Blog

Coding speed was never the real constraint. Discover what actually slows software delivery down in the closing article of our AI and productivity series.
 

Running multiple AI agents in parallel multiplies cognitive load before it multiplies output. What is the real cost of multi-agent AI? Read the article.

Reviewing AI-generated code demands more focus than writing it, yet chat-based AI interaction disrupts the flow state developers need. Read the article.

The Nephila team is experimenting with agentic coding. We're sharing some insights that emerged, which touch on a central issue in the current debate around AI and developer productivity. Read the article.

Vibe coding or software engineering? AI is redefining value in software projects: here's what we're learning at Nephila. Read the article.

Developers tried to jailbreak our AI Debug Challenge. Here's every attack, the layer that stopped it, and why engineering beat the model. Read the article.

We built a game for PyCon Italia 2026: thirty AI-generated Python snippets, three levels of bugs to catch. The AI writing broken code was easy. Read how we engineered everything around it.

Building up a web application requires three coding languages: HTML, CSS, and Javascript. To develop single-apge applications and user interfaces Nephila chooses Vue JS and React JS. 

The web is filled up by real time applications, which are increasingly complex and require different tools to be developed.  Let’s build one of these app with Channels, a Django framework that allows to go beyond http request-response mechanism.

django CMS 3.2, the brand new version of django CMS has been recently released. Let's discover the new features and improvements in this version.