Using AI Tools to Write Better Project Documentation
The project manager who refuses to use AI writing tools in 2023 is in roughly the same position as the project manager who refused to use spreadsheets in 1990. The tools are here, they work, and the professionals who learn to use them effectively will have a meaningful productivity advantage over those who do not. That said, using AI tools badly is a real risk — one that can introduce errors, erode credibility, and create exactly the kind of rework they were meant to prevent.
Understanding where AI writing assistants add genuine value for project documentation — and where they fall short — is the foundation of using them well.
What AI Tools Can Do for PMs
The most immediate value of tools like ChatGPT, Claude, and Microsoft Copilot for project managers is speed on first drafts. A project charter that would take two hours to produce from a blank page can be roughed out in ten minutes with a well-constructed prompt. Status reports, risk registers, lessons-learned documents, RACI matrices, meeting summaries, stakeholder communication plans — all of these benefit from a fast, structurally competent first draft that the PM can then edit for accuracy and context.
AI tools are also useful for adapting the same underlying content for different audiences. The same project update, rewritten for a technical team versus an executive steering committee, requires a significant shift in language, level of detail, and emphasis. An AI assistant can produce both versions quickly once it has the core facts.
Finally, AI tools can help overcome the blank-page problem that afflicts even experienced writers. Having a draft — even an imperfect one — to react to is cognitively much easier than constructing a document from nothing. Many PMs find that editing an AI-generated draft is both faster and less mentally taxing than writing from scratch.
How to Use Them Effectively
The quality of AI output is highly sensitive to the quality of the input. A vague prompt produces a vague document. A well-specified prompt produces something much more useful.
Providing specific context is the most important habit to develop. The AI does not know your project. It does not know your organisation's culture, the political sensitivities around a particular deliverable, the history of the project, or the specific format your governance framework requires. Every relevant piece of context you include in the prompt improves the output. Project type, industry sector, audience, purpose, format, tone, length — all of these are worth specifying explicitly.
Being explicit about format helps too. Do you want a bulleted executive summary or a narrative paragraph? A formal risk register table or a plain-language risk description? Structured document or conversational briefing note? The more precisely you describe what you want, the closer the first draft will be to what you need.
Reviewing and correcting factual errors is non-negotiable. AI tools will occasionally state things that sound authoritative but are wrong — fabricated statistics, misattributed quotes, incorrect technical details, plausible-sounding regulatory requirements that do not exist. The PM is always the last line of defence. Every factual claim in an AI-generated document needs to be verified before the document goes to a stakeholder.
What to Be Careful About
Sensitive project information is the most significant risk. Most AI tools send prompts to external servers. Pasting your project's financial data, personnel information, or confidential client details into a public AI tool creates a data governance problem that is entirely separate from the quality of the output. Many organisations have policies (or should have them) about what information may be shared with external AI services. Understanding and following those policies is the PM's responsibility.
Over-reliance is a more subtle risk. Writing is thinking. The process of drafting a lessons-learned document or a risk register forces the PM to think through the project carefully, surface insights that might otherwise stay implicit, and make connections between events and causes. If the AI does all of this and the PM just approves the output without genuine engagement, some of the cognitive value of the exercise is lost. The goal is to use AI to reduce the mechanical burden of writing, not to outsource the thinking.
Hallucinated facts that sound plausible are the most practically dangerous failure mode. Unlike obvious errors that prompt review, a confident-sounding but incorrect statement can pass through review undetected. The risk is highest in technical domains, regulated industries, and any document that includes data or calculations. If you cannot verify a specific claim in the AI-generated draft, remove it or replace it with something you can verify.
What Changes for PMs
AI writing tools do not make project management easier in the sense of requiring less skill. They change what skills matter most. Prompt construction — knowing how to specify what you want clearly and completely — becomes a practical professional skill. Critical review — the ability to identify what is wrong or missing in a plausible-sounding document — becomes more important, not less. And judgment about what information should and should not go into an AI tool becomes a routine professional responsibility.
The PMs who will benefit most from these tools are the ones who use them to do more of what matters — better stakeholder communication, more time thinking about risk, deeper engagement with the team — not the ones who use them to do less thinking.
XNM Consulting integrates practical PM tools and techniques into every engagement. For support on building project management capability in your organisation, visit our Program and Project Delivery page.