Teodor-Valentin Maxim

[Teodor-Valentin Maxim @fdd6445]

@SrinivasanSS52 Client: Use AI, deliver faster. What? Why it is taking you so long? I could've done this with AI myself.
Saas: Sure, boss. We'll use AI to iterate faster on features.

Client: WHAT THE HAIL?! WHY MY APP GOING DOWN?

[Teodor-Valentin Maxim @fdd6445]

@0xPrajwal_ Reality is if you can't do your job efficient enough without LLMs, then by default you bring me an additional cost to get you an LLM. And for a business, LLM contracts aren't cheap like a personal plan.

[Teodor-Valentin Maxim @fdd6445]

@0xPrajwal_ Candidate 1. Most likely to walk logically through a problem rather and come up with a solution without relying on a LLMs for help. I'm hiring you to think to resolve issues, not prompt your way through, plus ain't no way I'll pay for an overpriced enterprise contract to OpenAI.

[Teodor-Valentin Maxim @fdd6445]

@seelffff It doesn't make any sense. goose is just another harness, but you can't replace an Anthropic sub with goose because you never mentioned what model you're using and goose needs a model to run.

[Teodor-Valentin Maxim @fdd6445]

@chmielciu @shubh19 @plainionist Probably decided to use tRPC for a simple to-do app and now he is overwhelmed.

[Teodor-Valentin Maxim @fdd6445]

@plainionist No. You're depending on the service of labs like OpenAI or Anthropic, but at least you won't be 'left behind'.

[Teodor-Valentin Maxim @fdd6445]

@chribjel @opencode I can see your point. Still, I don't think it is necessary. Spend resources on improving the tool rather than the on-boarding process for clueless users.

[Teodor-Valentin Maxim @fdd6445]

@Layton_Gott So clanker, OpenClaw ain't a model.

[Teodor-Valentin Maxim @fdd6445]

@suni_code Buy yourself two shawarmas. Their free plan is enough for what you need.

[Teodor-Valentin Maxim @fdd6445]

@PawelHuryn Update your Claude.MD with the secret sauce "No mistakes.", as well. It compliments very well the "Research the codebase before editing. Never change code you haven't read." instruction.

[Teodor-Valentin Maxim @fdd6445]

@DmytroKrasun Or a lot of data is people asking language they should use and a lot of answers 'use X'. LLMs are a blackbox at the end of the day and we will never know the exact reason of why it chooses a specific tech stack when you're letting the model decide your stack, just approximations.

[Teodor-Valentin Maxim @fdd6445]

@DmytroKrasun It defaults to whatever stack they are trained on. If I go to ChatGPT and just ask to help me build something he is gonna recommend python. He will never say Ocaml, since there is a lot more of python code to train an llm on compared to ocaml.

[Teodor-Valentin Maxim @fdd6445]

@aigleeson Meanwhile, your whole work or repo can be replicated with just a deep research prompt in ChatGPT web ui. Please, use your tool to research how pointless the existence of your tool is.

[Teodor-Valentin Maxim @fdd6445]

@SeverusChud Okay, but what is required is that I need someone to remove the deaf chick mission and the stupid prom mission and the gamewill be perfect.

[Teodor-Valentin Maxim @fdd6445]

@zeeg They stole a lot of data upon which they built their products*.

[Teodor-Valentin Maxim @fdd6445]

@neogoose_btw Duh. Of course. They charge you per token input/output. Makes sense if you ask me.

[Teodor-Valentin Maxim @fdd6445]

@NickADobos @LukasHozda Gatekeeping is necessary. You wouldn't want to be operated on by an incompetent surgeon. You don't want your OS to be developed by incapable people, otherwise you should praise Microsoft for their products.

[Teodor-Valentin Maxim @fdd6445]

@vxunderground Okay, but 30,000 layoffs were only in US?

[Teodor-Valentin Maxim @fdd6445]

@rand_longevity Yet, you come from a country that makes you go into debt just for a diploma with no employment prospects.