How I built an AI-powered B2B e-commerce platform, and why it was the experiment that changed everything

Michał Zabielski describes how, with AI assistance, he built a working prototype of Endora Commerce in four weeks, and what that experiment means for anyone considering building their own B2B platform.

What gap did off-the-shelf platforms leave open?

The starting point was frustration: pricing policy in typical B2C solutions doesn’t scale to B2B requirements, and managing multiple warehouses, pallets, sets, and product groups mirroring an ERP/WMS/CRM structure, is an afterthought in many platforms.

What a month of building actually produced

• Multi-level pricing matching a client’s organisational structure.
• Real-time data synchronisation with external systems.
• AI-assisted admin functions for bulk catalogue categorisation and updates.
• Headless architecture, ready for integrations.

Zabielski is firm on one point: AI speeds up the work, it doesn’t replace domain knowledge. Well-written code without an analysis of real business requirements doesn’t automatically translate into business value, it takes technical skill combined with an understanding of the client’s processes. Endora Commerce, born from that experiment, is now being tested with the first clients who are validating the four-week build hypothesis in practice.

---

Curious what Endora Commerce could look like for your business? Book a call, we’ll show you the working prototype.