A plain-English overview of what GitHub Copilot actually does inside VS Code in 2026, written specifically for Business Central AL developers starting with AI-assisted development.
This is part of a series where we try to explore all features of GitHub copilot in VS Code.
What Is GitHub Copilot in VS Code? (A Business Central Developer's Guide)
If your mental model of GitHub Copilot is "autocomplete for code," it's time for an update. That was accurate in 2021. Today, Copilot in VS Code spans a much wider spectrum — from quick inline suggestions all the way to autonomous agents that can plan, edit multiple files, run terminal commands, and open a pull request while you're away from your desk.
This post is the starting point for a full series on using Copilot for Business Central AL development. Before we touch any AL code, it's worth being precise about what Copilot actually is, because the vocabulary matters — "Agent mode," "Chat," "subagent," and "cloud agent" all mean different things, and mixing them up leads to confused prompting later.
The Copilot spectrum, from lightest to heaviest:
- Inline completions — the original feature. As you type, Copilot suggests the next line or block based on context.
- Copilot Chat — a conversational side panel for asking questions, requesting explanations, or making targeted changes.
- Agent mode — give it a task, and it plans the work, edits across files, runs terminal commands, and iterates until the task is complete, checking in with you along the way.
- Cloud (coding) agent — an asynchronous version of Agent mode that runs on GitHub's infrastructure. You assign a task, close your laptop, and come back to a pull request.
- Subagents — isolated helper agents a parent agent can delegate to for research or sub-tasks, keeping the main session's context focused. (This deserves its own dedicated section later in this series — it's more useful than it sounds.)
Why this matters more for Business Central than for general software development
AL is a fairly specific language tied to a fairly specific object model — tables, pages, codeunits, extensions, and events. Generic Copilot tutorials, which are almost always written for JavaScript or Python, don't map cleanly onto BC work. A few reasons why:
Business Central extension development is fundamentally about extending base application objects rather than modifying them directly — Copilot needs to be told this constraint, because it isn't obvious from the code alone. Object ID ranges and naming conventions are team-specific and won't be inferred automatically; they have to be either stated in the prompt or taught once through custom instructions. And BC-specific patterns like event subscribers, table/page extensions, and the Business Central Test Toolkit rarely appear in the generic demos most Copilot content is built around.
What's next in this series
The next post covers environment setup: getting VS Code, the AL Language extension, and Copilot configured correctly for Business Central work — including a couple of settings that trip up newcomers. From there, the series moves into chat modes, slash commands, and the specific prompt patterns that get useful answers out of Copilot for AL development, before building up to real feature work, testing, and eventually agent and subagent workflows.
If you're a Business Central developer curious about where AI-assisted development actually fits into your day-to-day work, this series is built for you specifically — not adapted from a generic web-dev tutorial.
Regards,
Comments
Post a Comment