AI-Powered, Automated Sales Order Entry Product
->
OVERVIEW
As the first and sole designer, I created a trustworthy, AI-powered data entry platform focusing on simplicity.
Summary
Helped employees at manufacturing and distribution companies transition from manual, paper-based purchase orders to automated workflows. I focused on integrating AI familiarity and error-handling awareness into the product to reduce the learning curve and support their data-intensive work environment. To ensure ease of use, I designed interfaces aligned with users’ mental models.
Role
Product design
UX/UI design
Design system
Information Architecture
User interview
Usability testing
Website design elements
Company
Endeavor AI
Tools
Figma
Adobe Suite
Google Suite
Tailwind
React
Visual Studio Code
Duration
Jan 2025 - March 2025
Role
Company
Tools
Duration
Product Design
UX/UI Design
Design System
User Interview
Usability Testing
Website Design Elements
Endeavor AI
early stage startup



Figma
Adobe Suite
Google Suite
Tailwind
Visual Code
Jan 2025 - Mar 2025




FINAL DESIGN
Seamless AI-Automated Dashboard and Sales Order Entry Page.
I built the product from 0 to 1 to enable fast and accurate data confirmation based on the Usability Test insights.
Role
Company
Tools
Duration
Product Design
UX/UI Design
Design System
User Interview
Usability Testing
Website Design Elements
Endeavor AI
early stage startup



Figma
Adobe Suite
Google Suite
Tailwind
Visual Code
Jan 2025 - Mar 2025




AI Match Confidnece
Problem
Users were overwhelmed by ambiguous matches and had to rely on memory to cross-check ERP and customer item manurial data.
Solution
Introduced color-coded confidence levels and top match suggestions to reduce ambiguity and lighten cognitive load.
Single Order Page & AI Workflow Guidance
Problem
Repetitive screen transitions and unclear task boundaries caused cognitive overload and made it difficult for users to stay focused.
Solution
Unified the interface into a single page to reduce friction, and visually chunked complex tasks into guided units to help users complete their workflow efficiently.
AI Order Orchestration
Problem
Struggle to structure and prioritize large volumes of Purchase Orders, and cognitive overload from manually calculating item-level price differences.
Solution
Grouped orders based onclient-specific business logic and surfaced price discrepancies using badge indicators and contextual tooltips—reducing decision time and easing cognitive load.
Error handling
Problem
Unclear error feedback leaves users unsure of the root cause, whether it's a system issue or missing data, resulting in confusion and wasted time.
Solution
Designed clear, actionable error messages that explain why the error occurred and how to fix it, enabling users to resolve issues quickly.
DASHBOARD
Empower users to start confidently by reducing ambiguity and decision fatigue
Jobs To Be Done
When overwhelmed by piles of Purchase Orders, they have to quickly identify what to process first in order to efficiently create Sales Orders.
Problem
Users struggle to prioritize PO documents due to lack of structure, which leads to cognitive overload and workflow delays.
How To Solve
Designed the AI-determined match certainty using color coding and clear button labels to help users recognize their current task status and decide how much effort to invest.
ORDER PAGE
Enhance focus and efficiency by simplifying the workflow into a single page.
Jobs To Be Done
Users need to confirm the extracted information and matches item by comparing Purchase Order document and ERP item list.
Problem
Constant switching between paper and screen causes fatigue and breaks focus. Their unfamiliarity with AI automation also makes it difficult to adapt to the new workflow.
How To Solve
▪︎ To help users compare information without excessive eye shifting, I aligned each interface section with the corresponding position of the data in the original document and simplified tasks into a single-page workflow
▪︎ To help users recognize the edit feature intuitively, I applied the same UI box style from the Excel-like layout they already use at work, reducing the unfamiliarity of AI.
▪︎ Users struggled with the process after completion, so I designed clear explanation and guidance to prevent friction.
ORDER PAGE
Simplified task breakdown using a card-type layout and designed AI into intuitive UXUI
Jobs To Be Done
When handling large volumes of item and pricing data, users want to easily identify what actions to take.
Problem
Users struggled to remember customer and ERP item IDs. They are overwhelmed by item data and drained by price comparisons between ERP and customer systems due to the heavy information load.
How To Solve
▪︎ Designed a card-type layout to reduce user confusion by visually chunking workloads into clear, actionable units.
▪︎ Utilized the AI-calculated price difference feature in badge and tooltip designs to help users save time and reduce cognitive load.
▪︎ By displaying AI-identified top matches with ranking in the search UI, it reduced users' reliance on memory and made the task easier.
ERROR INTERACTION
Improve accuracy with clear, actionable error messages that help prevent user mistakes
Jobs To Be Done
Users require to catches every order entry error before they happen for anomalies, mismatched pricing, and missing order details.
Problem
Unclear error feedback leaves users unsure of the root cause, whether it's a system issue or missing data, resulting in confusion and wasted time.
How To Solve
Provide a clear explanation of why the error occurred and what users need to do to resolve it quickly.
DASHBOARD
Empower users to start confidently by reducing ambiguity and decision fatigue through clear guidance.
Jobs To Be Done
When overwhelmed by piles of Purchase Orders, they have to quickly identify what to process first in order to efficiently create Sales Orders.
Problem
Users struggle to prioritize PO documents due to lack of structure, which leads to cognitive overload and workflow delays.
How To Solve
Color coding in match certainty and clear button labels help users recognize their current task status and decide how much effort to invest.
Designed the AI-determined match certainty using color coding and clear button labels to help users recognize their current task status and decide how much effort to invest.
ORDER PAGE
Enhance focus and efficiency by simplifying the workflow into a single page.
Jobs To Be Done
Users need to confirm the extracted information and matches item by comparing Purchase Order document and ERP item list.
Problem
Users struggle to remember customer and ERP item IDs, while constant switching between paper and screen causes fatigue and breaks focus. Their unfamiliarity with AI automation also makes it difficult to adapt to the new workflow.
How To Solve
Simplified tasks into a single-page workflow with AI suggestions and a familiar Excel-style layout to improve usability and reduce friction.
ORDER PAGE
Simplified task breakdown using a card-type layout that helping users reduce confusion.
Jobs To Be Done
When handling large volumes of item and pricing data, users want to easily identify what actions to take.
Problem
Users are overwhelmed by item data and drained by price comparisons between ERP and customer systems due to the heavy information load.
How To Solve
Designed a card-type layout to reduce user confusion by visually chunking workloads into clear, actionable units. Badges help guide users through their tasks, while tooltips simplify complex actions by providing detailed, contextual information.
ERROR INTERACTION
Improve accuracy with clear, actionable error messages that help prevent user mistakes
Jobs To Be Done
Users require to catches every order entry error before they happen for anomalies, mismatched pricing, and missing order details..
Problem
Unclear error feedback leaves users unsure of the root cause, whether it's a system issue or missing data, resulting in confusion and wasted time.
How To Solve
Provide a clear explanation of why the error occurred and what users need to do to resolve it quickly.
AI Match Confidence
Empower users to start to work assuredly by reducing ambiguity and decision fatigue.
Jobs To Be Done
Users create sales orders by accurately distinguishing each client's Purchase Order Information.
Problem
Users struggled to remember customer and ERP item IDs. They are overwhelmed by item data between ERP and customer systems due to the heavy information load.
How To Solve
Color coding in match certainty and clear button labels help users recognize their current task status and decide how much effort to invest.
▪︎ Designed AI-informed match certainty fuction by using color coding and clear button labels to help users to decide how much effort to invest in worflow decision making.
▪︎ Displayed AI-identified top matches in the search UI, reducing users' reliance on memory, making the work easier.
Single Order Page
Enhance focus and efficiency by simplifying the workflow into a single page.
Jobs To Be Done
Users need to confirm the extracted information and match items by comparing the Purchase Order document and ERP item list.
Problem
Constant switching between screens causes fatigue and breaks focus. User unfamiliarity with AI automation also makes it difficult to adapt to new workflow.
How To Solve
▪︎ Aligned UI sections with the corresponding position of the data in the original document and simplified into a single-page workflow.
▪︎ Applied consistant UI box style from the Excel-like layout they already use at work to reduce the unfamiliarity of AI.
AI-driven Order Orchestration
Minimize user effort in performing tasks like calculation and prioritization.
Jobs To Be Done
When faced with a large volume of Purchase Orders, including items and pricing data, users have to quickly decide what to process first in order to create Sales Orders.
Problem
Users struggle to prioritize PO documents due to lack of structure and calcuate the price differences, which leads to cognitive overload and workflow delays.
How To Solve
▪︎ Utilized the AI-calculated price difference feature in badge and tooltip designs to help users save time and reduce cognitive load.
▪︎ Leveraged AI to prioritize and group sales orders based on each client's business logic requirements.
AI Workflow Guidance
Simplified task breakdown using a card-type layout that helps users reduce confusion.
Jobs To Be Done
When handling large volumes of items in a Purchase Order, users need to quickly and accuratly identify what actions to take.
Problem
Users were confused by unclear task boundaries and post-completion workflows.
How To Solve
▪︎ Designed a card-type layout to reduce user confusion by visually chunking workloads into clear, actionable units.
▪︎ Guided users through their tasks with badges, and tooltips simplify complex actions by providing detailed, contextual information to prevent mistakes.  
Error Handling
Improve accuracy with clear, actionable error messages that help prevent user mistakes.
Jobs To Be Done
Users are required to catch every order entry error such as anomalies, mismatched pricing, and missing order details before they happen.
Problem
Unclear error feedback leaves users unsure of the root cause, whether it's a system issue or missing data, resulting in confusion and wasted time.
How To Solve
Provide a clear explanation of why the error occurred and what users need to do to resolve it quickly.
DESIGN PROCESS
Established a fast-paced design iteration and execution process.
Get Context
User interviews,
Analyzed manual
and automated
user journeys.
Ideation
Sketching, Wireframing
Design
High Fidelity Prototyping
Usability Test
A/B Test, User feedback from internal testing
Iteration
Visual design exploration,
Design System
Understand context
User interviews focused on manual workflows and AI familarity.
Ideation
Sketching,Wireframing
Design
High Fidelity Prototyping
Usability Test
User feedback from internal testing
Iteration
Visual exploration, Design system
PRODUCT GOAL
Create an intuitive AI-automated platform for users unfamiliar with AI, and build a flexible prototyping system that enables fast, tailored solutions for diverse client needs.
Product
Business
Built an intuitive AI automation experience for newcomers.
Designed for rapid UX/UI customization to support specific business logic within a limited timeline.
PROBLEM
AI unfamilarity + UX Friction = Workflow Inefficiency
Understand context
User interviews focused on manual workflows and AI familarity.
Ideation
Sketching,Wireframing
Design
High Fidelity Prototyping
Usability Test
User feedback from internal testing
Iteration
Visual exploration, Design system
Jobs To Be Done
Users must accurately enter Purchase Order details to create a Sales Order, but they have to rely on memory to match customer items with vendor items.
INFORMATION ARCHITECTURE
Streamlined workflow from 6 steps to 2, with human control.
Product
Business
Designed to reduce the learning curve and support users' data-heavy workflows by aligning with their mental models, using familiar elements like Excel-style layouts and color coding.
Implemented an Design System to support the customization of clients' business logic by streamlining engineering workflows and saving time.
Information Architecture
USABILITY TEST
Identified users' preferred interaction patterns and the reasons behind them.
Product
Business
Designed to reduce the learning curve and support users' data-heavy workflows by aligning with their mental models, using familiar elements like Excel-style layouts and color coding.
Implemented an Design System to support the customization of clients' business logic by streamlining engineering workflows and saving time.
Information Architecture
“Hate scroll and transitions, distracting the focus.”
“Feel pressure to remember Item ID exactly.”
“Hate horizontal scroll, distracting the focus.”
Color coding helps users anticipate the effort required for upcoming tasks and prioritize their workflow.
Price information between a customer and a vender needs to be clear for calculation.
SOLUTIONS
Reduced the learning curve, clarify decisions, and simplify execution.
Product
Business
Designed to reduce the learning curve and support users' data-heavy workflows by aligning with their mental models, using familiar elements like Excel-style layouts and color coding.
Implemented an Design System to support the customization of clients' business logic by streamlining engineering workflows and saving time.
Information Architecture
AI Match Confidence
Applied AI to evaluate match scores and suggest the best-fit relevant alternatives.
AI-driven Order Orchestration
Leverage AI to prioritize sales orders and calculate price differences.

AI Workflow Guidance
Simplified task breakdown for smooth AI-human collaboration.

SOLUTION
Built an intuitive UX and UI to reduce the learning curve and make data editing easier. Devised a reusable design system to support business logic customization and accelerate development.
AI Match Confidence
Applied AI to evaluate match scores and suggest the best-fit relevant alternatives.
AI-driven Order Orchestration
Leverage AI to prioritize sales orders and calculate price differences.
AI Workflow Guidance
Simplified task breakdown for smooth AI-human collaboration.
Product
Business
Designed to reduce the learning curve and support users' data-heavy workflows by aligning with their mental models, using familiar elements like Excel-style layouts and color coding.
Implemented an Design System to support the customization of clients' business logic by streamlining engineering workflows and saving time.
Information Architecture
Design
I built the AI automated Sales Order Entry product from 0 to 1 to help users quickly and efficiently complete data confirmation across diverse purchase orders by simplifying the process from User Testing and Interview insights.
Color coding and Button labels that describe specific actions help users decide how much effort is required for the next task.
Combining the Extracted Information and Matches Results sections with the search experience into a single page helped users save time, reduce effort, and work more efficiently. The Excel-style layout optimized familiarity in AI automation workflows.
Card types help users intuitively break down their tasks and reduce confusion.
Tooltips and button labels help users predict outcomes and avoid mistakes. Badges guide users through workloads, and tooltips simplify complex tasks by providing detailed information.
ITERATION
Before
Improved visual hierarchy to enable more intuitive, refined workflows.
Before - Dashbaord
After - Dashboard
Before - 6 steps of Order Page
After - Single Order Page
DESIGN SYSTEM
Built a scalable design system to accelerate custom B2B demos.
As a sole designer, I built the entire design system for rapid design customization for demo to support specific business logics since the lack of a structured system for primary design components led to process delays.
Built desktop screens for web platform.
Screen Size
Space
Layout Columns
To ensure consistency, the screen is responsive to screen height, with a fixed width of 1440px.
Custom breakpoint in Tailwinds: 1440px
To help position elements in the interface, columns are proposed to serve as a guide in this step.
Standard spacing: 24px
Spacing scale: 8px
To help position elements in the interface, the design uses a 5 column layout.
Margins: 24px
Gutters: 56px
Column grid: 5
Used Tailwind CSS color tokens to ensure consistency between design and development, and to optimize collaboration with engineers.
Blue
Blue 800
#1E40AF
Blue 600
#2563EB
Blue 500
#3B82F6
Blue 100
#DBEAFE
Blue 50
#EFF6FF
Black/White
Zinc 900
#18181B
Blue 600
#2563EB
Blue 500
#3B82F6
Blue 100
#DBEAFE
Blue 50
#EFF6FF
Blue 50
#EFF6FF
Blue 50
#EFF6FF
Blue 50
#EFF6FF
Tertiary
Zinc 900
#18181B
Blue 500
#3B82F6
Blue 50
#EFF6FF
Blue 50
#EFF6FF
Blue 600
#2563EB
Blue 100
#DBEAFE
Blue 50
#EFF6FF
Blue 50
#EFF6FF
Error Red
Zinc 900
#18181B
Blue 500
#3B82F6
Text alignment is primarily applied Left Align.
Used System Icons. 16px, 24px, 40px
Headline Large - Inter Semibold 20px
Headline Medium - Inter Medium 16px
Title Large - Inter Semibold 14px
Title Medium - Inter Medium 14px
Title Small Text 2 - Inter Medium 12px
Body Large - Inter Regular 14px
Body Medium - Inter Regular 12px
Body Small - Inter Light 12px
Label Large - Inter Regular 12px
Used to convey dynamic information, such as a count or status. A badge can include text, labels, or numbers.
Used for single-step actions and as Calls to Action (CTAs), while using icons to quickly convey actions at a glance.
Typically used when users have multiple items to choose from. These are used in components like Sales Order Filter, See More Match Items, and Shipment Type, where an action is triggered based on the user’s selection.
Enabled users to easily search for the information and data they needed to complete their tasks using Endeavor’s data.
Used in forms to help users modify extracted information from Endeavor. Input fields are typically found within forms, but can also appear in modals, search interfaces, or within requested items in the Matched Results section.
Used to provide specific information that helps users take clear action and understand the next steps. These are commonly found in buttons, icons, and error notifications.
Set of components.
WHAT I LEARNED
As a designer, I recognized the importance of building trustworthy interactions between AI and users, and I transformed tight time constraints into opportunities for rapid yet thoughtful problem solving.
At Endeavor AI, I gained the ability to design efficiently and prioritize time-saving processes for the engineers — a critical skill in fast-paced startup environments. The company's vision was to deliver fully customized AI agents products within a week, a level of speed and adaptability that's rare in large organizations. This fast iteration cycle become my key differentiator and taught me how to design for impact under tight timelines.