Into human, tech, startup, building product, growth, and casual rant. Opinion is my own. The other half of Heapstack, a quintuplet of Algoways, a letter of abjd
@azamuddin91 you can't
if you need something like this implement either LLM Gateway (eg: LiteLLM) so you can monitor or host the Claude in server and only that server can access your codebase and not any other project
This is absolutely BONKERS this happened to Railway.
How can Google have no checks and balances on an action like this? Especially an *automated* action.
This literally crippled their entire business.
Thanks for reverting back to pre-1998 commit. We all deserve this. Thank you Mr Gemoy, undo 28 years of progress
I wonder if it will take another 32 years till we realize we were wrong. At that time i probably might still be able to say “i told you so”
And just like that, not the first instances Google did this, it happened to my company before and to others, that’s why i can’t recommend GCP for serious works
It appears Google Cloud has blocked our account, and so some services are unavailable
We’ve escalated this to Google and will keep people posted. Deepest apologies.
@leeweiserngmai1 the way "learning on the job" need to change, and we are still trying to figure things out. this one I can't answer because I came to a similar conclusion, team is shrinking, I can see the future where it's 1 project = 1 engineer, or 1 project = 1 product + 1 engineer, done that
Situation 3: Dev A and B argue, Dev A ask LLM to prototype, so does Dev B. Validate the outcome and discuss the learning. IMHO there are more learning when the outcome is validated with real user rather than on theoretical state.
Situation 1: dev A thinks approach X is correct, dev B thinks Y is the right way. They argue and try to convince each other.
Situation 2: dev A thinks approach X is correct, tells the LLM to implement it.
There is SO MUCH learning in Situation 1, lost when using LLMs....
@leeweiserngmai1 that one is using agent to counter your argument, I say it's still not ideal, because that learning stuck on Dev A. You need 2 Dev to argue so both can learn.
@leeweiserngmai1 I understand it slightly different, by using AI, dev is not learning, because AI won't ague back. So for me, regardless of AI or Not-AI, introduce that "argument", and introduce it later, because code is cheap. Like what we are doing right now, this discussion lead to think+learn
Follow up to the @deadpudds klia long term car park drama
Posting here coz she didn’t post it here. Follow her on instagram for the drama
Update: she has filed to @KPDNKK. If you kena, please file your claims also.