Once there’s an established standard for coding something, it’s usually wise to follow these conventions—whether in your company or for your personal projects.

Take Large Language Model (LLM) APIs as an example. Since OpenAI was among the first to provide widely-used models, their approach to creating a client and sending chat completion requests has become a blueprint for many other LLM providers.

You could interact with your API endpoint directly, without any extra libraries. But following a standard pays off: if you ever need to update your model version or switch providers, making the change is often as simple as updating an environment variable. This flexibility is especially useful when running models locally during development.

If you don’t use a consistent approach across your projects, you’ll likely end up spending a lot of time debugging when even minor changes occur. By reusing the same implementation throughout your codebase, it becomes much easier to adapt, minimizing the risk of sudden, unexpected errors.

New standards will continue to evolve—for instance, the Model Context Protocol (MCP) aims to connect AI models with various data sources and tools. Investing the time to understand and adopt these industry standards is a smart move for future-proofing your codebase and ensuring smoother collaboration within your team.