Architecture, real cases and technical thinking on applied AI.
Noise-free summary of what's changed this year in the Anthropic ecosystem.
A faster, cheaper model redefines what's worth automating.
Why we stopped setting up Pinecone for most projects.
The trick isn't adding more tools, but knowing how to present them to the model.
An AI system without evals is one you don't know is working. Here's how we build them.
From 3h of daily manual work to 20 minutes. The process we followed.
The Model Context Protocol changes how we connect AI with the tools you already use.
When a RAG system is worth setting up and when it's just unnecessary over-engineering.
Documentation, access, tests and a transfer session. The detail matters.
If you want to go deeper on any of these topics for your case, a short conversation usually resolves more than a white paper.