RAG for non-engineers: what it is and where it actually helps

Mon, 06 Jul 2026 00:00:00 +0000

Retrieval-augmented generation is a simple idea with an awkward name.

Before an AI answers you, it looks something up first.

That is the whole trick. A normal language model answers from memory, the way a student writes an exam from whatever is in their head. RAG hands that same student an open book and says: check the relevant page before you write. The model still does the writing. It has just stopped guessing about facts it was never taught.

How I build AI systems that run without me

Sat, 06 Jun 2026 00:00:00 +0000

“I build AI systems that run without me. Then I share what works.” That is the line at the top of this site. It reads like a productivity boast. It isn’t one. It is a house rule I set for myself after fifteen years of doing the opposite.

For most of my working life I built things that only ran when I was in the room. Custom projects. Client work. Every deliverable was bespoke, every account needed hand-holding, and the day I stopped, the income stopped with me. I collected promotions and delivered for names like Disney and Sony, and none of it changed the basic math: I was trading hours for money and calling it a career.

Writing on Xavier Tai

RAG for non-engineers: what it is and where it actually helps

How I build AI systems that run without me