Use it. Study it. Teach it.
We use AI in everyday work: writing and reviewing code, researching questions, processing information and building automations. Using it regularly gives us reasons to be both enthusiastic and demanding.
No permanent favorite.
We work across commercial and open models rather than treating any provider as a permanent default. Capabilities change quickly. So do costs, context windows, interfaces, tool use, latency and reliability.
A coding assistant, a research tool and a system extracting facts from documents face different demands. We compare models on the task in front of us, including how much checking and correction the result needs.
We also teach practical AI use. That means understanding both what current systems can do and where human review, source verification, security controls and judgment remain necessary.