Using GenAI to slow down, and then speed up
Used with intention, GenAI moves your hours from producing to thinking. The thinking gets more of them. The building gets fewer.
The thinking gets more of your time. The building gets less.
My first website built with Claude Code began as one long conversation, back and forth with Claude until the site worked.
When it was done, I asked Claude to read the conversation and write down the method behind it: the steps and the decisions. Then I built five more websites with that document as the starting point.
The long conversation was where the thinking happened. The document holds what that thinking decided. The five builds that followed started from it.
Now think about the last thing you handed to AI. How many of your minutes went to deciding what to ask for?
The Principle: Where the Time Goes
The website story has the shape this issue is about: the thinking came first, and the building drew on it. Used with intention, GenAI moves your hours from producing to thinking. The thinking gets more of them. The building gets fewer.
The word that comes to mind is pace, and pace measures how fast one activity runs. The measure worth watching is where the hours land.
In the Intentional Intelligence framework, three modes shape that split. Mirror reflects your assumptions back to you. Telescope shows you what you cannot see from where you stand. Hammer builds. Mirror and Telescope take the thinking hours. Hammer takes the fast ones.
Most PMs spend their hours on production. Strategic PMs spend more of them deciding what to produce.
Why This Matters
The decisions are what you keep. Claude can draft a page or a plan on request, so the drafting is the part that repeats easily. What took the long conversation was deciding how the site should be built, and writing those decisions down is what let five more sites start from them. The hours worth protecting are the ones where you decide what to make.
The thinking step is already in the job. One Project Manager posting from DYOPATH, October 2026, includes this line: "Translate business objectives into actionable project plans." That is one employer describing one job, and translating objectives into a plan is thinking work.
Mirror is where the move starts. Mirror is the mode that begins with your unfinished thinking: a plan you are unsure about, a set of assumptions you have not tested. You hand AI the work in progress and ask what it sees. That handover is where the thinking hours begin, and the prompt below is how you make it.
The AI Advantage
For the thinking: Mirror finds the assumptions you stopped seeing, and Telescope brings in the views you cannot see from your chair. These sessions take their share of the hours, and their output is questions.
For the communicating: Hammer keeps its speed. After a Mirror session you paste in what it turned up, and the draft has little left to guess.
Take one piece of work and two different first minutes:
First minute, Hammer: "Draft the scope change request for my sponsor."
First minute, Mirror: the prompt below, with that same ask pasted at the bottom.
The first gets you a document. The second gets you questions about what the document should say.
Any time-saved figure from me would be a guess. The challenge below gives you two numbers from your own week, and those are the ones that count.
This Week's Prompt
Use this prompt to find what you have not thought through before you build something:
Copy/paste this into ChatGPT or Claude:
WHO: Act as a candid project advisor who has watched many projects stall on untested assumptions and has no stake in my plan.
WHY: because I am about to ask AI to produce [the deliverable: a plan, an update, or a proposal] and I want to find out what I have not thought through before I build it.
WHAT: review my draft thinking below and:List the assumptions I am making that I have not tested.Name the one a skeptical sponsor would challenge first, and what they would say.Tell me which of the points above, if the answer turned out different, would change the content of the deliverable itself.
HOW: turn what you found into no more than five questions, ranked by how much each would change what I build. Stop at the questions. I will ask for the deliverable after I have answered them.
[Paste your unfinished plan, your notes, or the ask you were about to send.]
What this reveals: which of your assumptions are carrying weight, and which question deserves your next ten minutes.
This Week's Challenge
Pick one piece of work you were about to hand to AI this week. Run the prompt above first, then write down two numbers: minutes spent thinking and minutes spent producing.
Before: you open AI and ask for the draft in the first minute.
After: you spend the first five to ten minutes on the Mirror prompt, answer the questions that come back, then ask for the draft with those answers pasted in.
Keep the two numbers. If your week is already full, they tell you whether the thinking minutes changed the producing minutes, and they give you something to compare against next time.
Talk soon,
Paul
P.S. Contact me and tell me your two numbers, and what the Mirror prompt turned up. I'm watching how the split between thinking and producing moves for PMs, and the best examples become future issues. Contact Me