# Introduction AI code agents are a type of technology called large language models (LLM), which uses neural networks to analyze vast quantities of data. This type of model is based on the ability to recognize patterns in code and complete it with a plausible response. # How AI Code Agents Work AI code agents use a method called interpolation, which consists of remembering how they behaved in the past to provide a coherent response. However, this process can lead to errors of confabulation if not controlled. # Fine-tuning and learning from human feedback (RLHF) LLM models are further improved through fine-tuning on curated examples and training with human feedback (RLHF). This technique allows agents to become more useful and precise. # Implications for developers AI code agents may seem like a useful option, but they are also subject to common errors. It is essential to understand how they work and when to use them to avoid problems.