Alex - https://x.com/alexgraveley
Anton - https://x.com/atroyn
00:00 - intro
03:38 - copilot
05:04 - minion
09:24 - the core loop of minion
14:19 - vector programs
16:38 - prompting techniques
19:00 - fine-tuning and generalization
21:20 - model "knowing its done"
23:28 - what can LLMs do
24:19 - reasoning and fine-tuning
28:08 - how copilot was iteratively improved
31:30 - just try things
32:45 - intention vs behaivor
33:22 - human dataset generation
39:26 - AI enables iterative interfaces
40:32 - what happens when models get smarter
43:48 - you dont need a PhD
46:45 - no one knows yet