Day 1 Codex AI Requirements Setup & lab missions

Codex-Based Requirements Definition Mission

Learn the full planning workflow: summarize (clean up) an unstructured meeting conversation, define feature specs and exception flows, build a decision question list, then combine the final docs & share them by email.

Choose a step

AI Task 1

Clean up the conversation summary

Pull key summary notes from the raw meeting transcript

AI Task 2

Analyze meeting notes

Sort users, features and business policies from the meeting summary

AI Task 3

Define functional requirements

Turn feature candidates into testable, standard spec sentences

AI Task 4

Normal and exception flows

Add detail to user scenarios and exception cases (failure, duplicates, etc.)

AI Task 5

Open planning questions

Turn undecided items into a list of questions for stakeholders

AI Task 6

Combine docs and share by email

Finish docs with an AI lint check and send via the Gmail plugin

Skill Task 1

Define the service problem and Personas

[ADD-D01-T01] Define goals and needed data for Viewer·Creator·Advertiser·Admin

Skill Task 2

Sort Features into Domains

[ADD-D01-T02] Place key features like Members, Video, Playback into owning Domains

Skill Task 3

Review the Domain Map and responsibility boundaries

[ADD-D01-T03] Visualize Domain dependencies and draft Bounded Contexts

Mission 1

Pet Care Matching

[Assignment] Write planning docs from a pet sitter matching meeting transcript

Mission 2

Self-Service Study Room & IoT

[Assignment] Spec real-time space booking and smart device control

Mission 3

Short-Form & Live Commerce Domains

[Skill Assignment 1] Personas, Feature Domain placement and Bounded Contexts

Mission 4

Smart B2B Mobility Control

[Skill Assignment 2] Data ownership and Domain boundaries for a company-vehicle control platform

Step 0 Pull a key summary from a detailed transcript (cleanup)

📥 Step input

Raw 1st meeting transcript in back-and-forth chat form (meeting-notes-raw.md)

📤 Step output

Key summary notes file (meeting-notes.md)

🛠️ Detailed follow-along guide

  1. Get the raw notes: press the download button below to download meeting-notes-raw.md, the raw 1st meeting transcript, and put it in your work folder, or create the file content.
  2. Ask for cleanup: open your Codex/ChatGPT chat and send the Codex request example prompt below.
  3. Save the result: copy the clean summary notes AI pulled out of the raw chat and save them as meeting-notes.md. (This file is the Input for the next step 1 lab.)

Mission situation

The raw notes from the planning meeting mix back-and-forth talk and small talk between team members (CEO, planner, developer, designer), so they're too messy to pull requirements from directly. Learners must use Codex as a cleanup tool to produce planning notes that summarize only the key business requirements.

Learner mission

Write or download the following raw 1st meeting transcript as meeting-notes-raw.md.

meeting-notes-raw.md original (scroll)# 1st Kick-off Meeting Notes: Building an Expert Consultation Service

**Date**: 2026-07-10 14:00 - 15:30
**Attendees**: CEO, Planner (PM), Developer (Dev), Designer

---

**CEO**: Hello, everyone. Today's meeting is to discuss the basic requirements and set the direction for the "online expert consultation service" we're launching. The core is a process where users come to our platform, find experts in various fields, book a session and finish paying smoothly.

**Planner**: Yes. Let's start with the basic sign-up process. Do users have to sign up to use the service?

**CEO**: Yes, that's right. Users should finish signing up first before they can properly use features like expert search and booking. Without an account, they can only browse a little.

**Designer**: After signing up they'll search for experts — how are you thinking about expert profile pages and categories?

**CEO**: Experts should be grouped by field first. For example, categories like IT consulting, legal advice and design coaching. Users should be able to pick a category or type a search term to quickly find the expert they want.

**Designer**: So search results show a list of experts, and you go into a detail page to pick a time. How do they choose the date and time?

**CEO**: Yes, when a user picks an expert, the expert's available schedule should show as a calendar or timeline. Users should be able to freely pick the date and time they want and request a booking.

**Developer**: How does payment work right after the booking request? Is the booking confirmed only when payment happens together with the request?

**CEO**: The booking should move to "confirmed" only when payment is complete. If payment fails because of a card limit or another reason, of course that booking must not be completed — it should stay pending or be cancelled.

**Developer**: So the business rule is that if payment fails, the booking isn't created or the slot is released. Where do experts check their booked schedule?

**CEO**: Experts should have their own dashboard in our system, where they can watch and manage booking requests and confirmed schedules in real time.

**Planner**: Expert sign-up must be different from regular sign-up — do we need a separate verification or approval process? We can't let just anyone work as an expert.

**CEO**: Good point. Experts can't start right after applying. We'll require an admin approval policy: a site admin reviews their documents or qualifications and "approves" them, and only then do they appear in the expert list and join the booking system.

**Planner**: When a booking is done, sending notifications to users and experts would help usability. Which notification channel do you prefer?

**CEO**: We definitely need notifications, to tell people when a booking is confirmed or cancelled. But whether to send them by email or SMS — we need to look more at dev resources and cost. Let's not decide that yet.

**Developer**: Do we need to set rules for cancellation fees or refunds now too?

**CEO**: The cancellation fee policy needs legal review and talks with marketing too. It's hard to decide today, so let's leave it as an undecided item as well. We'll settle it at the 2nd meeting.

**Planner**: Got it. I'll gather today's decided policies and pending items into cleaned-up meeting summary notes for the first spec draft.

Codex request example (cleanup prompt)

PromptRead the raw meeting notes in meeting-notes-raw.md, remove unneeded conversation and context between speakers,
and write compact, informal summary notes of about 15 lines covering service users, feature requirements, system policies and undecided items.
Don't imagine and add anything that isn't in the original.

🤖 Codex's role in this step

Summarizer and cleanup tool: removes small talk and speaker context from the conversation and pulls out key summary notes needed for detailed design.

Step 1 Find the key points in the meeting notes

📥 Step input

Summary planning notes file cleaned in step 0 (meeting-notes.md)

📤 Step output

  • • List of service user roles
  • • List of feature candidates
  • • List of business rules
  • • List of undecided items

🛠️ Detailed follow-along guide

  1. Prepare the meeting notes: place meeting-notes.md, the step 0 summary. You can also press the download button below to get a cleaned example file.
  2. Ask AI to analyze: continuing the step 0 chat, send the Codex request example prompt below.
  3. Review the sorting: check that the user list (Customer, Expert, Admin) and the decided/undecided policy sorting meet the learner checks.

Mission situation

Now it's time to analyze the step 0 planning summary (`meeting-notes.md`) cleaned from the conversation between the CEO, planner, developer and others. Based on this summary, learners must accurately categorize the service's main objects (user roles, features, business rules, policies).

Learner mission

Prepare the cleaned meeting-notes.md summary file or use the example download below.

meeting-notes.md summary (scroll)We want to build an online expert consultation service.

Users can search for experts after signing up.
Experts are grouped by field.
Users can choose a session date and time.
A booking is confirmed when payment is complete.
If payment fails, the booking must not be completed.

Experts can check their own bookings.
Admins can approve expert sign-ups.

A notification must be sent when a booking is complete.
Whether to send notifications by email or SMS is not decided yet.
The cancellation fee policy is not decided yet either.

Codex request example

PromptRead meeting-notes.md and separate the following.

1. Service users
2. Features users perform
3. Features the system performs
4. Business rules
5. Things not decided yet

Don't add anything not in the meeting notes,
and mark unclear items separately.

Don't change any files yet; just show the analysis.

Learner checks

  • Are users split into Customer, Expert and Admin?
  • Is the link between payment success and booking confirmation clear?
  • Are the notification method and cancellation policy separated as undecided items?
  • Did Codex avoid adding features not in the meeting notes?

🤖 Codex's role in this step

Analyst: logically identifies user role groups, spec candidates and business rules from the cleaned planning summary.

Step 2 Write functional requirement sentences

📥 Step input

Step 1 outputs (user roles, feature candidates, business rules, undecided items)

📤 Step output

functional-requirements.md (standardized functional requirements table FR-001 ~ FR-004)

🛠️ Detailed follow-along guide

  1. Create a requirements template: make a functional-requirements.md file in your work folder and paste in the table frame from the learner mission below.
  2. Send the prompt: continuing the step 1 chat, copy the Codex request example prompt below and send it.
  3. Copy & save the result: copy the spec sentences AI made in table form (Markdown), paste them into functional-requirements.md and save.
  4. Filter vague wording: if finished sentences include vague words like "conveniently" or "quickly", ask AI again to make them measurable.

Mission situation

In step 1 you found the feature candidates the service needs. In this step you turn them into real requirement sentences. Write sentences that developers and planners understand the same way, not just feature names.

How to write requirements

[User or system] must be able to    [do a feature or action]    under [a specific condition].

Learner mission

Change the feature candidates from step 1 into the following format.

Requirement ID User Feature Requirement
FR-001 Customer Sign-up (blank for learners to fill)
FR-002 Customer Expert search
FR-003 Customer Choose session time
FR-004 System Confirm booking

Codex request example

PromptWrite the feature candidates from step 1 as functional requirements.

Apply these rules.

1. Give requirement IDs in order from FR-001.
2. Write only one feature per requirement.
3. Clearly name who the user is.
4. Include the condition under which the feature runs.
5. Write sentences that can be tested.
6. Don't use vague words like "conveniently", "appropriately" or "quickly".
7. Don't add anything not in the meeting notes.

Write the result as a Markdown table.

🤖 Codex's role in this step

Technical Writer: produces measurable, checkable sentences based on templates and standard formulas.

Step 3 Add normal and exception flows

📥 Step input

Step 2 output (functional-requirements.md table)

📤 Step output

  • • Design doc with normal/exception flows added
  • • List of requirement checks
  • • List of policies needing more confirmation

🛠️ Detailed follow-along guide

  1. Pick a requirement: find the FR-004 Confirm booking item to expand in the file.
  2. Run the prompt: continuing the step 2 chat, copy and enter the Codex request example prompt below.
  3. Add the flows: paste the detailed normal flow, exception flow and check text AI returns under that ID in functional-requirements.md and save the file.
  4. Filter made-up rules: check that AI didn't imagine exception policies not in the original, and left them as "needs confirmation".

Mission situation

You've written requirement sentences, but only the normal situation is defined. In real services, exceptions like payment failure, no data and double bookings happen often. In this step you add normal and exception flows to each requirement.

Learner mission

Pick the following feature from the step 2 requirements and define its normal/exception flows.

  • Expert search
  • Choose session time
  • Payment and booking confirmation
Template### FR-004 Confirm booking

- User:
- Precondition:
- Normal flow:
- Exception flow:
- Check:

Codex request example

PromptAdd a normal flow and an exception flow
to each requirement in functional-requirements.md.

Apply these rules.

1. Write the normal flow step by step, from the user's first action to the final result.
2. For exception flows, check failure, duplicate, no data and no permission cases.
3. Don't decide exception policies that can't be confirmed from the meeting notes.
4. Mark anything that needs a policy decision as "needs confirmation".
5. Write at least one check for each requirement.

🤖 Codex's role in this step

QC Explorer: reviews and suggests responses to design flaws and abnormal input errors (duplicates, timeouts, exception handling).

Step 4 Write undecided items and planning questions

📥 Step input

From the step 3 output, the 'list of policies needing confirmation and undecided items'

📤 Step output

open-questions.md (decision planning question table)

🛠️ Detailed follow-along guide

  1. Create the question file: make an open-questions.md file in your work folder.
  2. Send the prompt: continuing the step 3 chat, send the Codex request example prompt below to get the undecided items table.
  3. Record the questions: paste the table AI made into open-questions.md and save.
  4. Watch the role: check whether AI answered questions on its own, and make sure it kept the purpose (questions stay open for a decision).

Mission situation

While detailing the requirements, we found business policies that can't be decided right away. We need to turn these undecided items into specific, organized questions to get planning feedback from stakeholders or decision makers.

Codex request example

PromptReview the current requirements doc
and find anything not decided or vague.

Write each item as a question
the planner can confirm with stakeholders.

Include the following.

1. Question ID
2. Related requirement ID
3. Question to confirm
4. Why this decision is needed
5. Features affected if not decided
6. Priority
7. Status

Don't make up answers to the questions.

🤖 Codex's role in this step

Question Maker: for pending policies, separates affected features and importance, and lists the questions needed to reach agreement in the right format.

Step 5 Create the combined requirements doc and share by email (Gmail plugin)

📥 Step input

  • • Spec files such as meeting-notes.md
  • • **Gmail Plugin** connected in Codex Client

📤 Step output

  • • requirements-specification.md (combined)
  • • Real email sent through the Gmail plugin API

🔌 [Key prep] Set up and connect the Codex Client Gmail plugin

  1. Click the [Plugins (plugin store)] icon at the top right of Codex Client.
  2. Find the `Gmail` plugin in the list and press **[Install]**.
  3. After install, a Google sign-in window appears. **Sign in with your own Google/Gmail account**, then complete **OAuth email permission approval** so the AI agent can send email.

🛠️ Detailed follow-along guide

  1. Merge docs and first check: send the prompt from **Learner mission 1 (combine and AI check)** below in the AI chat to ask for the combined requirements doc requirements-specification.md.
  2. Fix vague wording: if the error results AI found include a WARNING, refine the spec with AI and build the final approved version.
  3. Tell AI to send with the Gmail plugin: to use the connected plugin, copy the **Learner mission 2 (Gmail plugin send)** prompt and send it in the AI chat.
  4. Check it was sent: open the recipient's inbox or your own 'Sent' folder to confirm the email really went out.

Learner mission 1 (combine and AI check prompt)

Prompt 1Read the following 3 files and write requirements-specification.md, a requirements doc that merges them.

- meeting-notes.md
- functional-requirements.md
- open-questions.md

At the same time, analyze errors in the requirements doc using the 6 rules below.

[Check rules]
1. Duplicate requirement IDs
2. Missing requirement IDs
3. Missing user roles
4. Missing normal and exception flows
5. Missing checks
6. Vague wording (vague words: conveniently, quickly, as needed, appropriately, normally)

Show the check results first in the output in the format below, then show the merged requirements-specification.md content under it.

[Output format]
PASS: no duplicate requirement IDs
ERROR: FR-004 check missing
WARNING: FR-007 uses vague word "conveniently"

Learner mission 2 (Gmail plugin send prompt)

Prompt 2Using the Gmail plugin, send an email that meets the conditions below, based on the just-checked requirements-specification.md
and the undecided questions in open-questions.md.

- To: [enter the email address of the CEO or decision maker who will give feedback]
- Subject: [Approval request] Review of the expert consultation service requirements doc and settling undecided policies
- Body:
  - Report that combining and checking the planning docs is done
  - The main open questions that urgently need planning decisions (summary of Q-001 ~ Q-004)
  - Deadline for feedback replies

[Alternative] Local mail client and sending by hand

If sending with the plugin is hard because of Google account security limits, press the buttons below to write or copy the email.

🤖 Codex's role in this step

Plugin Integrator: after finishing internal doc analysis, connects remotely to the Gmail plugin API and sends business emails on your behalf.

Lab Assignment 1 Pet Home Care and Matching Service

📥 Assignment input

  • • pet-care-meeting-notes.md meeting notes
  • • The Codex prompts you learned in the step 1–5 labs

📤 Assignment output

  • • pet-care-requirements.md (feature and exception flow requirements doc)
  • • An email approval request to settle undecided policies

💡 Mission background and goal

These are planning meeting notes for a platform that connects pet owners (customers) with professional pet sitters. The goal is to clearly spec owner identity and pet info registration, pet sitter qualification review and admin approval, the payment deadline after a match is accepted, real-time care status updates, and exception handling for cancellations and pet sitter no-shows.

🛠️ Self-guided lab

  1. Download the notes: click the button below to download the original pet-care-meeting-notes.md file.
  2. Write requirements: based on the downloaded notes, tell Codex AI to spec in detail the user roles (regular user, pet sitter, admin), features (sign-up, search, booking, payment, status change) and exception flows (payment timeout, cancellation penalty, no-show handling).
  3. AI lint check: after writing the plan, use the step 5 check rules (duplicate/missing IDs, missing normal/exception flows, vague wording) to have AI review the requirements doc closely, and build the final requirements doc as the approved version.

Learner self-lab materials

Press the button below to download the raw meeting notes for Mission 1.

Recommended AI lab prompt

Mission 1 Prompt[Role: IT business analyst]
Analyze the pet-care-meeting-notes.md file I just downloaded and write the requirements doc pet-care-requirements.md in this format.

1. Overview and service structure
2. User roles (owner, pet sitter, admin)
3. Key functional requirements list (ID format: PCR-001, PCR-002...)
4. Normal and exception flows (e.g. expiring after 1 hour without payment, 200% point compensation and a replacement pet sitter suggestion when a pet sitter no-shows, etc.)
5. Open questions list

Spec it in as much detail as possible, at a buildable level, and also apply the AI lint check rules from step 5 and show the self-check results at the top.

🎯 Main assessment criteria

• Is the 'admin approval' condition after pet sitter registration clearly specified?
• Are deadline exceptions like 'cancel if not paid within 1 hour' after a match request included?
• Are the compensation policy and replacement suggestion flow for pet sitter no-shows properly written into the plan?

Lab Assignment 2 Self-Service Study Café Booking and IoT Control Service

📥 Assignment input

  • • study-cafe-meeting-notes.md meeting notes
  • • The Codex prompts you learned in the step 1–5 labs

📤 Assignment output

  • • study-cafe-requirements.md (combined requirements doc with IoT exception handling)
  • • Hardware-software control test scenario spec

💡 Mission background and goal

These are meeting notes for a service at a self-service study café: booking study rooms on mobile and remotely controlling offline hardware (door locks, smart power cut-off, thermostats, etc.). You must build a complete IoT business spec by working out several safety processes: generating a PIN code when payment completes, syncing IoT device states and PIN limits at check-in/out, and emergency door opening and alarm alerts when IoT communication fails (e.g. door lock not working).

🛠️ Self-guided lab

  1. Download the notes: click the button below to download the original study-cafe-meeting-notes.md file.
  2. Write requirements: based on the downloaded notes, tell Codex AI to spec the user and hardware flows (activate 10 minutes before booking, notice 5 minutes before end, cut power at end).
  3. Detail failure handling: clearly write how to handle each possible scenario like IoT connection loss or a faulty door lock (server backup network, emergency alert to admin on send failure, emergency call, etc.) using fail-safe design.

Learner self-lab materials

Press the button below to download the raw meeting notes for Mission 2.

Recommended AI lab prompt

Mission 2 Prompt[Role: IoT systems business analyst]
Based on the downloaded study-cafe-meeting-notes.md, write study-cafe-requirements.md including the following.

1. Normal flow for study room booking and payment
2. Hardware control timing (activate PIN and turn on air conditioning 10 minutes before booking, audio notice 5 minutes before end, cut off immediately at end)
3. Exception and failure recovery flows (Fail-safe):
   - When the IoT door lock network goes offline, switch automatically to a backup wireless network and an emergency remote-open process
   - When the PIN code fails to send, provide an emergency manual open button and rules for an instant admin dashboard alarm
4. Requirement checks

Organize the flow in great detail as a timeline so developers and HW engineers can connect things with no confusion.

🎯 Main assessment criteria

• Are timeline conditions for hardware control (door lock, lights, air conditioning) before/after booking clearly stated?
• Does the requirements doc cover a two-layer emergency procedure for physical failures like an IoT device losing network?
• Are booking business rules like non-member booking limits and extension payment policy written as standard sentences?

Skill Task 1 ADD-D01-T01 — Define the Service Problem and Personas

🏷️ Skill tags

Task area: HYBRID Tech Skill: TECH-BA TECH-DDD AI Skill: AI-CTX AI-ANA AI-DOC AI role: Domain Analyst

📥 Input Context

  • • Service idea and existing Day 1 requirements outputs
  • • Candidate user data

📤 Output to submit

  • • 01_SERVICE_OVERVIEW.md
  • • 03_DOMAIN_MAP.md

💡 Task goal and background

Define the goals, actions, features and data needed for Viewer·Creator·Advertiser·Admin. Give AI the Domain Analyst role and a service scenario, and find goal conflicts and missing items between Personas.

🛠️ Steps

  1. Prepare the Workspace: put the input docs from earlier Days or LMS Tasks into the project Workspace.
  2. Assign the role: separate decided items, assumptions and undecided items in the input docs, and give AI the Domain Analyst role.
  3. Define Personas: define the goals, actions, features and data needed for Viewer·Creator·Advertiser·Admin.
  4. Check results: don't take AI results as-is; check facts, structure and exceptions against the Tech Skill criteria (TECH-BA, TECH-DDD).
  5. Record outputs: record your final choices, what you changed, and AI suggestions you rejected and why in the output.

Recommended AI prompt

ADD-D01-T01 PromptYou are the Domain Analyst for this project.

[Project Context]
Service idea, existing Day 1 requirements outputs, candidate users

[Task]
Define the goals, actions, features and data needed for Viewer·Creator·Advertiser·Admin.

[AI use]
Give AI the Domain Analyst role and a service scenario, and ask it to find goal conflicts and gaps between Personas.

[Output format]
1. Facts confirmed in the current Context
2. Suggested results and reasons
3. Gaps, conflicts and risks
4. Questions a person must decide
5. A checkable done Checklist

Don't present guesses as facts; always mark them as 'assumptions'.

🎯 Main assessment criteria

• Are each Persona's goals, actions, features and data organized in a table and clearly separated?
• Can technical outputs and AI review results be traced to each other?
• Is at least one reason recorded for accepting, changing or rejecting an AI suggestion?
• Can the next Task reuse the result as input Context without extra questions?

Skill Task 2 ADD-D01-T02 — Sort Features into Domains

🏷️ Skill tags

Task area: TECH Tech Skill: TECH-BA TECH-DDD AI Skill: AI-ANA AI-REV AI role: Domain Analyst

📥 Input Context

  • • ADD-D01-T01 output (Personas and functional requirements)

📤 Output to submit

  • • 01_SERVICE_OVERVIEW.md
  • • 03_DOMAIN_MAP.md

💡 Task goal and background

Place Members, Channel, Video, Playback, Engagement, Advertising, Ad Serving, Billing and Analytics features clearly in their owning Domains, and refine main and shared responsibilities.

🛠️ Steps

  1. Prepare Workspace input: prepare the step 1 output as the input doc.
  2. Assign the AI role: give AI the Domain Analyst role and ask for candidate owners and alternatives for each feature.
  3. Place in Domains: place Members, Channel, Video, Playback, Engagement, Advertising, Ad Serving, Billing and Analytics features in their owning Domains.
  4. Decide final boundaries: a person picks the final boundary from AI's alternatives and records the reason.

Recommended AI prompt

ADD-D01-T02 PromptYou are the Domain Analyst for this project.

[Project Context]
Service idea, existing Day 1 requirements outputs, candidate users

[Task]
Place Members, Channel, Video, Playback, Engagement, Advertising, Ad Serving, Billing and Analytics features in their owning Domains.

[AI use]
Have AI suggest candidate owners and alternatives for each feature, then a person picks the final boundary and records the reason.

[Output format]
1. Facts confirmed in the current Context
2. Suggested results and reasons
3. Gaps, conflicts and risks
4. Questions a person must decide
5. A checkable done Checklist

Don't present guesses as facts; always mark them as 'assumptions'.

🎯 Main assessment criteria

• Does every key feature have one main owning Domain, with shared responsibilities marked separately?
• Can technical outputs and AI review results be traced to each other?
• Is at least one reason recorded for accepting, changing or rejecting an AI suggestion?

Skill Task 3 ADD-D01-T03 — Review the Domain Map and Responsibility Boundaries

🏷️ Skill tags

Task area: AI Tech Skill: TECH-DDD TECH-ARCH AI Skill: AI-CTX AI-REV AI-DOC AI role: Domain Analyst

📥 Input Context

  • • ADD-D01-T01, ADD-D01-T02 outputs

📤 Output to submit

  • • 01_SERVICE_OVERVIEW.md
  • • 03_DOMAIN_MAP.md

💡 Task goal and background

Visualize order, dependencies and data ownership between Domains, draft Bounded Contexts, and hand them over as input for system architecture.

🛠️ Steps

  1. Visualize and structure: visualize order, dependencies and data ownership between Domains and draft Bounded Contexts.
  2. Run an AI Review: run an AI review with the Review Prompt 'analyze Domains with duplicate or missing responsibilities' and keep before/after versions.
  3. Check: don't take AI results as-is; check them against the Tech Skill criteria (TECH-DDD, TECH-ARCH).

Recommended AI prompt

ADD-D01-T03 PromptYou are the Domain Analyst for this project.

[Project Context]
Service idea, existing Day 1 requirements outputs, candidate users

[Task]
Visualize order, dependencies and data ownership between Domains and draft Bounded Contexts.

[AI use]
Run an AI review with the Review Prompt 'analyze Domains with duplicate or missing responsibilities' and keep before/after versions.

[Output format]
1. Facts confirmed in the current Context
2. Suggested results and reasons
3. Gaps, conflicts and risks
4. Questions a person must decide
5. A checkable done Checklist

Don't present guesses as facts; always mark them as 'assumptions'.

🎯 Main assessment criteria

• Are the Domain Map, responsibility table, key Entity candidates and applied AI review notes kept?
• Can technical outputs and AI review results be traced to each other?
• Is at least one reason recorded for accepting, changing or rejecting an AI suggestion?

Lab Assignment 1 Pet Home Care and Pet Sitter Matching Service

📥 Assignment input

  • • pet-care-meeting-notes.md meeting notes
  • • The Codex prompts you learned in the step 1–5 labs

📤 Assignment output

  • • pet-care-requirements.md (feature and exception flow requirements doc)
  • • An email approval request to settle undecided policies

💡 Mission background and goal

These are planning meeting notes for a platform that connects pet owners (customers) with professional pet sitters. The goal is to clearly spec owner identity and pet info registration, pet sitter qualification review and admin approval, the payment deadline after a match is accepted, real-time care status updates, and exception handling for cancellations and pet sitter no-shows.

🛠️ Self-guided lab

  1. Download the notes: click the button below to download the original pet-care-meeting-notes.md file.
  2. Write requirements: based on the downloaded notes, tell Codex AI to spec in detail the user roles (regular user, pet sitter, admin), features (sign-up, search, booking, payment, status change) and exception flows (payment timeout, cancellation penalty, no-show handling).
  3. AI lint check: after writing the plan, use the step 5 check rules (duplicate/missing IDs, missing normal/exception flows, vague wording) to have AI review the requirements doc closely, and build the final requirements doc as the approved version.

Learner self-lab materials

Press the button below to download the raw meeting notes for Mission 1.

Recommended AI lab prompt

Mission 1 Prompt[Role: IT business analyst]
Analyze the pet-care-meeting-notes.md file I just downloaded and write the requirements doc pet-care-requirements.md in this format.

1. Overview and service structure
2. User roles (owner, pet sitter, admin)
3. Key functional requirements list (ID format: PCR-001, PCR-002...)
4. Normal and exception flows (e.g. expiring after 1 hour without payment, 200% point compensation and a replacement pet sitter suggestion when a pet sitter no-shows, etc.)
5. Open questions list

Spec it in as much detail as possible, at a buildable level, and also apply the AI lint check rules from step 5 and show the self-check results at the top.

🎯 Main assessment criteria

• Is the 'admin approval' condition after pet sitter registration clearly specified?
• Are deadline exceptions like 'cancel if not paid within 1 hour' after a match request included?
• Are the compensation policy and replacement suggestion flow for pet sitter no-shows properly written into the plan?

Lab Assignment 2 Self-Service Study Café Booking and IoT Control Service

📥 Assignment input

  • • study-cafe-meeting-notes.md meeting notes
  • • The Codex prompts you learned in the step 1–5 labs

📤 Assignment output

  • • study-cafe-requirements.md (combined requirements doc with IoT exception handling)
  • • Hardware-software control test scenario spec

💡 Mission background and goal

These are meeting notes for a service at a self-service study café: booking study rooms on mobile and remotely controlling offline hardware (door locks, smart power cut-off, thermostats, etc.). You must build a complete IoT business spec by working out several safety processes: generating a PIN code when payment completes, syncing IoT device states and PIN limits at check-in/out, and emergency door opening and alarm alerts when IoT communication fails (e.g. door lock not working).

🛠️ Self-guided lab

  1. Download the notes: click the button below to download the original study-cafe-meeting-notes.md file.
  2. Write requirements: based on the downloaded notes, tell Codex AI to spec the user and hardware flows (activate 10 minutes before booking, notice 5 minutes before end, cut power at end).
  3. Detail failure handling: clearly write how to handle each possible scenario like IoT connection loss or a faulty door lock (server backup network, emergency alert to admin on send failure, emergency call, etc.) using fail-safe design.

Learner self-lab materials

Press the button below to download the raw meeting notes for Mission 2.

Recommended AI lab prompt

Mission 2 Prompt[Role: IoT systems business analyst]
Based on the downloaded study-cafe-meeting-notes.md, write study-cafe-requirements.md including the following.

1. Normal flow for study room booking and payment
2. Hardware control timing (activate PIN and turn on air conditioning 10 minutes before booking, audio notice 5 minutes before end, cut off immediately at end)
3. Exception and failure recovery flows (Fail-safe):
   - When the IoT door lock network goes offline, switch automatically to a backup wireless network and an emergency remote-open process
   - When the PIN code fails to send, provide an emergency manual open button and rules for an instant admin dashboard alarm
4. Requirement checks

Organize the flow in great detail as a timeline so developers and HW engineers can connect things with no confusion.

🎯 Main assessment criteria

• Are timeline conditions for hardware control (door lock, lights, air conditioning) before/after booking clearly stated?
• Does the requirements doc cover a two-layer emergency procedure for physical failures like an IoT device losing network?
• Are booking business rules like non-member booking limits and extension payment policy written as standard sentences?

Skill Assignment 1 (Mission 3) Short-Form VOD & Live Commerce Platform Domain Design

📥 Assignment input

  • • shortform-meeting-notes.md meeting notes
  • • Skill Tasks 1~3 (Persona, Domain, Bounded Context prompts)

📤 Assignment output

  • • shortform-domain-specification.md
  • • Short-form/live commerce Bounded Context doc

💡 Mission background and goal

These are meeting notes for a platform that tightly combines global short-form streaming and live commerce. Define the goals and actions of the 4 user types Viewer·Creator·Advertiser·Admin, place Members·Video·Playback·Engagement·Ad Serving·Billing features in owning Domains, then draft Bounded Contexts.

🛠️ Self-guided lab

  1. Apply Skill Task 1: work out the goals, actions and needed data of the 4 Personas Viewer·Creator·Advertiser·Admin.
  2. Apply Skill Task 2: split short-form VOD upload, streaming, targeted ads and payment features into 9 main owning Domains.
  3. Apply Skill Task 3: visualize data ownership and dependencies between Domains and submit a Bounded Context spec.

Recommended AI lab prompt

Mission 3 Prompt[Role: Domain Analyst & Solution Architect]
For a global short-form streaming and live commerce service, write shortform-domain-specification.md with these 3 sections.

1. Viewer·Creator·Advertiser·Admin Personas and key needed data
2. Owning Domain table for Members, Video, Playback, Engagement, Advertising, Ad Serving, Billing and Analytics features
3. Visualized dependencies between Domains, a Bounded Context draft and a check Checklist

🎯 Main assessment criteria

• Are the goals and data of the 4 user types (Viewer/Creator/Advertiser/Admin) separated?
• Are playback and ad owning Domains for VOD and Live Streaming clearly mapped?
• Are Bounded Context boundaries and reasons for judging AI results written?

Skill Assignment 2 (Mission 4) Smart B2B Mobility & Vehicle Control Platform Domain Design

📥 Assignment input

  • • fleet-meeting-notes.md meeting notes
  • • Skill Tasks 1~3 (Persona, Domain, Bounded Context prompts)

📤 Assignment output

  • • fleet-domain-specification.md
  • • B2B vehicle control Bounded Context & Responsibility Map

💡 Mission background and goal

These are meeting notes for a system covering company vehicle operations, real-time driver monitoring and commerce linked to fuel/maintenance. Carefully structure the Personas of a complex setup — company (corporate Admin), driver (Driver) and maintenance partner (Vendor) — and the owning Domains for vehicle data monitoring and payment.

🛠️ Self-guided lab

  1. Apply Skill Task 1: define the goals and needed data of the corporate admin, driver, partner garage and control Operator.
  2. Apply Skill Task 2: place real-time GPS monitoring, automatic trip logs, maintenance commerce and corporate billing features in owning Domains.
  3. Apply Skill Task 3: spec data ownership and a Bounded Context draft for connecting external maintenance/fuel network APIs.

Recommended AI lab prompt

Mission 4 Prompt[Role: B2B Enterprise Domain Analyst]
Based on fleet-meeting-notes.md for a company-vehicle mobility control platform, write fleet-domain-specification.md.

1. Personas and goals/needed data for Corporate Admin, Driver, Vendor and Dispatcher
2. Owning Domain placement for GPS Telematics, Fleet Management, Maintenance, Billing & Expense
3. Bounded Contexts visualizing data ownership and dependencies between Domains

🎯 Main assessment criteria

• Is data ownership separated between the B2B company organization and telematics (GPS monitoring)?
• Are the external fuel/maintenance commerce area and the corporate billing Domain boundary separated?
• Are Bounded Context dependencies and AI suggestions changed/rejected included?