iTranslated by AI

The content below is an AI-generated translation. This is an experimental feature, and may contain errors. View original article
👨‍💼

[PMBOK-Compliant] Practical AI-Driven Project Management: A Starter Kit for Cursor or Codex (Free)

に公開
AI
PM
Cursor
Codex
ai駆動pm
tech

Realizing AI-driven PM (Project Management)—this is a challenge that will dramatically change the work of project managers.

In traditional project management, a significant amount of time has been spent on routine tasks such as document creation, scheduling, and risk detection. By leveraging AI for these tasks, work hours can be made significantly more efficient, allowing PMs to devote more time to what they should naturally focus on, such as "decision-making" and "stakeholder coordination."

In this article, we will explain in detail, through a hands-on format, practical methods for achieving PMBOK-compliant AI-driven PM using Cursor or Codex. We will comprehensively share necessary information as an introductory guide, from the basic concepts of AI-driven PM to specific environment setup procedures and real-world application examples.

Note: In this article, PM refers to Project Management (PjM, not PdM).

What is AI-driven PM (Project Management)?

AI-driven PM (AI-driven Project Management) is a methodology that utilizes AI to support PM tasks, improve operational efficiency, and drive projects efficiently with high quality.

PMs receive consistent support from AI, ranging from project planning to execution, monitoring, control, and closure.

By making document generation and risk detection more efficient, it creates an environment where PMs can focus on tasks that only humans can do, such as decision-making and stakeholder coordination.

Hands-on practice in this article

By following this article to the end, you will be able to:

  1. Create requirement definition documents based on meeting minutes
  2. Generate risk registers
  3. Generate regular meeting agendas
  4. PMBOK-compliant project driving assistance

Furthermore, by mastering this methodology, you will be able to automatically generate various documents beyond those mentioned above (e.g., project charters, stakeholder registers, quality management plans).

Setting up the AI-driven PM Environment with Cursor

Here, I will explain using Cursor as an example. Since the same setup can be achieved with Codex, please adapt the instructions as necessary.

As some of you may know, Cursor is basically an AI coding assistant that improves development efficiency with AI.

In other words, while it is primarily a tool for IT engineers, use cases have recently emerged in various fields such as corporate planning, marketing, and writing. Since it can also be utilized for project management, this article explains the practical methods in detail.

The setup flow consists of the following 3 steps:

  1. Install Cursor
  2. Turn ON "Privacy Mode" in Settings (to prevent training on your data)
  3. Place the starter folder on your local PC

I will explain them in order.

① Install Cursor

Download and install it from the official Cursor website.

👉 https://cursor.com/ja

You are all set once you see this screen after installation.

Cursor Installation Screen

Click the Pane icon in the top right to display the AI chat, file explorer, and terminal.

Cursor Pane

② Settings → Turn ON "Privacy Mode" (to prevent training on your data)

From the menu, go to "Cursor" -> "Settings" -> "Cursor Settings" and type "privacy mode" in the search box.

Privacy Mode Settings

Select "Privacy Mode" and toggle the switch to ON.

When this setting is enabled, Cursor will no longer send your prompt content as training data for the AI models.

In other words, this is a privacy protection setting to prevent your project information and confidential data from being used for AI training.

It is highly recommended to always keep this mode ON when handling sensitive code, such as for corporate use or client projects.

③ Place the starter folder on your local PC

Access the following GitHub repository.

👉 https://github.com/MasahiroOkamura-MAC/AI-driven-Project-Management/tree/main

For those who can use Git: Clone it or download it to your local machine.
For those without a Git environment: Download the folder via "Download ZIP".

Starter Folder Download

Extract the Zip file and select "Open Folder" from Cursor.

Open Folder

It is correct if the ".cursor" folder, "AGENTS.md", and "ECサイト開発案件" (EC site development project) folder are displayed.

Structure Diagram

Confirm that the commands are displayed by typing a slash "/" in the AI chat.

Slash Command

Now you are ready for AI-driven PM!

Hands-on: Actually running AI-driven PM

Now that the environment is set up, it's finally time to put it into practice.

AI-driven PM operates by executing prompts within Cursor's AI chat.

① Create a Requirement Definition Document based on meeting minutes

Type "/" in the AI chat field and select the command to create a requirement definition document.

Requirement Definition Document Command

Next, drag and drop the sample "議事録.txt" (meeting_minutes.txt) into the AI chat.

Inputting meeting minutes

Input it in the following format:

/aipjm/02_planning/07_要件定義書
Generate based on @議事録.txt

Generating Requirement Definition Document

Once the AI finishes processing, "要件定義書.md" (requirement_definition.md) will be generated inside the docs folder.

Output result

② Generate a Risk Register

Similarly, you can generate a risk register.

Risk Register Command

Specify the requirement definition document and the meeting minutes as context.

Generating Risk Register

Once the AI completes the task, the risk register will be generated.

Risk Register Output

③ Generate Regular Meeting Agenda

Set the prompt as follows, just like with the requirement definition document.

MTG Agenda

Once the AI finishes processing, the agenda is completed.

Agenda Completed

④ PMBOK-Compliant Project Driving Assistance

As a PM, you likely encounter troubles frequently.
In such cases, you can consult without hesitation in AI-driven PM.

  • The delivery date is delayed. How should I respond?
  • There is a gap in understanding the development scope with the client. How should I respond?

By attaching meeting minutes or describing the situation in detail as needed, you can receive higher-resolution advice.

[Column] It is Not Always Necessary to be PMBOK-Compliant

In real projects, flexible responses are required.

In some cases, it may be more effective to modify Command files or AGENTS.md, or to generate using a prompt like the following:

@Meeting Minutes This is the previous meeting's minutes. Generate an agenda for tomorrow's regular meeting in the following format:
{Describe your company's optimized meeting minutes template here}

[Summary] Collaboration Between AI and PMs Will Become the "New Standard"

AI-driven PM is not currently intended to replace all PM tasks.
It is essential for humans to verify the deliverables generated by AI.
Nevertheless, having AI generate drafts can significantly improve the efficiency of PM work.

[Next Time] Adding Context via MCP

The most important thing for AI to make correct judgments and proposals is "context."

By integrating with tools that support MCP (Model Context Protocol) servers (such as Google Drive, Notion, Jira, Asana, etc.), even more accurate AI-driven PM becomes possible.

[Promotion]

MAKE A CHANGE Co., Ltd. is promoting AI-driven PM (Project Management).
We can assist with the introduction of AI-driven PM using existing tools (Cursor, Codex, Notion AI, etc.) as well as the development of original AI tools.

👉 https://makeachange.co.jp/

Company Logo

@masa_oka108 (X)

#AI #PM #Cursor #Codex #ProjectManagement #AIDrivenPM

Discussion