In the ever-evolving world of software development, staying ahead of the curve is crucial. One of the latest tools making waves in the industry is GitHub Copilot, an AI-powered coding assistant developed by GitHub and OpenAI. This innovative tool is designed to boost productivity, improve code quality, and foster innovation among developers. I have been diving deep into settings, use cases, instructions and prompt generation for the past couple of months, and am very excited about the progress being made. If you are not already “pair programming” with GitHub Copilot, I urge you get a license and start dive into the world of AI-enabled programming. It has completely changed not only my workstream, but my understanding on many different coding principles. This is going to be the beginning of a series of ongoing blog posts to help support you in your GitHub Copilot development journey.
What is GitHub Copilot?
GitHub Copilot is more than just a fancy auto-complete tool, right Tine Staric? 😉 It suggests code, functions, and documentation in real-time, integrating seamlessly with Visual Studio Code (and other IDEs). This powerful assistant accelerates development by autocompleting syntax and patterns, improves code quality by suggesting best practices and catching errors, and enhances learning by helping developers understand complex constructs.
Model Considerations
One of the biggest complaints I have heard about AI user adoption is that AI doesn’t seem to be that knowledgeable, or it’s not that helpful. I always like to refer to working with AI as it’s much like training a Junior Developer. They may need a little more guidance, but they can prove to be a huge asset when equipped with the correct skills and tools.
When working with your pair programmer, GitHub Copilot, you have the ability to select the model it should use. Each model brings its own strengths to the table. Here are a few of the models I use most frequently.
- GPT-4.1 and GPT-4o: These models shine when it comes to generating clean syntax and maintaining consistency across your codebase. They are ideal for writing boilerplate code, routine functions, and data processing logic that is tailored to your coding style.
These models are often the default starting point for many developers, which is where the “fancy autocomplete” comes into play. When your development work demands deeper reasoning, modular architecture, or you need your pair programmer to understand the deeper business logic (especially when it comes to AL development for Business Central), this is where premium models come into play.
Premium models like Claude 3.5, Claude 3.7 Sonnet, and Gemini 2.5 Pro (just to name a few of my favorites) offer enhanced capabilities that go beyond just syntax.
- Claude 3.5 and 3.7 Sonnet: We all know you can’t go to University to become an AL Developer (as it’s not one of those widely used programming languages). Being an AL Developer for Business Central requires us to understand the business logic and generate modular code. Therefore, Claude models are the most widely used models for AL development. They excel at interpreting complex requirements and translating them into structured, reusable components.
- Gemini 2.5 Pro: This model is particularly great at planning and multi-step reasoning. I use it to help me plan out my customizations, and actually have Gemini help write the prompts that I will use with the Claude models to actually help execute my AL development. If you are interested in building out your own agents, I’ve also found Gemini to be helpful with writing out those instructions.
As models continue to evolve, switching between them based on task type can significantly enhance your productivity. For example, you might use Gemini to help structure your development plan and generate a high-quality prompt. I then pass that prompt to Claude to produce modular AL code (with hopefully fewer questions and fewer iterations). Lastly, I might use one of the GPT models to go back through to write tooltips, captions and documentation. This layered approach allows you to leverage each of the models strengths while staying in the flow of development with GitHub Copilot.
If you are looking for some more information about the types of models available for GitHub Copilot, check out this AI model comparison. It breaks down the recommended models by task.
If you are looking for more information about premium requests, meaning the models that could potentially cost you more money, check out these concepts on billing for requests in GitHub Copilot. There is also a direct link to take you to the model multipliers.
If you are wondering why a specific model is not available to you, and you are part of an organization, reach out to an administrator of your organization, and have them Enable different models through your organizations settings.
How do I get GitHub Copilot?
There are several different GitHub Copilot plans which vary depending on your needs, and if you are using GitHub Copilot as an individual, as part of an organization or an enterprise.
GitHub Copilot Free is available to individual developers who do not have access to Copilot through an organization or enterprise. This free plan does limit the features available to you (think premium requests), but it is a good way to dip your feet in and start working with AI.
GitHub Copilot Pro is a paid option that currently allows unlimited completions, access to premium models. GitHub Copilot Pro+ is the highest plan for individual developers, which allows for more premium requests, and full access to most models.
There are then plans for GitHub Copilot Business and GitHub Copilot Enterprise.
You can learn about all of the different plan offerings here (because they do tend to update quite frequently).
Looking Ahead
The upcoming release of the GitHub Copilot Playbook promises to further enhance the development experience. I’ll break down the different modes to work with GitHub Copilot, explore instructions that can aide in overall development, and guidance on building your own reusable prompts. I’m excited about how AI has improved my coding experience, and my goal is to share tips and tricks to assist you in your AI adoption journey.
Resources
Changing the AI model for GitHub Copilot Chat
BC Development Notebook 