Latest Articles

After 25 years of building development teams and watching technology reshape the job market, I've learned to spot patterns. The current AI revolution isn't just another tech trend—it's fundamentally altering the entry point to our industry. And if we don't act deliberately, we risk creating a generation of missing mentors.

Data portability isn't just a GDPR compliance checkbox—it's a fundamental business strategy that determines whether you control your digital destiny or remain permanently tethered to vendor ecosystems. After 25+ years of building API architectures and data exchange systems for enterprises and government institutions, I've learned that the organizations thriving in today's digital landscape are those that treat data portability as a core architectural principle, not an afterthought.

The AI development landscape is evolving rapidly, and developers are discovering that while vibe coding is fun, frameworks provide the structure needed for reliable, cost-effective outcomes.

During my 13 years as CTO, I spent most of my time untangling data integration nightmares. Not the sexy kind you see in tech conferences—I'm talking about the 3 AM calls when someone's ERP system suddenly can't talk to their e-commerce platform. The kind where you realize a company has built their entire operation around a vendor's promise, only to discover they're now hostage to someone else's business model.

In today’s digital landscape, data is the lifeblood of organizations. Yet, too often, data gets locked into proprietary systems, siloed databases, or vendor-specific formats. This lack of portability stifles innovation, increases dependency, and creates inefficiencies. As an integration and data specialist, I’ve seen firsthand how embracing file-based data structures can unlock flexibility, resilience, and control—empowering both businesses and individuals to harness their data’s full potential.

What starts as a "simple" request—like adding a button to duplicate products—can uncover layers of technical and architectural challenges. When I built a Chrome extension for Ergonode PIM, the goal was straightforward: replace a cumbersome, multi-step process with a single click. But delivering that simplicity required solving for SPA navigation, dynamic DOM manipulation, and Chrome extension constraints. This is the story of how we turned a routine task into a lesson in enterprise efficiency—an

Every year, the same ritual: my kids’ birthday parties. Friends and family from all over the country park in my street, and the hassle begins. "Do you have a discount code for me?" I did, but visitors couldn’t use it. I had to log in, enter four license plates, and pay over €30 upfront—with a resident discount, but still. It wasn’t just inconvenient for me; it was especially frustrating for parents with newborns, elderly guests, and anyone who just wanted to park independently.

The AI landscape is shifting toward smaller, more efficient models—and Mistral Small 3.1 is setting the new standard. With 24 billion parameters, a 128k-token context window, and multimodal capabilities (including image recognition and OCR), this open-source model delivers both power and agility. It achieves 150 tokens per second, supports 21+ languages, and outperforms competitors like Gemma 3 and GPT-4 Omni Mini in benchmarks.