My local AI assistant will ask for Codex's help.
Just wanted to share what I have been experimenting with between Codex and Qwen running locally in Lm Studio.
I am very new to AI, only now understanding what the "b" meant in the list of models.
I started with using the Codex Desktop app and using the free version. I'm still learning what the free version limits are. I haven't run into any stops saying I need to purchase something yet. I was amazed once I found you can give Codex access to your files. My first question was how to make a script on a Raspberry Pi that uses vlc player to monitor a security camera. The script needed to launch on reboot and reconnect the stream if it drops. To my surprise it said okay, I'll make it for you. It found my current script on my desktop and edited it, and the script works exactly like I asked. I was expecting it to spit out the text of the script, and then I'd copy-paste it.
Anyway, ive been experimenting with local Ai now. I have a 4080 super, I find Owen 3.5 9b fits nicely on it, and it seems to be quite smart. But sometimes it can't figure some stuff out. So I'd go to Codex to help figure it out. Codex made a suggestion to create a bridge between lm studio/qwen and Codex. I thought that's an interesting idea and had Codex do the steps to create that connection. Now Qwen, whenever it has a problem or if I say "Ask Codex", Qwen will send a file to a specific folder that Codex watches and read and reply with steps to do whatever is in question.
This is absolutely wild to me!
Side note. Is this dangerous? I do understand that having an Ai have access to my files, It could delete or edit something that I wasn't expecting it to do. I have mostly sentimental stuff on a separate server, I dont think it has access to that, but im not sure haha