Review bid completeness and validation results in one place

03 · DESIGN PROCESS & KEY DECISIONS

Design Process

A collaborative design process helped translate complex operational challenges into prioritized use cases, future-state workflows, and interactive prototypes.

Discovery & Use Case Prioritization

Facilitated workshops and workflow walkthroughs to identify pain points and prioritize high-value use cases based on business impact and feasibility

Process image 1

Current-State Workflow Discovery

Mapped the current-state workflows, identified manual tasks and dependencies, and uncovered opportunities where AI agents could simplify or accelerate operations

Process image 2

Future-State Journey & Experience Design

Translated prioritized use cases into conversational user flows and interactive prototypes, exploring how AI agents could handle tasks while keeping employees informed and in control

Process image 3

Prototyping & Client Iteration

Collaborated with product, engineering, and client stakeholders to review prototypes, validate feasibility, and refine the experience through weekly feedback sessions

Process image 4

KEY DESIGN DECISIONS

Simplify repetitive actions

Reduce manual steps through streamlined workflows

Keep employees in control

Preserve human review and approval at critical steps

Automate coordination

Let AI agents manage routine tasks, updates, and follow-ups

Make progress visible

Provide clear status updates and visibility into ongoing tasks

04 · Business Impact

Operational Value

13% →80%

Increase in automation rate demonstrated through the pilot

~2hrs → 10min

Reduction in average transaction turnaround time

1.7M

Annual transaction volume considered in the solution's design

05 · WHAT I LEARNED

Bringing People & AI Together

This project reinforced that successful enterprise AI design goes beyond task automation. It requires understanding how people work, where friction exists, and how AI can improve the end-to-end experience

Governance shifts from a final checkpoint to a continuous layer across how AI is built, assessed, and monitored.

Understand before automating

Identify workflow pain points and prioritize the manual tasks with the greatest impact

Balance automation with human control

Automate repetitive work while preserving review, validation, and approval at critical steps

Design for the entire workflow

Consider multiple users, handoffs, and dependencies to create connected end-to-end experiences

Explore more work