arrow_left_alt

Blog

Sylius Behind the Scenes: MCP Commerce

September 4, 2025

Imagine you're shopping for winter boots, but you're not exactly an outdoor gear expert. Instead of scrolling through endless product pages, you simply ask ChatGPT, "What boots should I get for hiking in the Polish mountains during winter?" The AI doesn't just spit out generic results - it considers your location, the typical weather patterns, your activity level, and recommends specific products from actual stores, complete with reasoning behind each choice.

This isn't science fiction. It's happening right now, and Sylius just made the first open-source eCommerce platform to officially support it.

When AI becomes your shopping assistant

AI engines like ChatGPT are no longer just chatbots - they're becoming shopping channels. You can already search for products and make purchases directly through these platforms. The AI acts as your personal shopping consultant, guiding you through decisions you might struggle with on your own.

Let’s look at another example. When you need a laptop for graphic design work, wouldn't you rather have an intelligent assistant walk you through the options than browse through hundreds of product listings yourself? The AI can provide contextual product descriptions, explain why certain specs matter for your specific workflows, and even factor in things like your budget, preferred software, and project complexity.

This shift toward AI-powered shopping feels natural because it mirrors how we actually want to shop - with guidance, context, and personalized recommendations.

The MCP revolution

Behind this transformation is something called the Model-Context Protocol (MCP). The MCP is a standardized protocol to connect language models (such as ChatGPT) with external tools, APIs, and systems. MCP allows AI models to make structured calls (similar to function calls) during conversational interactions.

The MCP Server acts as a bridge between the language model and the Sylius application logic. It exposes specific tools (like product searches or order creation) and executes them based on requests from the AI.

Just like APIs and server-rendered content became communication standards that applications had to support, MCP is emerging as the next standard for AI agent communication. It's how AI systems will access and interact with data from various sources.

For eCommerce platforms, this means more than just helping customers find products. Store employees could ask AI questions like "Which regions have the longest delivery times?" or "What's the status of order #12345?" The applications extend far beyond the purchase process into customer service, logistics, and business intelligence.

We're seeing the same pattern that played out in the 2000s with search engines like Google, Bing, and Yahoo - except now it's happening with AI solutions. The interface for interacting with your store might not be a browser or mobile app, but an AI assistant like ChatGPT. Everyone wants this capability, but not everyone is willing to invest in it. There's too much uncertainty - how to approach it, whether the investment will pay off, or if the standard might change completely in just two weeks.

Despite all this uncertainty around this emerging technology, Sylius decided to start working on this solution.

First steps towards MCP integration

First, they started wondering how to give their users the tools they need to be compatible with this emerging standard. The challenge was clear: how do you create a communication layer between Sylius and MCP?

They decided to dive in with two-week sprints, focusing on understanding the technology and validating their direction. While many might want to explore this standard, the barrier to entry is high - it's unclear how to start, there's limited time for experimentation, and calculating ROI on such emerging tech is nearly impossible.

Despite these challenges, they went for it.

They initially built the solution in Python, which is better suited for AI integration work. They created an MCP server that communicates with AI agents and sends requests to the Sylius API. They used the API to deliver data to the server, and this server communicates with ChatGPT. To configure it, they had to “tell” GPT that they're providing additional information and give it the server address, so it knows it can always use this tool during conversations to pull data.

Here's how it works in practice: when someone is interested in finding winter hats, the AI communicates with the server, the server communicates with Sylius, finds out what hats are available in the store, and returns this information to ChatGPT. Since ChatGPT is a natural language model, it can interpret this store's data and provide context, asking questions like "Do you like wool hats?" or suggesting any other options available in the shop. This leads to the server receiving specific questions and learning more details, such as whether a particular hat is currently available in a given size.

This means AI engines like ChatGPT can now receive information from Sylius stores - asking about winter clothing options, getting product recommendations based on conversation context, or checking order shipping status. 

The breakthrough came when they successfully moved their Python prototype to PHP, allowing them to release it as a proper Sylius plugin. This makes integration much more accessible for the Sylius community.

Barriers, difficulties and challenges...

At the beginning, Sylius wanted to leverage the “deep search” tool in ChatGPT. However, they encountered an unexpected problem. This feature allows ChatGPT to prepare more extensive and comprehensive text on topics of interest. Unfortunately, this function turned out to be research-only - it doesn't support interactive functionality. They ended up using the tool in a different context than intended, essentially enriching ChatGPT's knowledge with store content rather than enabling real interactions. Because of this limitation, Sylius couldn't utilize this option for their intended purpose.

Working on AI integration isn't just about technical challenges - they've encountered varying legal regulations across different regions. Currently, some AI features available in the United States remain limited in the European Union. For instance, while US developers can create integrations that connect AI with email, calendars, and personal notes, European users have more restricted access to these capabilities. They have GitHub integration available, but other integrations remain unavailable despite being already developed by companies like Google.

This regulatory patchwork creates challenges for global platforms like Sylius, but it also drives innovation as they work to provide value within different legal frameworks.

{{cta-technology-sylius="/comp/cta"}} 

The future of AI-driven commerce

What they're building could contribute to something much larger - imagine a global marketplace where you can purchase products from multiple international stores through a single AI interface. The AI would handle currency conversion, shipping logistics, and product comparison across vendors worldwide.

But that's just one possibility. The real power is in making commerce more intuitive and accessible. When AI can understand context - your location, climate, activity level, preferences - shopping becomes less about filtering through options and more about getting intelligent guidance.

Why this matters for Sylius

As the first open-source eCommerce platform to officially support MCP, they're not just following trends - they're helping shape the future of online shopping. This integration comes with their MIT license, staying true to their open-source values. This solution could prove to be a better option for stores that want to remain competitive without paying high commissions on global marketplaces. 

The process was intense but rewarding. Their engineering team has pushed boundaries, solved problems they'd never seen before, and created something really new. Every challenge taught them something and made the final product stronger.

What's next

The situation changes daily, and they're continuously refining their solution. This is just the beginning of what's possible when eCommerce platforms embrace AI-native communication standards. They want you along for this ride. Found a bug? Have a better idea? Want to contribute some code? Jump in - that's how open-source works best, and your input genuinely matters. Commerce is becoming more conversational, more contextual, and frankly, more intelligent. With MCP Commerce, Sylius is ready for whatever comes next.

👉 Learn how to enable it in your Sylius instance

<div class="rtb-text-box is-blue-50">Need a Sylius specialist? Contact us, we'll be happy to help!</div>