Tool calling is gaining more traction after two years of static content generation by LLMs.
I've taken a slightly different approach in 0dev. 0dev agents generate python code for their tasks on the fly, gather the information and finally execute it after getting confirmation.
I'm working on a few enhancement in the baseline features of the agents before releasing a demo video but if you're interested in the concepts in this space, particularly the OSS end of the spectrum, you can have a look at it here:
Top comments (0)