Sketch it on a napkin, before you build it
Before you open Claude, take ten minutes to sketch what's actually in your head. This is that sketch: rough, fast, and just for you. It's the raw material you'll bring into the brainstorm.
Seven things to scribble
This is a session with yourself, not a document for anyone else. Here's the same thinking, scribbled out for Retention Pulse.
What goes in each one, and why
The problem
What's the actual problem you're solving? And what's your evidence it's real, not just a hunch?
Products come from problems. If you can't point to evidence the problem exists, you're about to build a beautiful solution to nothing. Start here, every time.
The job
If this thing did one job brilliantly, what would it be?
One sentence. The north star everything else has to serve. If a feature doesn't serve the job, it doesn't make the cut.
The pieces
What real-world things does it deal with?
The nouns. These become the information your tool works with, so just list the real things in your world that it touches.
The connections
Where does that information already live, and what does this need to plug into?
You almost never start from scratch. Your data is already sitting in software you use every day, and Claude can read it straight through your browser.
The features
What does it actually DO with all that?
The verbs. This is where the tool earns its keep, so write what it does to the pieces, in plain words.
The view
What do you open and look at?
The one surface you'll actually use. Keep it to what you genuinely need to see, nothing more.
The edges
What does it deliberately NOT do?
The most important box, and the one nearly everyone skips. "What it won't do" is what keeps the build small enough to actually finish.
That scribble is what you carry into the brainstorm
You don't walk in cold anymore. You walk in with a problem, a job, and a rough shape. That's where Claude turns your napkin into a real plan, and the loop begins.
Go to The Theory →