BLOG
[3/4] Our Approach to Project Management: Relentless Efficiency

- The Work Nobody Wants to Do
- We Don't Want to Spend Our Time on Project Management
- Automating Task Management
- The Challenges We Had
- Two-Way Sync with Google Sheets
- Automating Task Creation
- 1. Automatic Title Generation and Translation
- 2. Automatic Dependency Management and Blocked-Status Updates
- 3. Automatic Subtask Generation for Implementation Tasks
- Automating Reporting
- Automatic Report Generation System
- Automatic Posting to Google Chat
- Concrete Results
- 1. Much Shorter Task Creation Time
- 2. A Dramatic Reduction in Mental Burden
- 3. Bulk Operations Are Now Possible
- 4. Better Overview of Information
- 5. Reporting Fully Automated
- The Essence of DX
In the previous article, we explained how we have systematized project management and how we teach it within the company. Through guidelines and training, we practice project management based on a reliable methodology.
But that alone is not enough. To run project management with high quality and efficiency, thorough automation is essential.
This time, we describe in concrete terms how we have achieved digital transformation (DX) of our project management.
The Work Nobody Wants to Do
One of the biggest challenges in project management is "the sheer amount of effort."
Traditional project management required work such as the following:
-
Creating tasks one by one in GitHub
-
Setting the title, description, assignee, estimate, start date and completion date individually for each task
-
Defining dependencies between tasks manually
-
Assigning tasks to sprints
-
Updating statuses
-
Creating daily, weekly and sprint reports
Creating a single task took about 30 seconds.
Creating 20 tasks in sprint planning took 10 minutes on that alone. On top of that, it meant clicking many buttons, moving through many screens, and filling in many fields.
For engineers and product managers alike, this was work nobody wanted to do. It was too tedious, too cumbersome, and too much of a mental burden.
As a result, project management got put off, task updates stalled, and progress became unclear. A vicious cycle had formed.
We Don't Want to Spend Our Time on Project Management
As we explained in the first article, we are not a management company.
What we really want to spend our time on is:
-
Increasing the value of the product
-
Responding to users' voices
-
Thinking about how to make the product better
We do not want to spend our time on project management and reporting.
At the same time, we do not want to neglect project management either. We want to run it with high quality and efficiency.
To do that, we committed to thorough automation.
Automating Task Management
The first thing we tackled was making task management more efficient.
The Challenges We Had
Task management in GitHub had the following challenges:
-
No bulk operations: setting assignees or sprints in bulk was difficult
-
Poor overview: it was hard to see many tasks at a glance
-
Not suited to external stakeholders: GitHub is an unfamiliar tool for product owners and external stakeholders
Two-Way Sync with Google Sheets
We built a system that automatically syncs GitHub Projects and Google Sheets in both directions.
With this system:
-
Bulk editing in a spreadsheet: easy operation with copy and paste
-
Better overview: many tasks can be seen at a glance
-
Anyone can use it: product owners and external stakeholders can view and edit in a tool they already know
-
Sync with GitHub: changes in the spreadsheet are automatically reflected in GitHub, and changes in GitHub are reflected in the spreadsheet
Task creation time went from 30 seconds to a few seconds.
Even more importantly, the mental burden dropped dramatically. There is no longer any need to click many buttons and move through many screens; tasks can be managed with simple operations in a spreadsheet.
Automating Task Creation
Making task management more efficient is still not enough. Creating a task requires entering a lot of information.
We automated much of the work involved in creating tasks.
1. Automatic Title Generation and Translation
We give tasks a hierarchical structure:
-
L1 (major category): e.g., Setup, Feature Development, Bug Fixes
-
L2 (mid-level category): e.g., Account Creation, User Management, Error Handling
-
L3 (minor category, specific content): e.g., Sentry alert notification settings, password reset feature
-
Task code: e.g., 0-1-3 (the L1-L2-L3 numbers)
When this information is entered in Japanese in the spreadsheet, the system automatically:
-
Generates an English title: translates the Japanese into English automatically
-
Formats it properly:
0-1-3: Setup - Account Creation - Sentry Alert Notification Settings -
Reflects it in GitHub: sets both the English title and the Japanese title (a custom field)
This lets members register tasks they think of in Japanese in GitHub right away, without the effort of translating them into English.
2. Automatic Dependency Management and Blocked-Status Updates
When tasks depend on each other, managing those dependencies is cumbersome work.
Our system:
-
Tracks dependencies automatically: when dependencies are defined in the spreadsheet, they are automatically reflected in the GitHub issue description
-
Detects blocked status automatically: if a task it depends on is not complete, the task is automatically recognized as blocked
-
Updates titles automatically: a 🚫 icon is automatically added to the title of a blocked task
-
Notifies assignees automatically: when a task becomes blocked or unblocked, the assignee is automatically notified with a GitHub comment
For example:
When blocked:
@username ⚠️ This task is now blocked This task cannot be started because the following dependencies are not yet completed: - ⏳ #145 - Setup - Account Creation - API Implementation
When unblocked:
@username ✅ This task is now unblocked! All dependencies have been completed. You can now start working on this task. 🚀 Ready to begin!
This means members no longer have to track dependencies by hand and can start tasks at the right time.
3. Automatic Subtask Generation for Implementation Tasks
Development tasks often need four subtasks: frontend implementation, frontend review, backend implementation and backend review.
When a task whose L3 starts with "Implementation:" is created, the system automatically:
-
Creates a parent task: with the
(parent-issue)suffix -
Generates four subtasks automatically: Frontend Implementation / Frontend Review / Backend Implementation / Backend Review
-
Sets dependencies between subtasks automatically: Frontend Review depends on Frontend Implementation, and so on
As a result, a single entry automatically creates five structured tasks.
Automating Reporting
Another major burden in project management is reporting.
Creating daily, weekly and full-sprint reports by hand takes a lot of time. On top of that, they need to be reviewed for accuracy, and sometimes they need to be multilingual.
Automatic Report Generation System
We built a system that automatically generates four types of reports.
1. Daily Report (every business day at 20:00 NPT)
-
Status of tasks that reached their due date (completed, in progress, not started)
-
Tasks due on the next business day
-
Overall sprint progress rate
-
Changes to the sprint (tasks added or removed)
2. Weekly Report (end of the week at 21:00 NPT)
-
Status of tasks due this week
-
Tasks due next week
-
Overall sprint progress rate
-
Cumulative sprint changes for the week
3. Sprint Report (sprint end date at 21:30 NPT)
-
Completion status of the whole sprint
-
Cumulative changes across the whole sprint period
4. Daily Scrum Summary (AI-generated)
-
AI summarizes the daily report plus the daily scrum meeting notes
-
Summary of sprint progress
-
Highlights of the daily scrum
-
Comments for the team
Automatic Posting to Google Chat
All reports are automatically:
-
Posted to the project space in Google Chat: in both Japanese and English versions
-
Saved to a database: available for later reference and analysis
-
Calculated on a business-day basis: weekends and public holidays are excluded for accurate calculation
As a result, no one has to create reports by hand anymore. Everyone involved can check automatically generated reports every day, every week and at the end of each sprint.
Concrete Results
We started a trial run of our automation initiative in October 2025.
Results so far:
1. Much Shorter Task Creation Time
-
Before: about 30 seconds to create one task
-
Now: a task can be created in a few seconds
-
Efficiency gain: about 10 times faster
2. A Dramatic Reduction in Mental Burden
More important than the time saved is the reduced mental burden.
-
Before: clicking many buttons, moving through many screens, filling in many fields
-
Now: simple operations such as copy and paste in a spreadsheet
Work that nobody wanted to do has become work people can do without a second thought.
3. Bulk Operations Are Now Possible
-
Setting assignees in bulk
-
Setting sprints in bulk
-
Copying and editing tasks in bulk
Operations that were difficult in GitHub are now easy in a spreadsheet.
4. Better Overview of Information
Many tasks can now be seen at a glance in a spreadsheet, which makes it easier to grasp the project as a whole.
5. Reporting Fully Automated
Creating reports by hand has disappeared completely, and that time can now be used for other things.
The Essence of DX
Our DX of project management is not just about "efficiency."
Its essence is making it possible to spend our time on what really matters.
By no longer having our time taken up by project management and reporting, everyone on the team can:
-
Focus on increasing the value of the product
-
Spend more time responding to users' voices, requests and frustrations
-
Spend time improving the product
-
Have discussions about building a better product
This is exactly what we are aiming for.
We are not a management company. We are a product co-creation partner. That is exactly why we run project management with high quality and efficiency, so that we can spend our time on what really matters.