To say that AI has changed the software industry is an understatement. Its rapid advancement and widespread availability has not only slashed development times for established engineers, but also opened up software creation to just about everyone, regardless of experience.

That’s a huge plus for creativity and innovation. However, there's a world of difference between building an app that works and building one that thousands of merchants can safely rely on.

At VS Labs, we're big believers in AI-assisted development. It's become an important part of how we build and improve our Shopify apps. But we also know that generating code is just one part of the job.

Behind our growing app suite is a team of experienced engineers, supported by more than 25 years of eCommerce experience through Visualsoft. We use AI to move faster, but it's our people and processes that decide what gets built, how it works and when it's ready for merchants.

The catch with vibe coding

Vibe coding has made huge waves in software development, and it's easy to understand why. Instead of writing code line by line, you tell an AI tool what you want to build, and it gets to work. With just a series of well-worded prompts you can add features, fix problems and make changes until you've got something that resembles the app you had in mind. 

For someone with a good idea but little to no coding experience, that's incredibly powerful. An app that might once have taken weeks of work can potentially be prototyped in a single afternoon, no devs required. 

Engineers are getting plenty out of these tools too, using them to handle repetitive work, experiment with ideas and speed up everyday tasks.

The important distinction between those two use cases, however, is that experienced teams understand that code that runs isn’t the same thing as code that’s ready for retail. It’s not even close.

AI can produce an app that looks the part and seems to work as intended, but it can't guarantee that everything happening behind the scenes has been properly considered.

Security, permissions, dependencies, unusual customer behaviour, future maintenance. There are plenty of things that might not be obvious when you're looking at a working prototype, and without the experience to recognise potential problems, it's easy to assume everything's fine. 

That’s relatively low risk when building something just for yourself, but when other businesses are relying on it, the stakes are considerably higher. 

Understanding cause and consequence 

Shopify apps are a perfect example of why that distinction matters. Retailers can install one in just a matter of moments, but behind the button there can be a lot going on. 

Depending on what it does, an app might access customer info, interact with orders, add features to a storefront or connect with other services. Even something as seemingly straightforward as a wishlist can involve customer accounts, product data and multiple integrations.

That means mistakes can have serious consequences. A poorly implemented update might interfere with existing functionality, a security vulnerability can put sensitive info at risk, and an unexpected issue during peak trading can cost businesses revenue they may never recover.

There's also the ongoing work to consider. Shopify regularly introduces new features, updates APIs and changes platform requirements, so apps need to keep up.

That's why building a dependable Shopify app takes a lot more than knowing how to write code, even if AI is doing it for you. It’s crucial to understand how the Shopify platform works, how different components interact and what happens when thousands of merchants start relying on your software.

Keeping humans in charge

At VS Labs, we use an approach called agentic engineering. It involves AI agents taking on substantial parts of the development process, working from defined plans and carrying out tasks that would otherwise take up valuable engineering time. But we never hand an agent an idea and let it loose.

Planning and development

Development starts with identifying something worth building. Ideas are captured as GitHub issues, assessed for priority and reviewed against what merchants actually need. Once an idea is approved, our Product Factory turns it into a structured development plan, setting out the requirements, scope and criteria for completion.

From there, AI agents can get to work writing code, updating dependencies, preparing translations and generating tests. They can also help with browser testing, changelogs and documentation, freeing up our engineers to concentrate on more involved development work.

Review, testing and release

Every development task completed by an AI agent results in a pull request, which must be reviewed and approved by an experienced engineer before it's merged into the codebase.

That means checking more than whether the code works. Security, permissions, billing and keeping data from different Shopify stores properly separated all need careful consideration.

Next is testing. Something can pass every automated check and still be frustrating to use, so our engineers also assess how features behave in Shopify Admin, on storefronts and through Shopify POS. We're building apps for real merchants and shoppers, not just trying to satisfy a test suite.

Even when everything looks good, a feature doesn't go straight into a live app. Changes move through release candidates before reaching a stable release, with engineers making the final decision about what's ready to ship.

Ongoing security, support and accountability

Getting an app ready for launch is only half the battle. Keeping it secure, reliable and working as intended for thousands of merchants is an ongoing responsibility.

At VS Labs, that means regular maintenance, compatibility updates and bug fixes, alongside the security work that goes on behind the scenes.

Our engineers follow established security practices, including regular training, careful management of app permissions, dependency checks and external penetration testing. We also monitor our apps for potential issues, so we're not relying on merchants to tell us if something's gone wrong.

If a problem does arise, there's an experienced engineering team ready to investigate and resolve it. AI can help with that work, but responsibility for the response stays firmly with our people.

With more than 25 years of eCommerce experience and ongoing relationships with hundreds of retailers, we're also familiar with the pressures merchants face, from preparing for peak trading and managing customer experiences to connecting online and physical stores.

That knowledge feeds into the decisions we make about our apps, helping us identify useful improvements, anticipate potential problems and provide informed support when merchants need it.

That experience also helps us get more out of AI. We can take advantage of the speed and efficiency it offers, while applying the engineering expertise and retail knowledge needed to build apps that merchants can rely on.

Explore our Shopify app suite and discover what experienced engineering, supported by AI, can deliver for your store