AI-Assisted Python for Nonprogrammers

Chapter 15. Agentic Coding

As you learned in Chapter 14, an LLM’s training inputs determine what it can produce as output. LLMs can write English sentences because they were trained on English sentences. They’re able to write poetry because they were trained on poetry. And they’re able to write Python code because they were trained on Python code. The more training a model gets in a particular style, vocabulary, and domain, the more it’s able to generate high-quality output in that domain, using that particular style and vocabulary.

This raises important questions: can we ask AI to write software for us? If so, will the quality be any good? Can AI systems replace human programmers? If so, then what is the role of the human in writing software?

In this chapter, you’ll learn about agentic coding, in which people don’t write code themselves, but ask AI to write it for them. You’ll learn about the history of agentic coding, some of the thinking in favor of and against it, and then how to get started using some of the most popular agentic coding systems. By the end of this chapter, you will have written some Python code via an AI agent.

Agentic coding represents one of the biggest, fastest-moving, most consequential, and most controversial topics the software world has ever known. Things will continue to change and evolve over the coming years, and these changes will have a profound effect on how software is designed and written, how engineers are hired, and even what sort of training ...