Top Tweets for #cloudllm

4 months ago

When Should Enterprises Run AI Locally Instead of Cloud LLMs? https://t.co/NNQI7RNsr6 #LocalAI #CloudLLM #EnterpriseAI #AIInfrastructure #PrivateAI #OnPremAI #GenerativeAI #AIStrategy #EnterpriseTechnology #LLM #ArtificialIntelligence

342

もと｜AIで生活快適化（QOLを��げよう） @moto_ai_jp

5 months ago

#CloudLLM に #LocalLLM のことを尋ねる度に #OpenRouter について書かれることが多く、どうせ課金しないと使えないんだろうと避けてきた。しかしいろいろなLLMのハブになるということを知り、便利だから使ってみようとアカウントを作り課金し使ってみると、理由がガッツリわかってしまったw

GUBA

@gubatron

9 months ago

#cloudllm orchestration 1 shoted a full featured tetris in 6m 46s using 4 specialist agents (researcher, architect, programmer and tester) I think i could build a tetris game, that good in 2 days. https://t.co/eKNrC1cTc5 Does this mean in 60 minutes of unsupervised work that it could build what I could in 20 days? Perhaps more if I instantiate more agents. kind of having a Neo moment where I'm starting to believe my agent framework will actually empower me to be 1000x more productive all by myself. what happens with bigger swarms and enabling agents with more tools and their full context capacity? can we use them to improve the orchestration framework itself? to make it easier to use, productice it for other teams to start using AI this way, instead of dealing with the constant yes/no interaction of tools like codex/claude code? we become the bottlenecks if we're hand holding agents the whole time. we can't expect agents to work like humans, they can do better on their own with the right setup.

gubatron's tweet photo. #cloudllm orchestration 1 shoted a full featured tetris in 6m 46s using 4 specialist agents (researcher, architect, programmer and tester)

I think i could build a tetris game, that good in 2 days.
https://t.co/eKNrC1cTc5

Does this mean in 60 minutes of unsupervised work that it could build what I could in 20 days? Perhaps more if I instantiate more agents.

kind of having a Neo moment where I'm starting to believe my agent framework will actually empower me to be 1000x more productive all by myself.

what happens with bigger swarms and enabling agents with more tools and their full context capacity?

can we use them to improve the orchestration framework itself? to make it easier to use, productice it for other teams to start using AI this way, instead of dealing with the constant yes/no interaction of tools like codex/claude code?

we become the bottlenecks if we're hand holding agents the whole time.

we can't expect agents to work like humans, they can do better on their own with the right setup.

123

Top Tweets for #cloudllm

Last Seen Hashtags on Sotwe

Trends for you

Most Popular Users