adrienne
@adrienne
Breaking the devops philosophy, I'm about to deploy a large batch of changes to the GM Farcaster AI bot that I'm pretty sure will break things and not work as expected, but going to do it anyway b/c I think it'll be easier to identify issues in production than locally, and its low stakes since no one really cares if it doesn't work 🤞 🤞
6 replies
1 recast
26 reactions
JB Rubinovitz
@rubinovitz
Do you have any thoughts on agent devops/observability? Seems like it’s still early days.
2 replies
0 recast
2 reactions
adrienne
@adrienne
Havent had to give much thought about it. But if observability became necessary with distributed systems bc they were too complex for any single person to debug, I’d assume it becomes even more so with the addition of agents everywhere
1 reply
0 recast
1 reaction
conorbronsdon.Eth
@conorbronsdon
Charity Majors and I covered this in our recent podcast (which I already evangelized to Adrienne 🫡) https://youtu.be/bTOWNyhnwEQ?si=c444uw4u32P8omv_ + We built a free agent observability & evaluation suite for Galileo: https://www.galileo.ai/agentic-evaluations
0 reply
1 recast
1 reaction
