@maven_hq: Context Management is a key component of the effective LLM usage. Context engineering for agents and multi agent systems is a great place to start. #maven #llms #ai #aiagents

tsosms :

you pay $ 200 and you will get bigger context window

gn0m3ch0mpsky :

This is why Claude code will compress its context window at checkpoints.

David :

Even AI has a short attention span

giulio :

mrvalgard :

why it not prio last answer? it's logical in conversation we ref last response

globalcycleanalysis :

🙄 Just tell me how

rat :

There’s a better talk that was recently given on the "AI Engineer” YouTube channel

stevo :

How do you have time to learn everything about the world?

gnarlinator :

LLMs are just like me 🥹

Philbert :

Where’s then link of the video ? Haha my brain is rotting 🤣

jesus citrus :

In claude do /compact, in coursor do /summarize and you good. Context over 50% is a start for hallucination so keep it below👍

Saintdogz :

This is very important video for the ones that use AI daily. I run my own LLM, the first thing i learned is to separate topics into new chats, to reset the context for the reason you explain in this video! Really nice video bro!💯

PanGagatek :

always just one prompt and one answer 🥰 just refine your prompt each time instead of continuing discussion 🥰

Aki :

Well it seems that I ask question and while AI tries to form answer it already runs into this..

Saib :

😁😁😁

LobosHermanos :

ADHD has gotten so bad fr even AI is getting it

👽 :

Try to reduce the overall context to only what's absolutely necessary and provide additional context in the prompt itself. Keep the context window as small and clean as possible always. This way the quality actually improves the longer your chat becomes...

pvsda :

chat history impacts context management??