By the end of this chapter you can take one idea from nothing to a finished, working thing in a single sitting, using every skill from the first half of the course in order.
Everything so far has been a piece: context, a constitution, a project, delegation, editing, judgment, safety. This chapter is where the pieces become one motion. You take a real idea and build it, start to finish, and watch how naturally the skills you already have line up into a single arc.
This is the chapter that was scattered across everything else, now collected into one walkthrough. No new concept. Just the whole thing, done once, so you feel the shape of a real build in your hands. After this, you have done it, and done beats understood every time.
AVATAR OPENER · ~90s
Watch: one idea taken from nothing to a finished thing in a single sitting
HeyGen avatar · generated, consistent presenter
Before any building, one rule that the best builders never skip, even when the task seems too small to need it: design before you build. Decide what you are making and what done looks like before you make anything. The instinct to jump straight to building is exactly the instinct that produces the wrong thing fast. A few minutes of shape first saves an hour of redoing later.
THE ARC OF A BUILD
Every real build follows the same five moves. You have practiced all of them separately. Here they are in order, the way they actually happen.
Idea
One real thing you want, said in a sentence.
Design
What is it, who for, what does done look like. On paper first.
Set up
A Project, a CLAUDE.md, your constitution loaded.
Build
Delegate it. Plan, grill, then let it run.
Review
Read it like an editor. Redirect. Ship it.
A first build is not a new skill. It is your old skills, in sequence.
You have every one of these moves already. The build is just running them in order, once, on something you actually want.
SEE IT
Let us build something real and small: a simple one-page website for a thing you do. A market stall, a service, a side project. Here is the opening move, the design conversation, before a single line exists.
Move 1 and 2: idea, then design first
I want to build a simple one-page website for my weekend bread stall.
Before we build anything, help me design it. Interview me:
who is it for, what is the one thing it needs to make happen,
what are the three things a visitor must see, and what does
done look like for a first version?
Ask me one question at a time. Do not write any code until we
agree the plan.
Notice that is the grill-me move from Chapter 9, pointed at a build. You answer the questions, you agree a small first version (a name, what you sell, when and where, a way to reach you), and only then do you say go. The model builds it. Then you do the Chapter 10 move: read it like an editor, redirect the parts that are generic or not you, and push it from a first draft to a real one.
Keep it small
A first version that exists beats a perfect one that does not. Ship the small thing.
One risk, justified
Make one real choice that is yours, not the templated default. Be able to say why.
Review every part
Nothing ships on the first draft. Run the editor pass over what it built.
Then ship
Done and out beats perfect and hidden. You can always improve a live thing.
NOW YOU TRY · CREATE
Build one real small thing, end to end
Pick one small, real thing you can finish in a sitting: a one-page site, a simple tool, a tidy document with a purpose. Run the full arc. Design it first with the grill-me move. Set up a Project with your constitution and a short CLAUDE.md. Delegate the build. Review it like an editor and redirect once. Finish with something that exists, even if it is small.
Right if you end the sitting with a real thing that exists and works, not a plan for one, and you can name the one choice in it that is specifically yours.
Show the worked solution
The whole drill is proving to yourself that the arc holds. Take the bread stall site. Design: the grill-me conversation lands on a first version that shows the name, what you bake, where and when to find you, and a phone number, aimed at locals deciding whether to come Saturday. Set up: a small Project, your constitution loaded so the copy sounds like you, a three-line CLAUDE.md (plain voice, no marketing fluff, honest about being a one-person stall). Build: you delegate it, agree the plan, say go, and it produces a page. Review: the first draft calls your bread "artisanal" and your stall a "culinary experience", both of which your voice rule forbids, so you redirect: cut the puffery, say it is sourdough baked Friday night, sold Saturday until it runs out. The one justified risk that is yours might be a blunt line like "we sell out by ten, come early." You end the sitting with a real page. It is small, it exists, and one line in it could only have come from you. That is a build. Everything bigger is just this arc, again, with more parts.
WATCH FOR
✗You skip design and jump to building. That is how you build the wrong thing fast. Agree what done looks like first.
✗You aim for perfect and finish nothing. A small thing that exists beats a perfect thing that does not. Ship the first version.
✗You accept the build on its first draft. Run the editor pass. The first draft is generic until you redirect it.
✗You let it pick every default. Make one real choice that is yours and that you can justify. That is what makes it yours.
WHAT YOU LEARNED
The takeaways
A first build is not a new skill. It is your existing skills run in sequence: idea, design, set up, build, review.
Design before you build, even when it feels too small to bother. Decide what done looks like on paper first.
Keep the first version small and ship it. Done and out beats perfect and hidden.
Make one justified choice that is yours, and run the editor pass before anything ships.
Your project · step thirteen
This is the chapter where your thread project becomes real. Run the full arc on it: design the first version, set it up properly, build it, review it, and finish the sitting with something that exists. It does not have to be finished forever. It has to be real today. The rest of Part 5 makes it richer. This chapter makes it exist.
You spent twelve chapters collecting the moves. Today you ran them in one breath and a real thing came out the other end. You are not learning to build any more. You build.