Conversation
Jarkko Sakkinen
jarkko
Edited 1 month ago
One area where I think AI is quite nice: I like to tell roughly how a sequence should look like and ask AI to do me a snippet of Mermaid.
It's far from truth but then I can continue editing the snippet manually. My biggest difficulty with Mermaid is that I use it so rarely that I forget the syntax so this will help me to recall all of that. That's actually gist of this whole exercise.
It does not even require powerful AI: I use a local Ollama running llama 3.2 on my RTX 4090. The sequence does not need to be correct as long as it has about the right lanes and some stuff going back and forth.
This makes me wonder tho what is the value of paying Mermaid Charts, which does this on a paid subscription (that's where I got the inspiration in the first place).
This makes me wonder also all of the paid stuff. If you are smart software engineer in the first place you can get 90% of the "efficiency benefit" with a local 2k-4k GPU, which mostly recalling syntax of e.g., Mermaid.
It's far from truth but then I can continue editing the snippet manually. My biggest difficulty with Mermaid is that I use it so rarely that I forget the syntax so this will help me to recall all of that. That's actually gist of this whole exercise.
It does not even require powerful AI: I use a local Ollama running llama 3.2 on my RTX 4090. The sequence does not need to be correct as long as it has about the right lanes and some stuff going back and forth.
This makes me wonder tho what is the value of paying Mermaid Charts, which does this on a paid subscription (that's where I got the inspiration in the first place).
This makes me wonder also all of the paid stuff. If you are smart software engineer in the first place you can get 90% of the "efficiency benefit" with a local 2k-4k GPU, which mostly recalling syntax of e.g., Mermaid.
0
0
0