AI Writing
Why I Built an AI Prompt Library for Scope of Work Documents
Aug
For many of us, especially those who began our project management careers before Agile, Scrum, and other frameworks became mainstream, I feel there can be a reluctance to adopt ‘shiny new tools.’ This is particularly true for using AI to write PM docs like the Scope of Work.
After all, this is something we should know how to do ourselves, without needing to rely on AI for assistance. Or is it?
So, for me, I was hesitant to hand over any pure writing tasks to AI. I was comfortable using it to edit, and perhaps for proofreading, but actual bare-bones writing? Not sure.
However, one of my colleagues began creating a ‘prompt library’ for various AI-driven tasks. Some were designed for brainstorming, others for ‘ideation’. She was also using ChatGPT to identify blind spots and gaps in her documentation. Looking at the quality of the documents she produced, I could see the immediate benefit.
I decided to see how I could apply this approach, starting with a small project.
Last week, I decided to use Claude as my dedicated research and drafting assistant. After some trial and error, I created a suite of prompts, ranging from simple to complex.
I’ve now compiled these prompts into an Excel spreadsheet and added it to our Scope of Work template pack. I hope you find it useful.
The following section provides practical ideas on how you can use these prompts to research, write, and review your own SOW documents.
Download: Scope of Work Templates (MS Word + Prompt Library)
Getting Started
As a project manager, I’ve written tons of Scope of Work (SOW) documents.
The process is time-consuming and has several challenges:
- Inconsistency: My SOWs, and those of my colleagues, often varied in quality and depth depending on the project’s complexity or the time available.
- Inefficiency: I found myself constantly “reinventing the wheel,” trying to recall the best phrasing for a constraints section or searching old documents for a good risk register template.
- Incompleteness: It was far too easy to overlook a critical component, like procurement management or a detailed communication plan, especially under tight deadlines.
Here’s the thing: while the structure of an SOW is standardized, the thinking required for each section is not. I needed a system to refine the writing process while giving me time to work on the strategic more thought side of things. As mentioned above, this led me to develop a Scope of Work Prompt Library, a structured set of prompts designed to leverage AI as a highly skilled co-author.
This tutorial explains why I built it and, more importantly, how you, as a Project Manager, can use it to transform your SOW writing process.
Scope of Work prompt library – Excel spreadsheet
Benefits
The main benefit of the library is that it frees up time spent on low-level writing/drafting tasks. Instead of spending hours crafting paragraphs from scratch, I can get working drafts quickly so I can spent more time validating the content.
The library achieves this through a specific prompt structure:
- Persona: Defines the AI’s voice and expertise (e.g., “Act as a Senior Project Manager”).
- Task: Gives a clear, actionable command (e.g., “Draft a detailed project scope statement”).
- Context: Grounds the AI with real-world project details.
- Format: Specifies the desired output (e.g., “A formal paragraph followed by ‘Includes’ and ‘Excludes’ tables”).
By layering this structure across three tiers (Simple, Advanced, and Complex) I can modify the output to match the project’s needs.
Step-by-Step Tutorial
Imagine you need to write the “Project Scope” section (3.1) of your SOW. Here’s how you would use the library.
1: Select Your SOW Section
Navigate to the relevant section in your SOW template. For this example, we’re working on 3.1 Project Scope.
2: Choose Your Prompt Tier
Assess your project’s requirements to select the appropriate level of detail.
- Simple: Ideal for a quick internal draft, a small-scale project, or when you just need to get initial thoughts on paper.
- Advanced: The standard choice for most formal client-facing or cross-departmental projects. It ensures all necessary components are included professionally.
- Complex: Reserved for high-stakes, strategically critical, or exceptionally intricate projects where every boundary condition must be meticulously defined.
3: Customize and Execute the Prompt
Copy the chosen prompt from the library and customize the context with your specific project details. The more specific your context, the better the AI’s output.
4: Review, Refine, and Integrate
The AI-generated text is your first draft, not your last. Your job as the Project Manager is to review it for accuracy, add nuanced details only you would know, and ensure it perfectly aligns with stakeholder expectations. Then, integrate it into your SOW document.
Practical Examples
Here are three real-world examples demonstrating how to use the Simple, Advanced, and Complex prompts for different SOW sections.
Example 1: Defining Project Risks (Section 3.7)
Scenario: You are managing a software upgrade project for a critical financial system.
Simple Prompt Application:
Persona: Project Manager. Task: Identify the top three risks for the project. Context: My project involves a critical financial system software upgrade. Major risks include key staff leaving and unexpected technical challenges with data migration. Format: A simple three-row table.
Result: A quick, high-level list of risks perfect for an initial team discussion.
Advanced Prompt Application:
Persona: Risk Manager. Task: Create a risk register. Context: For the financial system upgrade, identify risks like ‘Data Migration Errors,’ ‘Vendor Delays,’ and ‘User Adoption Resistance.’ Assess severity and impact. Format: A table with columns for Description, Severity (H/M/L), and Impact (Cost/Time/Quality).
Result: A formal risk register that can be included in the SOW and used for ongoing monitoring.
Complex Prompt Application:
Persona: Senior Risk Analyst. Task: Develop a detailed risk response plan. Context: For our financial system upgrade, detail the response for the highest-priority risks. For ‘Data Migration Errors,’ a mitigation action is to perform three test runs. A contingency plan is to have a data-cleansing team on standby. Format: A table with columns for Risk, Severity, Impact, Mitigation Actions, and Contingency Plan.
Result: A strategic, actionable plan that demonstrates foresight to stakeholders and gives the project team clear direction.
Example 2: Delineating Project Deliverables (Section 3.6)
Scenario: You are launching a new corporate website.
Simple Prompt Application:
Persona: Project Manager. Task: List the major project deliverables. Context: The project is to launch a new corporate website. Format: A numbered list.
Result: A straightforward list like “1. Website Design Mockups, 2. Fully Functional Website, 3. User Training Guide.”
Advanced Prompt Application:
Persona: Project Manager. Task: Create a list of project deliverables with descriptions. Context: For the new corporate website, describe deliverables like ‘Website Design Mockups’ and how they relate to the goal of a modern user interface. Format: A table with columns for No., Deliverable, and Description.
Result: A more informative list that adds context for stakeholders, explaining the purpose of each deliverable.
Complex Prompt Application:
Persona: Product Owner. Task: Generate a comprehensive Deliverables Log. Context: For the new website, define the acceptance criteria for the ‘Fully Functional Website’ deliverable. Criteria include: ‘Pages load in under 2 seconds,’ ‘Passes WCAG 2.1 AA accessibility standards,’ and ‘All contact forms successfully send emails to the sales team.’ Format: A detailed table including Deliverable, Description, Acceptance Criteria, and Approval Authority.
Result: An unambiguous log that forms the basis of user acceptance testing and official project sign-off, leaving no room for dispute.
Example 3: Establishing the Meeting Cadence (Section 5.2)
Scenario: You are managing a 6-month project with a distributed team and executive sponsors.
Simple Prompt Application:
Persona: Project Coordinator. Task: Propose a weekly meeting schedule. Context: The team is remote. We need a team check-in and a stakeholder update. Format: A simple list of meetings, their day, and purpose.
Result: A basic schedule: “Team Meeting: Mondays, Stakeholder Update: Fridays.”
Advanced Prompt Application:
Persona: Project Manager. Task: Define the meeting rhythm for the project. Context: My project requires a daily stand-up for the dev team, a weekly tactical meeting with team leads, and a monthly steering committee review with sponsors. Format: A table with columns for Meeting, Purpose, and Frequency.
Result: A structured plan that clarifies the purpose and frequency of each interaction type.
Complex Prompt Application:
Persona: Communications Manager. Task: Design a comprehensive meeting and governance plan. Context: For this 6-month project, create a matrix for all recurring meetings. The ‘Steering Committee’ meeting’s purpose is ‘Strategic review and budget approval.’ Required attendees are ‘Project Sponsor, Program Director, PM.’ Inputs are the ‘Monthly Status Report.’ Outputs are ‘Go/No-Go decisions on change requests.’ Format: A detailed meeting matrix.
Result: A complete governance framework that ensures the right people are in the right meeting with the right information to make the right decisions.
Download: Scope of Work Templates (MS Word + Prompt Library)
Lessons Learned
I need to rethink how to use AI more strategically. I’ve also used it to create some macros that I can plug into Excel, which has been super helpful. I’ve always struggled to macros to work using the VB editor. More on that next week. For me, developing a suite of prompt templates has offered several benefits:
- Efficiency: Drafting a SOW has gone from a multi-day effort to a matter of hours.
- Consistency: Every SOW draft I produce feels more complete. Nothing gets missed, such as niche compliance requirements.
- Focus: By automating mundane writing tasks, I get enough distance from the doc to see where it can be improved.
Have you started to create your own Prompt Library? Interested to know how you’re using it.
Connect with us
