AI Tools · 8 min read
ChatGPT saves PMs the most time on four workflows: synthesising user research (turns 50 interview notes into themes in minutes), writing first-draft PRDs (cuts writing time by 60-70%), generating competitive analysis frameworks (structures messy competitor info fast), and creating stakeholder-ready summaries from technical documentation. Where ChatGPT consistently underperforms: anything requiring your specific product context, real usage data, or knowledge of your users beyond what you paste in.
This is ChatGPT's highest-value use case for PMs. If you have 20-50 user interview transcripts or support ticket exports, ChatGPT can identify recurring themes, pain points, and user mental models significantly faster than manual analysis — and without the cognitive fatigue of reading everything twice.
The prompt that works: Paste your research notes and say: "You are a UX researcher. These are notes from [N] user interviews about [product area]. Identify the 5-7 most common pain points, group supporting quotes for each, and suggest one product hypothesis for each pain point. Format as a structured report."
The result won't be perfect — ChatGPT can miss nuance and occasionally misgroup themes — but it gives you a strong foundation to edit and refine rather than starting from a blank page. The editing pass takes 20 minutes versus 3 hours of manual synthesis.
Writing a PRD from scratch is one of the most time-consuming parts of a PM's week. ChatGPT won't write your PRD for you — it doesn't know your users, your constraints, or your technical architecture. But it will write a structurally sound first draft in your preferred format that you can fill in with specifics.
The prompt that works: "Write a PRD for the following feature. Use this structure: Problem Statement, User Stories, Acceptance Criteria, Success Metrics, Out of Scope. Feature: [one-paragraph description]. Target user: [description]. Key constraint: [main limitation]. Write it in clear, non-technical language suitable for a startup audience."
The output will have all the right sections with placeholder-quality content. Your job is to replace the generic sections with your specific knowledge. This approach typically cuts PRD writing time from 3-4 hours to 45-60 minutes.
You have tabs open with 6 competitor websites, pricing pages, and G2 reviews. Turning this into a useful competitive matrix takes hours. ChatGPT can structure this fast. Paste competitor information (product descriptions, feature lists, pricing) and ask: "Create a comparison table for these competitors evaluating them on: pricing model, key features, target customer, weaknesses. Then write a 2-paragraph summary of where the market is underserved."
The table format is particularly useful for stakeholder communication — it's far more readable than prose analysis and easier to update when competitors change.
Engineering hands you a 20-page technical specification. You need to understand the implications for the product roadmap and communicate them to non-technical stakeholders. Paste the doc (or key sections) and ask: "Summarise this technical document for a non-technical product audience. Focus on: what this enables that wasn't possible before, what the main constraints or limitations are, and what this means for the user experience. Keep it under 300 words."
Prioritisation decisions: ChatGPT cannot tell you which features to build next. It doesn't know your users' actual behaviour, your company's strategic priorities, or the relative engineering costs. Prompts like "help me prioritise my roadmap" produce generic frameworks, not useful prioritisation. Data analysis: ChatGPT cannot analyse your actual product data — it can only help you think about how to interpret numbers you provide. For actual analysis, use Mixpanel, Amplitude, or a data team. Anything requiring your company's internal context: If the answer depends on knowledge that lives inside your company (customer contracts, architecture decisions, team dynamics), ChatGPT will hallucinate or produce irrelevant output.
Both are excellent for PM workflows, but they have different strengths. ChatGPT (GPT-4o) tends to produce more structured, list-heavy output — good for PRDs, feature specs, and competitive tables. Claude tends to produce more nuanced, contextually aware prose — better for user-facing copy, strategic memos, and anything requiring judgment calls. For a PM toolkit, using both is better than choosing one. The practical workflow: use ChatGPT for structural tasks (PRD structure, feature matrices), use Claude for narrative tasks (executive summaries, user communication, strategy documents).
ChatGPT's Custom GPTs allow you to create a pre-configured assistant with your company's context loaded in. Build a custom GPT with your product's user personas, your tech stack details, your PRD template, and your writing style guide. This eliminates the repetitive context-setting in every prompt. A well-built custom GPT for your product can produce output that requires 50% less editing than a generic ChatGPT prompt.
What to include in your PM Custom GPT: your standard PRD template, 3-5 example user personas (anonymised), your product's core value proposition, your company's writing style guide (formal vs conversational, preferred terminology), and any domain-specific terminology your product uses.
For identifiable user data (names, emails, specific account details), no — this creates privacy and data protection risks. Anonymise user research before pasting. For internal product documentation, check your company's data policy; many companies have restrictions on pasting proprietary information into external AI tools. ChatGPT Enterprise (which doesn't use your data for training) is the appropriate option if you regularly work with sensitive product data.
We help product teams build ChatGPT and Claude workflows that save 4+ hours per PM per week. Book a free session.
Book Free Strategy Call