Product managers spend a surprising share of the week turning one kind of text into another: interview notes into insights, insights into specs, specs into updates for four different audiences. That translation work is where ChatGPT earns its place.
The judgement calls, what to build and what to kill, stay with you. Here is how to split the work.
Specs and PRDs
Turn this problem statement and these notes into a first-draft
PRD: problem, users, goals and non-goals, requirements, open
questions, success metrics. Mark anything you are guessing with
[ASSUMPTION]. [notes]Review this spec as an engineer would. List ambiguous
requirements, missing edge cases, and anything that will cause
rework. [spec]The review prompt is often worth more than the drafting one. It catches the questions engineering will ask in refinement.
User research synthesis
Here are notes from [number] interviews. Group recurring themes,
quote the strongest evidence for each (with interview number),
and flag themes that appear only once. Do not invent quotes.
[notes]Strip names and anything identifying before pasting. Use initials or participant numbers.
Prioritisation
Score these ten items on impact, effort and confidence using the
evidence I give. Show the reasoning per item in one line, then
rank them. Where evidence is thin, say so. [items and evidence]Treat the output as a starting point for discussion, not a verdict. Its value is making your assumptions visible.
Stakeholder updates
Turn this week's notes into three updates: one for engineering
(detail, blockers), one for leadership (outcomes, risks, asks),
and one for sales (what is coming and when). Each under 150
words. [notes]Other jobs it handles well
- Writing release notes from a list of tickets.
- Drafting survey questions without leading wording.
- Turning a competitor's public changelog into a comparison table.
- Preparing for a tough roadmap review by arguing the other side.
Best AI tools for product managers
A general assistant covers most of the above. Beyond that, the useful additions tend to be whatever is built into tools you already use: AI features in your issue tracker, docs and analytics. A new standalone tool is rarely worth another tab unless it replaces one.
What to keep human
- Deciding what not to build.
- Reading the room in a stakeholder conversation.
- Anything where the answer depends on context that is not written down.
Questions people ask
Can ChatGPT write a PRD?
It can write a solid first draft from your notes. It cannot know your constraints unless you tell it, so expect to edit heavily.
Is it safe to paste product plans into ChatGPT?
Only in a tool your company has approved, with a work account. Unannounced plans are usually confidential.
Should I use a project for each product?
Yes. Attach the strategy doc and one good spec as examples so every chat starts with context.
Where to take this next
For a ready-made set, see the free product manager prompt pack. For an assistant set up around your product and stakeholders, The Pack builds the instructions and task files from your profile.
Laisser un commentaire :