March 21, 2026
The Myth I Keep Seeing
There’s a growing belief right now that AI can just build everything. I was on a call recently with someone who told me they built an entire app with AI. And honestly, they did. But from my experience, what they really built was the starting point.
An app in isolation might “just work.” But does it handle edge cases, security, real data, and integrations?
The moment that app has to plug into a real system, real data, and real users, the work changes completely. That’s usually where the story actually begins.
Generate a landing page.
Generate a funnel.
Generate the logic.
Ship it.
And in isolation… it looks convincing.
You paste in a prompt.
It outputs HTML.
Maybe even some JavaScript.
It looks complete.
But real production environments don’t work in isolation.
They’re messy.
They’re complex.
They’re layered with years of decisions, integrations, and custom code.
That’s where the real work still lives.
And it’s exactly what I’ve been doing lately.
This perspective also comes from operating as a small, hands-on shop working inside real client environments. Not greenfield startups. Not massive engineering teams. Existing websites. Existing revenue. Existing complexity.
Many teams right now are smaller than they used to be. Budgets are tighter. Expectations are higher. AI is helping fill gaps, but it also means fewer people are responsible for more moving parts.
That’s the environment I’m working in.
Working directly with clients. Collaborating with marketing teams. Maintaining existing systems. Building new functionality. And making sure everything still works when it goes live.
We’re all learning what this new model looks like.
This is just what it looks like from my seat right now.
What I’m Actually Building Right Now
Right now I’m building a custom funnel inside a large WordPress + WooCommerce site for an existing client.
This isn’t a standalone funnel builder.
This isn’t a blank slate.
And it’s not something a third‑party plugin could cleanly handle.
We need complete control over the experience:
- Control over design
- Control over logic
- Control over conditions
- Control over how it integrates with the rest of the site
- Control over performance and keeping things lightweight
This is a live ecommerce environment with:
- Existing customers
- Active orders
- Custom functionality
- Third‑party integrations
- Historical data
- Ongoing marketing campaigns
The funnel itself is triggered post‑purchase.
Depending on what the customer bought…
Depending on their purchase history…
Depending on other conditions…
They see different offers:
- Different pricing
- Different products
- Different flows
All dynamic.
And building this custom, with AI as a tool, means we can shape every detail.
- Not fighting plugin limitations
- Not forcing the business into someone else’s structure
- Designing the exact flow the business actually needs
This is where the gap between AI-generated output and real implementation becomes obvious.
AI Can Generate the Pieces
I use AI constantly while building.
I’m working inside tools like Cursor:
Switching between models.
Using different agents depending on the task.
Sometimes I’m working in terminal-based tools.
Sometimes inside editor chat.
Sometimes iterating quickly across multiple approaches.
This is also where human experience comes into play.
- Knowing which models to use
- Knowing when to switch
- Knowing when one tool is better for architecture and another for implementation
Sometimes solving one problem means using multiple models:
- One to think through the approach
- Another to generate code
- Another to help refactor safely inside an existing codebase
AI helps me:
- Scaffold logic
- Generate UI structures
- Prototype flows
- Debug code
- Explore architecture options
But those are pieces.
Someone still has to assemble the system. That includes reviewing specs, translating requirements, and communicating with clients and stakeholders while the solution takes shape.
Integration Is the Real Work
The designs I might receive from others (marketers now send me HTML, which still makes me smile)… or even the HTML AI generates… are static.
But the real funnel isn’t static.
It needs dynamic WooCommerce data:
- Product titles
- Pricing
- Images
- Descriptions
- Purchase logic
- Conditional rules
That means guiding AI to:
- Pull real product data
- Handle dynamic pricing
- Reference existing products
- Respect business logic
- Work within current architecture
And all of that has to happen inside a live site that’s already doing revenue.
That changes everything.
The Things AI Doesn’t Know
AI doesn’t know:
- What custom code already exists
- What integrations might conflict
- How checkout flows are structured
- What performance constraints exist
- How the marketing team plans to evolve this
- What technical debt lives underneath
- What other features or products are already in the pipeline
- How multiple development teams might interact with the same code
- What historical decisions shaped the current architecture
- Which parts of the system are fragile even if they look stable
It just builds what you ask for.
Which is useful.
But dangerous if you don’t understand the environment.
There’s also the human context. Some of these codebases I’ve worked in for years. I carry knowledge that isn’t documented anywhere:
- Why something was built a certain way
- What was tried before
- What broke
- What had to be rolled back
This is where experience still matters.
The Environment Matters
This work isn’t just writing code.
It’s managing environments:
- Working locally
- Connecting to staging
- Deploying to production
I’m also managing Git workflows:
- Staging changes
- Committing updates
- Pushing to remote
AI helps write the code.
But I still need to understand:
- Git workflows
- Branch strategy
- Deployment flow
- Rollback safety (make sure there's a revert if something goes sideways over the weekend)
- Testing process
These aren’t things AI owns.
They’re things the builder owns.
Working With Marketing Teams
Another layer to this work is collaboration.
Marketing teams bring:
- Campaign ideas
- Offer structures
- Experiments
- Conversion goals
My job is translating those ideas into something that actually functions.
From idea to implementation:
- Landing page
- Product page
- Offer logic
- Checkout behavior
- Post‑purchase flow
It’s not just a page.
It’s a system.
Building Inside a Moving System
The hardest part is doing all of this while the site is live.
- Customers are purchasing
- Orders are processing
- Campaigns are running
So every decision matters:
- Will this slow checkout?
- Will this conflict with plugins?
- Will this break analytics?
- Will this create edge cases?
- Will this scale?
These are the questions that shape the architecture.
And they’re the questions AI won’t ask on its own.
The Role of the Modern Builder
AI hasn’t removed the need for developers.
It’s changed the role.
The modern builder is now:
- Part developer
- Part architect
- Part integrator
- Part translator
- Part system designer
- Part AI pilot
The tools are faster.
But the responsibility is still human.
The Craft Isn’t Going Away
If anything, this work has reminded me how much the craft still matters.
Not just writing code.
- Understanding systems
- Working inside complexity
- Translating ideas into reality
- Guiding AI instead of trusting it blindly
AI can generate the funnel.
Someone still has to make it work, test it, and deploy it safely.
And right now…
That’s the work I’m doing every day.
- AI Can Generate OK Code. Someone Still Has to Do Everything Else. - March 21, 2026
- The Work Isn’t Going Away. It’s Changing. - February 28, 2026
- I’m Not Scaling. I’m Stabilizing. - February 15, 2026