GitHub APAC marketing team automates event ops with GitHub Copilot and Actions

GitHub’s marketing team for Japan and Korea has automated its entire event operations pipeline — from planning and registration screening to post-event cleanup — using GitHub Copilot, Issues, and GitHub Actions. The workflow, detailed in a September 2026 blog post by the team’s marketing lead, replaces a manual process that took up to two days with a system that sets up a new event from a single GitHub Issue.

The automation was built by writing plain-language runbooks, then handing them to GitHub Copilot to generate the underlying code. The team did not write the scripts themselves; instead, they described their existing manual steps in a Markdown file called AGENTS.md and used Copilot to translate those instructions into automated workflows.

How the pipeline works

The process begins when a team member opens GitHub Copilot and describes an event idea — for example, “I want to run a webinar about AI-assisted development in November.” Copilot reads the AGENTS.md file, which contains the team’s naming conventions, fiscal quarter mappings, time zone rules, and email templates. It then proposes a campaign name adhering to those rules, drafts two versions of an invitation email, and asks clarifying questions defined in the runbook.

Once the human approves the details, Copilot files a GitHub Issue with the correct labels. The moment an “event-setup” label is applied, a GitHub Actions workflow takes over. That workflow creates landing pages, generates tracking links, and configures CRM fields — tasks that previously required manual work across multiple tools.

Registration screening runs on a daily cron schedule. Every morning, a workflow pulls the latest registrant list for every open event, applies screening criteria, and shares the cleaned-up list via a shared channel. For invite-only events, the workflow checks registrants against criteria such as whether the person is a developer at an enterprise account, a student, or a competitor, before approving access.

A single variable called DRY_RUN acts as a safety switch. When enabled, every workflow runs in simulation mode — creating nothing in external systems, filing no issues in other repositories, and sending no lists. This lets the team rehearse changes before they go live.

Key facts about the automation

Aspect	Detail
Manual process duration	Approximately two days per event
Automated setup time	A few minutes after issue creation
Core tools used	GitHub Copilot, GitHub Issues, GitHub Actions
Safety mechanism	DRY_RUN variable for rehearsal mode

Why this matters for Indian developers and marketing teams

The approach is relevant beyond GitHub’s own team. The blog post emphasises that the method works for any repetitive task that runs through a tool with an API or CLI — including event platforms, CRMs, form builders, and analytics services. For Indian startups, SaaS companies, and digital marketing teams that manage frequent webinars, community meetups, or product launches, the pattern offers a low-code path to automation.

The team’s decision to keep a human in the loop — Copilot drafts, but a person decides — also addresses a common concern about automation rigidity. Because the initial conversation with Copilot is flexible, the team can customise individual events without breaking the downstream pipeline.

Not reinventing the wheel

The blog post acknowledges that commercial marketing automation platforms exist. However, the team found that the diversity of APAC markets — different languages, segments, lead definitions, and CRM fields across Japan, Korea, and other sub-regions — made packaged tools expensive to customise. Building the automation on GitHub’s own tools meant that changing a workflow is as simple as opening a pull request.

The post also notes that the same GitHub primitives used for software development — history, visibility, review, and a URL for every decision — now apply to marketing workflows. Every change is tracked, reviewed, and reversible.

Source: GitHub Blog AI — https://github.blog/ai-and-ml/github-copilot/marketing-ops-as-code-automating-events-from-planning-to-follow-up-on-github/