A concrete example: the cheap demo versus the magical agent

The first article uses an example that makes the point well. Imagine an assistant that receives this email:

"Hey, just checking if you're around for a quick sync tomorrow."

A "cheap demo" agent sees only that sentence and nothing else. Its code might be perfectly functional, but with no surrounding context it can only produce something robotic:

"Thank you for your message. Tomorrow works for me. May I ask what time you had in mind?"

A "magical" agent is given rich context before the model is ever called. The system gathers your calendar (which shows you are fully booked tomorrow), your past emails with this person (which show an informal tone is appropriate), your contact list (which identifies them as a key partner), and tools to send an invite. Only then does it generate:

"Hey Jim! Tomorrow's packed on my end, back-to-back all day. Thursday AM free if that works for you? Sent an invite, lmk if it works."

The lesson is that the difference in quality came from the context, not from a smarter model or a cleverer algorithm. The same model produced both replies. The author summarizes it this way: the code's primary job is not to figure out how to respond, but to gather the information the model needs.