Home › Blog
Blog
Short, honest write-ups from real projects: where beginners should start, WeWeb vs React, Xano in production, low-code vs vibe coding, slow Xano/PostgreSQL APIs and candidate matching. Most articles are also in French and Arabic.
· WeWeb · React · Low-Code · Frontend · Claude Code · Beginners
I built the same app’s frontend in WeWeb first, then in React. If you’ve never coded, I’d start with WeWeb. If you already know some code, go with React.
Read the article → · also in Français · العربية
· Xano · Backend · Low-Code · PostgreSQL · Production · Claude Code
Yes. I used Xano as the backend of a real ATS/CRM for about two years. What worked, what we had to fix, and how I use it with Claude Code now.
Read the article → · also in Français · العربية
· Low-Code · Vibe Coding · AI · Beginners · Xano · WeWeb
If you want to learn how apps work, I’d start with low-code. If you just need something built, vibe coding is fine. If you want both, the order matters.
Read the article → · also in Français · العربية
· Low-Code · Full-Code · Beginners · Learning to Code · Xano · WeWeb
If you studied code or want to learn the details, go full-code. If you’ve never written a line of code, low-code is an easier way in.
Read the article → · also in Français · العربية
· Full-Stack Development · Xano · React · Low-Code · Backend · Node.js · PostgreSQL · AI-Assisted Development
AI has made moving from low-code to full-code easier than ever. But easier to rewrite does not automatically mean worth rewriting.
Read the article → · also in Français · العربية
· Xano · PostgreSQL · Performance
Production APIs on a Xano/PostgreSQL ATS reached 5–40+ seconds under load. The causes we found and the index, join, pagination and trigger fixes that worked.
· OpenAI · Google Vision · Async workflows
A CV upload waited ~32 seconds for OCR and OpenAI. We redesigned it: create the candidate and answer immediately, then enrich it in the background.
· OpenAI · Embeddings · Vector search
How we ranked a ~35k-candidate pool against jobs: persisted OpenAI embeddings, trigger-based re-indexing, language and distance filters, and a 0–100 score.