Before we get into use cases, three non-negotiable rules. If you skip these and dive straight into the tools, you will eventually publish something embarrassing, misleading, or both.
Zero-Trust Fact-Checking. AI is a vibe engine, not a database. It will confidently generate statistics that don’t exist, attribute quotes to people who never said them, and describe competitors based on information that’s 18 months out of date. Never publish a technical claim or a stat without manual verification. The PMM is accountable for what goes out under the brand. Always.
Context Over Keywords. AI can write a resume, but it doesn’t know your company’s philosophy, your specific market position, or your “opinionated” take on the category. The more context you inject across your messaging framework, your ICP docs, your brand voice, the more useful the output. Garbage in, garbage out applies here more than anywhere.
Data Sovereignty. Unless you’re on an enterprise instance with data agreements in place, assume that what you type could eventually inform future model training. Don’t input unreleased product roadmap details, confidential customer data, or anything you wouldn’t want visible outside the company.
A few use cases that consistently deliver real productivity gains for PMMs:
Idea generation and iteration. First drafts, 10 variations of a headline, tone shifts, perspective changes. Use AI to generate the options; use your judgment to select and refine.
Notebooks as briefs. Feed an AI 50 pages of customer research, win/loss reports, and competitive intelligence, and ask it to generate persona summaries or messaging hypotheses. The synthesis speed is genuinely transformative.
Content transformation. A long-form whitepaper becomes a blog series. A slide deck becomes a sales email sequence. A transcript becomes a structured summary. One piece of content, many outputs.
Persona stress-testing. Load your ICP and persona docs, then ask the AI to interview your messaging as a skeptical enterprise buyer. It will find the gaps you’ve stopped seeing.
Competitive battlecard drafts. Feed it two sets of positioning and ask it to highlight the key differences, weaknesses to exploit, and objections to prepare for. Saves hours.
Data leakage. Inadvertent sharing of unreleased features or roadmap details is a real risk. Establish team norms around what can and can’t go into AI tools before you build workflows that rely on them.
The source of truth problem. AI is a great synthesiser and a terrible fact-checker. The output reads confidently regardless of whether it’s accurate. You are the one accountable for verifying it.
AI carefully adopted, with appropriate guardrails, is a genuine force multiplier for a PMM team. Adopted carelessly, it produces polished-looking work that erodes credibility the moment a knowledgeable audience reads it.
Full post with complete use case list: Using AI at Work →