AI doesn’t replace ecommerce development teams—it changes the way they work.

Over the past year, the world of e-commerce has been awash with promises about AI.

“Build a storefront in minutes.”
 “AI will write code instead of developers.”
 “Every brand will be able to build on its own.”

And at first glance, for a moment, it really seemed that was where we were headed. Shopify is bringing AI into almost every layer of its system:
from Sidekick for store management, to AI tools for developers, to adapting to a world where customers search, compare, and shop through AI engines and intelligent agents.

At BOA Ideas, we work with these tools every day, too: testing, developing, experimenting, and in many cases enabling clients to use them themselves. But the more we work with AI on real e-commerce projects, the clearer something becomes: AI can generate code quickly, and even quite well, but it still can’t take responsibility for an e-commerce system over the long term. And that difference is much bigger than it might seem at first.

Shopify isn’t just “adding AI”: it’s changing the way we work

One of the most interesting things happening right now is that Shopify isn’t treating AI as just another feature; it’s building the future of the platform around it.

We can already see this in several areas:

• Sidekick - an AI assistant connected to store data

• Shopify Dev MCP - an AI tool for developers

• Storefront MCP - building AI agents connected to website data

• The Shop app, with AI-powered product discovery

• Adapting websites for a world of AI discovery and LLM search

• Direct purchases from AI conversations and chats

This isn’t a coincidence. Shopify understands that the way people shop is going to change.

Gradually, we’ll see:

• Less manual searching

• More recommendation engines

• More conversational commerce

• More AI agents managing parts of the buying process

Or, as we put it at one of our meetups:

“Brands that aren’t accessible to AI agents simply won’t be part of potential customers’ shopping journeys.”

AI can genuinely accelerate development capabilities

And it’s important to be clear about one thing: these tools really are powerful.

AI can already help with:

• Creating sections

• Writing basic code

• Automations

• Generating reports

• Writing content

• Building flows

• Analyzing data

• Identifying website issues

• Improving operational workflows

For example, Sidekick can:

• Identify products with no image

• Help create coupons

• Generate sales reports

• Analyze collection performance

• Suggest marketing copy

• Help with day-to-day operational tasks

And this really does change how quickly we can work. The problem starts when we confuse:

“AI can generate and assist”

with:

“AI can build a healthy e-commerce system.”

The new problem is no longer slow development

In the past, most e-commerce projects suffered from slow development. Today, we’re starting to see a completely different problem: development that’s too fast, without proper oversight.

Because AI can generate:

• A lot of code

• A lot of components

• A lot of automations

• A lot of “solutions”

But it doesn’t really understand:

• Long-term maintenance

• The system’s ability to scale

• The impact on performance

• Architectural consistency

• Operational dependencies between systems

And this is exactly where projects start to get complicated.

One of the biggest risks: AI-generated technical debt

This may be the topic people are talking about the least right now. For example, AI tools can generate Shopify sections very quickly.

But in practice, we’re seeing cases where:

• Every section is rewritten from scratch

• The code becomes duplicated and bloated

• Performance suffers

• Accessibility breaks

• The structure becomes difficult to maintain

• Different areas of the site lack consistency

Shopify itself still limits some of its AI tools’ capabilities, too.

Based on our experience:

• Sidekick is better at creating new sections than editing existing ones

• Many outputs look similar to one another

• Complex requests still require manual work and significant QA

And that makes sense.

AI is good at creating; it’s still less capable when it comes to big-picture, systems-level thinking.

That’s why developers’ roles are changing

And this may be the most interesting change happening right now.

Developers are becoming less “the people who write every line of code by hand” and more:

• Architects

• Reviewers

• QA owners

• Integration managers

• Systems thinkers

Because in a world where AI can write code in seconds, the real value lies in the ability to:

• Understand the implications

• Manage complexity

• Perform precise QA

• Keep the system healthy over the long term

Because not everything needs to be built

One of the things we strongly believe in at BOA is this: first, make the most of what Shopify already offers, and only then build custom solutions.

The same is true of AI.

Not every idea needs a:

• Rebuild

• Custom app

• AI agent

• Complex automation

• Or a new flow

Sometimes, a simpler solution based on Shopify’s native capabilities is more stable, faster, and much easier to maintain.

We encourage clients to experiment for themselves, and that’s a very important part of our approach. If clients can and want to:

• Create content

• Write prompts

• Test ideas

• Build simple flows

• Or experiment with development

We’re all for it. Technology doesn’t have to “stay with the developers.” But it does need to be used properly.

That’s why, in many cases, we recommend:

• Setting up a test environment

• Testing before production

• Working with QA

• Consolidating code in an organized way

• Understanding the implications before deploying changes live

Because in real e-commerce, even a small change can affect:

• Performance

• SEO

• Checkout

• Integrations

• Tracking

• Accessibility

• Conversion rate

• Day-to-day operations

So, how should you work with AI on Shopify?

1. Start by checking feasibility - not every AI idea makes business or technical sense.

2. Work in a test environment - don’t build directly in production.

3. Let AI help - don’t let it make decisions on its own.

4. Do real QA.

5. Implement changes in an organized way.

In conclusion - what’s next for the world of e-commerce?

We’re probably entering a world where brands can do more themselves, the shopping journey becomes more conversational, and AI agents become a more significant part of how customers discover, compare, and buy products.

At the same time, the ability to manage complexity will matter more than ever.

Because in the end, the site with the most AI won’t win. The winner will be the site that runs reliably, scales the right way, stays easy to maintain, and intelligently connects AI, operations, and real-world e-commerce.

At BOA Ideas, we help brands understand where AI can truly advance their website and operations, and where it’s best to pause, assess, and build for greater stability.

 

---