Raphael De Lio @raphaeldelio.dev ยท Mar 24

Enterprise agents wonโ€™t scale because you gave them better prompts. ๐—ง๐—ต๐—ฒ๐˜† ๐˜€๐—ฐ๐—ฎ๐—น๐—ฒ ๐˜„๐—ต๐—ฒ๐—ป ๐˜†๐—ผ๐˜‚ ๐˜๐—ฟ๐—ฒ๐—ฎ๐˜ ๐˜๐—ต๐—ฒ๐—บ ๐—น๐—ถ๐—ธ๐—ฒ ๐˜€๐˜†๐˜€๐˜๐—ฒ๐—บ๐˜€.

0 likes 1 replies

?

Replies

Raphael De Lio ยท Mar 24

A lot of AI engineering content still assumes orchestration is the hard part: - Which model? - Which framework? - How many tools? - What memory layer? But once agents touch real infrastructure, the bottlenecks start looking a lot more like classic software engineering.