The eCommerce platform landscape is changing quickly. For years, the platform decision was relatively straightforward: choose the system that offered the right combination of features, integrations, scalability, and cost, then build your eCommerce experience around it. That model isn’t exactly disappearing, but it is becoming a lot less rigid.
In a recent Human Element roundtable, some of our eCommerce experts discussed where platforms like Shopify, BigCommerce, Adobe Commerce, Shopware, Znode, and other emerging players are headed.
Our panel this time included: Paul Briscoe, director of development; Gary Goodman, senior strategist; Rafael Shkembi, technical team lead; Nidhi Patel, technical team lead; and Kevin Gardner, development manager. I’ve condensed their discussion into this post, which covered everything from serious B2B functionality to composable architectures, AI-assisted coding, agentic commerce, and even the possibility that the eCommerce platform itself could eventually become less important.
Here’s a little teaser about that last point in the conversation:
If you make it to the end of this post, you’ll get more detail on how Paul sees this working.
Generally speaking, though, there wasn’t unanimous agreement on every point in the conversation — which is probably a good sign. But several major themes emerged.
eCommerce Is Moving Beyond the Monolithic Platform
One of the biggest changes happening across eCommerce is the continued separation of the storefront from the systems running behind it.
Traditionally, platforms such as Adobe Commerce or Shopware operated as relatively self-contained environments. The platform handled the catalog, checkout, business logic, CMS, routing, templates, and much of the customer-facing experience. Increasingly, those capabilities are being separated.
Modern commerce architectures might combine an eCommerce engine with a completely separate frontend, search provider, CMS, ERP, PIM, middleware layer, and other specialized services. APIs become the connections holding everything together. That shift can be seen across platforms.
Shopware has embraced an API-first architecture in which its own administrative and storefront experiences rely heavily on APIs. Adobe has similarly expanded its composable approach. Even platforms that historically emphasized simplicity are building more tools that allow commerce functionality to exist outside the traditional storefront. This changes the role of the eCommerce platform itself.
As Paul described during the discussion, the platform running behind a commerce experience can sometimes become less important than the pathways connecting the systems around it.
For increasingly complex businesses, the eCommerce platform may no longer be the center of the technology universe. It can instead become one component within a broader commerce ecosystem. But that flexibility comes with a cost.
Composable commerce can create a best-of-breed technology stack, but every additional service creates another integration point that needs to be developed, managed, secured, monitored, and maintained. Separating the frontend from the backend can also mean developing and deploying multiple applications instead of one.
Composable architecture can be extremely powerful. But it can also become extremely complicated. The right architecture therefore isn’t necessarily the one with the most components or the newest technology. It’s the one that supports the actual needs of the business.
Shopify Is Betting Heavily on Commerce Beyond the Storefront
Shopify is particularly interesting because its direction suggests a future in which the traditional eCommerce storefront may not always be where the transaction begins.
Consumers are already using AI tools to research products, compare options, and make purchasing decisions. The logical next step is for some of those systems to participate directly in commerce transactions. Shopify appears to be positioning itself for that possibility.
Instead of thinking exclusively about the Shopify storefront as the customer experience, the platform can increasingly serve as the commerce infrastructure behind experiences happening elsewhere.
An AI agent, marketplace, conversational interface, or another external application might interact with product information, pricing, inventory, and checkout capabilities without requiring the shopper to navigate a traditional eCommerce site in the same way they do today.
That doesn’t mean the storefront disappears. But it does mean merchants need to start thinking beyond it.
For direct-to-consumer brands especially, Shopify appears well positioned for this shift because of its large merchant ecosystem, streamlined checkout experience, and continued investment in AI and agent-driven commerce. Its B2B capabilities, however, are still developing.
B2B Remains the Platform Stress Test
If you really want to understand the limitations of an eCommerce platform, give it a complicated B2B business.
B2B commerce tends to expose architectural weaknesses quickly because the requirements go well beyond displaying products and processing credit-card transactions.
Businesses may need:
- Customer-specific catalogs and pricing
- Multiple buyers within a company account
- Roles and permissions
- Approval workflows
- Quotes and requests for quotes
- Purchase orders
- Real-time ERP pricing
- Multiple storefronts
- EDI
- Complicated fulfillment rules
- Deep ERP integration
Few platforms handle every one of those requirements elegantly out of the box.
Shopify has made substantial progress in B2B, including stronger support for companies, catalogs, and combined B2B/D2C experiences. But complex requirements such as quoting, multi-stage approvals, EDI, and certain real-time ERP interactions still frequently require third-party applications or custom development.
BigCommerce has historically offered stronger native support for some B2B requirements, including company users and roles. Its B2B Buyer Portal provides a separate experience for many of those functions, although our team has seen mixed reactions from merchants about how disconnected that experience can sometimes feel from the primary storefront.
Platforms built more specifically around B2B, such as Znode, approach the problem differently. Their advantage can be the ability to support complex catalogs, multiple storefronts, integrations, and sophisticated B2B structures without assembling quite as many separate applications.
Adobe Commerce also continues to offer a mature collection of B2B functionality, particularly for sophisticated enterprise implementations, although the overall Adobe ecosystem represents a significantly different investment and architectural commitment than lighter SaaS alternatives.
And that highlights a larger truth:
There is no universally “best” B2B eCommerce platform.
The better question is which platform creates the least friction between your business requirements and the technology needed to support them.
The Checkout Experience Still Matters
Architecture gets a lot of attention, but sometimes the simplest platform advantages remain the most powerful. Checkout is one example.
Shopify has created an enormous advantage through a checkout experience that consumers recognize immediately. Returning customers often already have their information stored through Shop Pay, dramatically reducing friction between product discovery and purchase. That familiarity matters.
Other platforms can create excellent checkout experiences, but Shopify benefits from something difficult to reproduce: a massive network effect. For high-growth consumer brands, that remains one of the reasons Shopify can be difficult to compete against.
For B2B organizations, checkout might be far less important than pricing logic, ERP integration, quoting, or account management.
Again, context matters more than feature-count comparisons.
Vibe Coding Is Probably the Wrong Name for Something Very Real
One of the most interesting parts of our conversation centered on “vibe coding” — using generative AI tools to create software through natural-language instructions.
Is it a gimmick? Our team’s consensus was essentially no. But the term itself may create the wrong impression.
“Vibe coding” sounds like someone asks an AI tool to build something, likes what appears on the screen, and sends it into production. That is exactly the kind of development process businesses should avoid.
A better way to think about the change is an AI-driven development lifecycle. AI can already help developers generate code, write tests, troubleshoot issues, create documentation, explore solutions, and accelerate repetitive development tasks. Those capabilities are only going to improve.
But production software still requires architecture, security reviews, testing, compliance considerations, version control, deployment processes, monitoring, and human judgment.
As Paul explained during the discussion, AI-assisted development works best when it still follows the practices the software development community has spent decades refining.
AI can help write the code. But that doesn’t eliminate engineering. In many ways, it makes engineering judgment more important. Someone still needs to know whether the generated solution is correct, secure, maintainable, and appropriate for the business.
The Developer’s Role Is Changing
One prediction raised during the roundtable was that eventually most software development may begin with AI. The developer then becomes less responsible for manually producing every line of code and more responsible for directing, reviewing, testing, and architecting the systems AI helps create.
We are already seeing the early stages of that transition. Developers across the industry are using AI coding assistants today, and AI-generated code is already making its way into production systems. But this doesn’t mean developers lose their jobs. It just means the leverage available to a skilled developer increases.
The most valuable technical skills may increasingly involve understanding architecture, business logic, integrations, data models, security, deployment, and how systems interact — rather than simply how quickly someone can type code. Knowing what should be built becomes at least as important as knowing how to build it.
JavaScript Isn’t Going Anywhere
Another question was whether today’s dominant web technologies — particularly JavaScript, Node.js, React, and similar frameworks — are built to last.
Frameworks will certainly change, but JavaScript probably isn’t going anywhere anytime soon. JavaScript has been fundamental to interactive web development since the 1990s. The libraries layered on top of it have changed repeatedly.
Prototype gave way to jQuery. jQuery gave way to modern frameworks such as React and Vue.
Those frameworks will eventually evolve too. But underneath them, JavaScript remains deeply embedded in how modern web applications operate.
The rise of headless commerce has arguably made JavaScript even more important because many modern storefronts now operate as standalone applications communicating with commerce platforms through APIs.
Rather than betting on a specific framework lasting forever, businesses should prioritize architectures that minimize unnecessary dependencies and allow components to evolve independently.
Technology will change and good architecture gives you room to change with it.
Could eCommerce Eventually Become Platformless?
The most provocative idea from the roundtable came near the end: What if the future of eCommerce includes significantly less reliance on eCommerce platforms?
It sounds radical until you break down what a platform actually does.
- A shopping cart is fundamentally data.
- A customer record is data.
- An order is data.
- Pricing, inventory, shipment status, and product information are all data.
Platforms provide extremely useful interfaces, workflows, business rules, and infrastructure around that information. Today, rebuilding those capabilities individually would usually make little economic sense. But AI-assisted software development could change that equation.
Imagine a future where an AI-driven development environment can rapidly create the frontend experience, connect directly with payment providers, communicate with an ERP, generate administrative interfaces, monitor errors, and continuously adjust the application. At that point, does every merchant still need a traditional eCommerce platform? We’re not sure.
In some B2B environments, versions of this architecture already exist. The ERP is the true system of record. Orders, pricing, inventory, and customer information all originate there. The “eCommerce site” is sometimes little more than an interface sitting on top of those systems.
Agentic commerce could push that concept further.
Instead of:
Customer → eCommerce Storefront → eCommerce Platform → ERP
the flow could eventually look more like:
Customer or AI Agent → Commerce Interface → ERP / Business Systems
The middle layer becomes thinner.
Potentially much thinner.
The Platform Decision Still Matters — But the Question Is Changing
None of this means businesses should abandon eCommerce platforms. Shopify, Adobe Commerce, BigCommerce, Shopware, Znode, and others solve enormous amounts of complexity that businesses would otherwise have to build and maintain themselves.
That value isn’t disappearing overnight. But businesses evaluating eCommerce technology should start asking a broader set of questions.
Not simply:
Which platform has the features we need today?
But:
How easily can this architecture adapt to what commerce might look like tomorrow?
Can your systems expose product and customer information through APIs?
Can your ERP communicate effectively with your commerce layer?
Can components be replaced without rebuilding everything around them?
Can your technology support commerce experiences beyond the traditional storefront?
And can your development processes take advantage of AI without sacrificing security, maintainability, or good engineering practices?
The next generation of eCommerce probably won’t be defined by one dominant platform. It will be defined by how effectively businesses connect data, systems, interfaces, and increasingly intelligent software. Platforms will remain an important part of that ecosystem. They just might not always be the center of it.
