AI WORLD HQ
Picture a team that keeps doing the same chores: checking every proposed change to shared code, writing the same summary, or tagging the same kinds of requests. Each person does it a little differently, and nobody is sure who keeps the process up to date. GitHub's two recent announcements are about exactly this kind of team work.
What Copilot Automations are
GitHub Copilot is GitHub's AI assistant, built into the platform where many teams store and share their code. GitHub says Copilot Automations are shared workflows, which means a group can build them together and keep them running.
A workflow is a set of steps that happens in order. GitHub describes three building blocks you can combine:
- Prompts: the instructions you give the AI, written in plain language (for example, "summarize this change in three sentences").
- Triggers: the event that starts the workflow, such as someone opening a new request for review.
- Tools: the other apps or actions the workflow is allowed to use.
GitHub says the aim is to design a workflow around a task, then let the whole team maintain it. The announcement says these are rolling out, which means not every account may have them yet.
What Copilot Fusion changes
An AI model is the trained system that actually produces the answer, and different models handle different jobs with different strengths. Normally you pick one for each task.
GitHub says that with GitHub Copilot Fusion, you choose Fusion once and Copilot decides which models to use. It calls this "routing" (sending each request to a suitable model) and "orchestrating" (coordinating those models so they work together). Think of it like a dispatcher who assigns each job to the right worker, instead of you deciding for every single job. Fusion is generally available, and GitHub's announcement does not list the individual models it uses, so check GitHub's own documentation for those details.
How to think about the two together
Automations shape the work: what starts a task, what steps follow, and who maintains it. Fusion affects the AI side: which model handles each step. The announcements describe them separately, so the exact way they connect is something to confirm in GitHub's documentation.
What isn't confirmed yet
The announcements don't say which plans include these features, and they don't give step-by-step menus. Because Automations are rolling out, your screen may look different from what's described, or the feature may not be there yet. Don't assume it's available until you see it.
Wrap-up
GitHub's two announcements point the same way: less repeated setup for teams, and less need to pick an AI model by hand. Automations are about shared, maintained workflows, and Fusion is about letting Copilot route tasks across models. The next step is simple: open GitHub's documentation for Copilot today, find the section on Automations, and note one repeated task on your team that a shared workflow could handle.