PAPER NO. 08
Building the
Storefront.
Two years ago I repeatedly rebuilt basic stores and kept reaching the edge of what I understood.
Today I direct a largely custom storefront with AI. The distance between those two points was not a leap. It was a sequence of steps I could not skip.
WHERE THE STORE IS NOW
From an idea
to a live change fast.
An idea for the ONE Blades site can occur almost anywhere.
I can capture it by voice, give the AI the change I want, and later turn that instruction into working Liquid without personally writing the code line by line.
That is where the store is now.
Large parts of the site have been rebuilt in custom Liquid. I can describe what I want, inspect what comes back, place it, test it, correct it and keep moving.
I am telling you where I am now because the distance matters.
The rest of this paper is about where I started.
And where I started was nowhere near here.
The point is not really the storefront.
The point is the climb required to become capable of directing the storefront.
There is one rule underneath that climb.
You cannot skip the steps.
Two years ago I struggled to build a basic store.
Today I direct a custom one with AI.
Nothing about that jump was a jump.
The blind
years.
Two years ago I was building stores by hand, alone, with very little understanding of what I was doing.
I would take a trial, build, test, refine and try to understand the tools.
Then I would hit a wall I could not get past.
So I would start again.
Then again.
I repeated that process more times than I want to admit.
Not because rebuilding the same thing was a good strategy.
Because I kept reaching the edge of what I understood and did not yet know how to move beyond it.
This was before AI became a meaningful part of how I worked.
Looking back, that period was frustrating but useful.
I was learning how the store was structured.
What themes did.
What sections did.
What I could change.
Where I kept getting stuck.
At the time it looked like failed attempts.
In retrospect, it was accumulated understanding.
The
turn.
Then I started using AI to help me understand what I was looking at.
At first, that was the job.
Explain this.
What does this setting do?
Why is this broken?
Where does this piece belong?
As the models improved and my doctrine became deeper, the work accelerated.
But I was still far from fluent.
Even earlier this year, I was running into limits.
A friend mentioned Liquid, the templating language Shopify uses, and showed me that the store could be shaped far beyond what I had been doing through the standard editor.
I did not suddenly decide to become a developer.
I changed the question.
Instead of asking: how do I learn to write all of this code?
I started asking: can I understand the outcome well enough to direct AI to write the code?
That was the turn.
I stopped trying to become the coder.
I learned to become the director.
From there, the work changed quickly.
The models could write.
I could describe.
Test.
Reject.
Correct.
And increasingly understand enough of the surrounding system to know where the code belonged and whether the result was right.
Doctrine made that much more effective because the AI already had the important ONE Blades context available while it worked.
What custom code
actually changed.
Standard ecommerce tools are basic for a reason.
They allow someone to create a clean functional store without becoming a developer.
That is useful.
But ONE Blades eventually needed more than a collection of standard commerce sections.
The site had to explain a new category.
Help skaters choose equipment.
Teach setup.
Hold technical guides.
Support ONE Academy.
Explain the creator system.
Tell the founder story.
And still sell the product without burying the customer under all of that information.
Custom Liquid gave me much more control over that architecture.
Instead of asking what the theme already had, I could increasingly ask what the customer actually needed the page to do.
The inline figure skating guides are one example.
They are not there simply to attract traffic.
They exist to answer real questions from skaters, coaches, parents and performers.
The site needed room for education as well as commerce.
Custom development made it much easier to treat those as parts of one system.
The scale changed
because the economics changed.
The current ONE Blades site contains a large amount of custom Liquid.
Pages that would once have required substantial developer time can now be built, tested and revised through a conversation between me and the AI.
That does not make the work free.
I still spend the time directing it.
Testing it.
Finding the mistakes.
Rewriting weak sections.
Checking mobile.
Checking desktop.
Making sure the page actually does what it was supposed to do.
But the economics of experimentation are completely different.
I can test an idea that I would previously have hesitated to commission from a developer.
If it is weak, I can remove it.
If it is close, I can refine it.
If I realize the entire structure is wrong, I can rebuild it.
AI did not make custom work effortless.
It made iteration affordable enough to become normal.
What happens
when I get stuck.
I want to make the mechanics clear because otherwise this can sound more magical than it is.
I still do not sit down and write large Liquid sections myself.
The AI may generate hundreds of lines.
Sometimes I can replace the whole section.
Sometimes replacing everything would be unnecessary or risky, so I ask for the exact part that needs to change.
I locate that part.
Replace it.
Preview it.
Check what happened.
If something broke, I take the result back and continue the conversation.
Over time, I have learned enough of the surrounding structure to place things more confidently.
That is very different from being able to write the code from memory.
I do not need those to be the same skill.
Why you cannot
skip the steps.
This is the rule the whole paper is really about.
And it is bigger than storefronts.
I could not have gone directly from the first basic store I built to the way I work now.
Even if today's tools had been handed to me then, I would not have known enough to direct them well.
I had to climb.
Get to the first step.
From the first step, the second becomes visible.
Reach the second, and the third begins to make sense.
You cannot fully see step ten from step one.
And trying to leap to it can make you miss the understanding the earlier steps were supposed to give you.
The climb is not in the way of the goal.
The climb is how you get eyes for the goal.
That is what two years of rebuilding taught me.
Each apparently small step changed what I was capable of seeing next.
The tools became more capable.
But I also became more capable of directing them.
You build the company
by becoming capable
of building it.
This is the part that goes beyond the storefront.
Building a company repeatedly puts you in front of something you do not yet know how to do.
That is uncomfortable.
But it is also the mechanism.
The store improved because my ability to direct the store improved.
My ability improved because I kept working at the edge of what I understood.
The same pattern appears everywhere else.
Product.
Operations.
Writing.
Content.
Coaching.
The work keeps demanding a version of you that does not quite exist yet.
So you build that version too.
You are never
actually finished.
The site has been rebuilt many times.
It is still not finished.
I no longer think finished is a useful state for something like this.
There is simply the strongest version I know how to build today.
Then I learn something.
A customer gets confused.
I see a weak sentence.
A better route through the site becomes obvious.
A new capability becomes available.
So the site moves again.
Some of that work happens away from the desk.
An idea arrives while walking, training, showering or waking up.
Once I can see the next improvement, I can go back and make it.
Finished is not the goal.
Better is.
The next paper in this series, Direct to the People You Serve, moves from how the storefront was built to why the business behind it is structured the way it is.
No distributors.
No retail partners.
No middlemen between ONE Blades and the skaters it exists to serve.
DO THIS NOW
Find and climb
your next step.
You do not need to be building a storefront to use this.
You need to identify the one step directly in front of you and stop trying to leap to step ten.
Name the thing you cannot yet do.
Be specific.
Not: I want a better business.
Something like: I cannot yet build a product page that explains this clearly.
A vague wall gives you no visible next step. A specific one usually does.
Find the single step directly in front of it.
Not the staircase.
One step.
If you cannot code, the next step may not be learning to code.
It may be getting AI to build one small section and learning how to place and test it safely.
Take it while you still feel lost.
You may not feel ready.
That does not automatically mean the step is wrong.
Being at the edge of what you understand is often exactly where the next piece of understanding gets created.
Look for the next step after you land.
Do not waste energy pretending you can perfectly map ten stages you have never experienced.
Take the current step.
Then look again from the new position.
Accept that better creates another step.
Build the thing.
Use it.
Watch what happens.
Let reality show you the weakness you could not see before the thing existed.
Then improve it.
THE BUILD SERIES · PAPER NO. 08
You cannot see step ten
from step one.
Take the step you can see.
PAPER NO. 09