I ran across an interesting post on 𝕏 by Bindu Reddy that said:

English is the new programming language. AI agents will become ubiquitous!

Stop learning python or C. Learn to think like a machine, talk to models and design these agents

The builders of tomorrow will create and control AI agents

I’m currently using a local LLM fine tuned for coding as a consultant / documentation replacement while working on my latest AI project. I use it for helping with code snippets and quick examples from Python libraries I only have passing (but growing!) familiarity with, or with more complex constructs where examples help follow the flow.

But you have to understand that LLM output is only a good starting point. It usually isn’t entirely accurate, but for my purposes, it works well. I’m not copy/pasting code, but trying to use the LLM’s examples to better understand flows and interactions.

The key is talking to the LLM about complex tasks you have already broken down into smaller units. I’m still solving the big picture issues. The AI is helping with the more granular details.

I’m finding myself spending more time working on the unit tests, verifying the initial implementation the AI assistant has helped me with.