Day 5 Summary Codex AI Architect Technical Design

Day 5. Service Design (UI/UX Review) Training Summary

Based on the created screen specifications and path design, we derive usability vulnerabilities from the user's perspective, review layout consistency between screens, and master the architectural UI/UX design guidelines that summarize the flow depth that achieves the final goal.

Part 1. Introduction background
Part 2. Practice objectives
Part 3. Follow along
Part 4. How it works
Part 5. Training Guide
Part 6. Final Task

Limiting barriers to usability design

Errors in understanding between planning, design, and development at the project site directly lead to database omission and loss of consistency. Find out why ensuring consistency at the design stage is important.

understanding error

Due to the obsession with function-oriented design, there is a phenomenon that novice users who log in for the first time end up quitting the app midway because they do not understand the UI.

vocabulary error

Button positions, sizes, and labels (ID vs. customer ID) are configured differently on each screen, increasing user learning fatigue.

shared error

It reduces final payment conversion rates by forcing users to go through an excessively long nine-step screen transition to achieve their goal (booking a consultation).

ChatGPT vs Codex Client (Workspace-based)

Unlike ChatGPT, which is limited to a single conversation context, we compare the characteristics of Codex Client, which recognizes the entire folder structure and dependency relationships between multiple files.

Compare items ChatGPT (General Conversational AI) Codex Client (Workspace-based AI)
Context awareness Comprehensive comparative analysis of multiple screen plan files and user inflow depth movement paths Three-dimensional tracking of interchanged UI button design, size, alignment error, and status message specifications such as Login, Signup, My Page, etc.
Review screen consistency Only conveys convenience feedback on a single text screen Self-recommendation of consistent standard naming and DB column names based on project standard terminology dictionary
Scope of Impact Analysis Level of suggestions for fragmentary schema errors or hypothetical table modifications Reverse inference on list of API specifications and screen modification areas according to subtle changes in schema (e.g. column addition)

Summary of final results to be created today

Day 5 List of final products the learner produces during the session and the purpose of those products.

Task 01. Data structuring

Entities and relationships are extracted and rendered from the natural language plan. Mermaid ERD scripts and statements (Output: erd-schema.md)

Task 02. Review screen consistency

Fragmented Korean/English planning terms defined in standard English column format Project Standards Glossary (Output: project-glossary.md)

Task 03. Improving user flow

API and screen modification list according to addition of reservation approval system (WAIT_APPROVE) Change impact analysis report (Output: change-impact-report.md)

Mission 01. Design record

Archives the reasons for normalization and data design decisions and rejected alternatives. Design Decision Log (DDL) (Output: design-decision-log.md)

Mission 02. Evaluate service quality with AI

Contains DB improvement measures for concurrent payment errors, referential integrity loss of withdrawal, and transaction rollback. Edge Case Analysis Report (Output: edge-case-analysis.md)

Input / Output data pipeline

This is a mapping between the input values used in the practice stage and the results generated through Codex Workspace inference.

01
data structure
02
Terminology standardization
03
Impact of change
M1
decision making
M2
Exception scenario
Task 01. Review the screen from the user’s perspective
📥 Input (planning meeting minutes)
• day5_meeting_notes.md (meeting minutes) - User flow scenario and reservation process content in the form of irregular lines
Codex Workspace
💾 Output (ERD specification)
erd-schema.md with Mermaid ERD code - Visualization of Primary Key/Foreign Key relationships of User, Expert, Reservation, and Payment entities

STEP 1. Task 01 - Review the screen from the user’s perspective

Enter meeting scenario lines to derive precise normalization relationships (1:N, 1:1) and Mermaid codes.

🔍 Codex request prompt (data structuring)

Based on the screen specifications of the provided reservation application details form, please review whether the information structure is clear from the perspective of a novice user who is new to the service, point out elements that impede usability, and write a report suggesting improvements (UX Review Report).
  • Check whether the PK and FK settings of the four major entities, User, Expert, Reservation, and Payment, do not conflict with the scenario logic.
  • When rendering a Mermaid diagram (`erDiagram`), check that the cardinality notation is correct.

STEP 2. Task 02 - Review screen consistency

We build a master dictionary of domain terms by comparing and analyzing the gap between Figma planning terms and DB development specifications.

🔍 Codex request prompt (generating a glossary of terms)

Please compare the design details of each screen in the provided day5_signup_flow.md, analyze elements that lack consistency in button placement rules, terms used, and feedback messages, and create a screen comparison report.
  • Check whether the English standard naming and column type information that matches the business domain are presented correctly.

STEP 3. Task 03 - Improving user flow

Understand the scope of API and table impact when a release plan change adds an 'Approval' step to the scheduling process.

🔍 Codex request prompt (understand scope of influence)

Please analyze the reservation detailed scenario flow in the provided day5_signup_flow.md and create a user journey improvement report (User Journey Report) that can minimize the click steps and screen movement until the user completes the consultation reservation.
  • Check whether status changes in the Reservations table, new API specifications, and button state transition chain effects on the planning screen have been detected.

STEP 4. Mission 01 - Become a first-time user (Design Decision Log)

Log rejection reasons and agreements regarding normalization decisions, physical separation decisions for reservation history, etc.

🔍 Codex request prompt (decision record)

You are a self-employed person in your 50s who is somewhat unfamiliar with using digital smartphone apps. While looking at the membership registration and first screen UI specifications written in the provided day5_signup_flow.md specification, please write down your emotional reaction and improvement requirements about what is most confusing and difficult when signing up from your perspective.
  • Check whether the background of the decision to separate history tables (DEC-003) and the alternatives (performance degradation when unifying tables) are clearly logged.

STEP 5. Mission 02 - Evaluating service quality with AI (Edge Case)

We audit system vulnerabilities to prevent referential integrity violations (withdrawals) and transaction collapses.

🔍 Codex request prompt (Edge case derivation)

Based on the provided day5_signup_flow.md specification, please give a score out of 5 for each item such as 1) information readability, 2) operation accessibility, 3) error resilience, and 4) movement efficiency, and build a service quality evaluation (Usability Evaluation) that points out strengths and painful areas for improvement along with a comprehensive evaluation score.
  • Check the plan to maintain reference consistency and distributed transaction rollback response by introducing soft delete (logical deletion).

How it works and its core mechanisms

Schema context-aware reasoning

Beyond simple text matching, it identifies the referential integrity (FK) relationships and transaction life cycle of entities and automatically deduces natural language proposals into Data Definition Language (DDL) queries and state machine rules.

Meta term clustering

We derive a master dictionary of enterprise terms by semantically clustering differently written terms (customer name vs. nickname), and synchronize the consistency of the data dictionary based on this.

Technical design synergy of five Tasks/Missions

We check how the data and analysis information derived from each step goes through the technical design loop to maximize quality.

[Novice user usability review] (Task 01. Identifying inconveniences)
[UI style consistency check] (Task 02. Unified standard)
[User Journey Shortening Design] (Task 03. Movement Line Optimization)
[Collection of virtual customer trouble logs] (Mission 01. First impression feedback)
[Usability numerical quality diagnosis] (Mission 02. Score prescription)

This usability diagnosis pipeline eliminates usability errors and design inconsistencies before writing a single line of code, ensuring that users can easily and conveniently understand and use the product.

💡 Educational design specifications

  • Difficulty level: ★★★★☆ (Advanced - includes database design normalization and distributed transaction concepts)
  • Estimated lab time: 1 hour 30 minutes ~ 2 hours (comprehensive design practice)
  • Practical usability: ★★★★★

Learner questions/discussion topics

This is a list of topics that are great for discussion during training or to use as interview questions.

Q1. Membership Withdrawal Strategy

“In case of a membership registration error, all entered data is lost. How should we balance this from a security perspective and a user experience (UX) perspective?”

Q2. Terminology Debt

“When a project progresses halfway through without creating a dictionary to review screen consistency, how does the ‘terminology debt’ cost incurred within the development team affect the actual schedule?”

Planning and design stage final training task

This is a diagram of the overall practice tasks and missions covered on Day 5.

Task topic How to use AI
Task 01 Review the screen from the user's perspective AI examines irrationality in screen placement from a novice user's perspective and develops countermeasures
Task 02 Review screen consistency Establish rules by comparing and analyzing the unification of terminology and button position symmetry across multiple pages
Task 03 Improving user flow Flow design that reduces the number of user clicks by shortening unnecessary individual guidance screens
Mission 01 Become a first-time user A smart device injects an unfamiliar, elderly persona to recreate a virtual sign-up predicament.
Mission 02 Evaluate service quality with AI Comprehensively evaluate usability, understanding, convenience, etc. to convert grades and scores
← Day 5 Classroom entrance View full roadmap