GitHub's Copilot app is moving beyond code completion into territory that could reshape how leaders manage their work. Ashley Willis, a senior director of developer relations at GitHub, recently shared how she built 40 automated workflows using Copilot's new automation features, transforming her daily routine from context-switching chaos into structured, AI-assisted decision-making. The shift highlights an emerging use case for AI coding assistants: not just writing code, but orchestrating the scattered workflows that consume leadership attention.

What Problem Are These Automations Actually Solving?

Willis describes a problem that resonates across senior leadership roles. Her work lived in fifteen different places: calendar invites, email threads, Slack channels, GitHub repositories, and planning documents. Her brain was the only system connecting them. Meetings bled into each other. Decisions were made in threads she never saw. Action items appeared in documents she'd never been shown, only to resurface weeks later as casual requests for updates.

The GitHub Copilot app, a standalone desktop application for macOS, Windows, and Linux, is built for working with agents rather than just chatting with them. It allows users to run parallel sessions across repositories on separate branches, and includes canvases, which are bidirectional work surfaces where humans and agents operate on the same plan, terminal, or browser session in real time. Automations are scheduled prompts that run against real work context: calendar, email, messages, and GitHub repositories. They connect through MCP servers and integrations, giving them visibility across all the places work actually happens.

How to Set Up Automations That Actually Stick?

• Start with a diagnostic prompt: Willis opened a chat and asked the Copilot app to scan all her work surfaces and suggest automations she hadn't thought of. The app immediately suggested about six, which she then refined and customized to match her thinking style.
• Organize by workflow category: Willis grouped her automations into five categories: daily preparation, staying informed about launches, career development tracking, relationship protection, and administrative cleanup. This structure makes it easier to maintain and refine automations over time.
• Refine iteratively, not perfectly: Willis notes that first drafts weren't perfect, and that's intentional. You teach the automations how you think, give them voice, and adjust them as you learn what actually helps. The process is collaborative, not one-and-done.

Willis built automations in five distinct categories, each addressing a different layer of leadership work. Her morning routine now includes several automations that run before she opens anything. Meeting Prep pulls her calendar and builds context for every meeting, with different formats for one-on-ones versus large syncs versus external calls. Pre-Meeting Access Check verifies she actually has access to the documents and links referenced in meeting invites, eliminating the panic of showing up to a meeting only to discover the agenda is locked. Daily Triage Digest sweeps GitHub, email, and messages for anything that needs her attention.

For staying informed, Willis built Ship Decoder, which finds everything GitHub shipped in the last 24 hours and explains it in plain language she can use in conversations. Launch Radar runs weekly and surfaces upcoming launches that touch her team's space, preventing her from being blindsided. These two automations alone probably save her an hour a day of scrolling through channels trying to piece together what happened.

The career development category surprised Willis most. She built automations that actively work on her own professional growth. Daily Wins Recap runs every evening and summarizes what she actually accomplished. Willis notes that her default mode is to check something off and immediately move to the next thing without recognizing it. Then performance review season arrives, and she's staring at a blank document trying to remember eight months of work. This automation keeps a running record so she doesn't have to. It counters the "what did I even do today?" spiral that hits hardest on busy days.

Willis

"On the days when imposter syndrome is loud, I need something that talks back to it with facts. The robot believes in me even when I don't," Willis explained.

Ashley Willis, Senior Director of Developer Relations at GitHub

The relationship protection category is where Willis wanted to be most honest. Commitments and Follow-Up Tracker searches her own messages for things she said she'd do and flags what she hasn't done yet. This automation is humbling and essential because when she tells someone "I'll look into this" and then forgets, that's a trust problem. The automation protects the trust. The kudos she writes are still hers. The noticing is still hers. The automation just makes sure her brain doesn't steal recognition from the people who deserve it.

Willis emphasizes a critical distinction: these automations don't replace the human parts of her job. They enable connection by handling the scaffolding. Before this system, she'd walk into conversations distracted or running on fumes because her brain was full of operational noise. Now when she sits down in a one-on-one, she's actually present. When she writes recognition for her team, it's specific and real.

Why Does This Matter Beyond GitHub?

Willis's experience points to a broader shift in how AI coding assistants are being used. The GitHub Copilot app was designed for agents working on code, but the automation layer reveals something deeper: AI agents can handle the operational scaffolding that keeps leaders from doing the thinking, connecting, and creating their roles actually require. This is not about automating the human parts of leadership. It's about automating the noise so the human parts become possible.

The distinction matters because it reframes what AI assistance means for knowledge workers. Rather than replacing judgment or decision-making, these tools can reclaim the mental space needed for judgment and decision-making to happen at all. Willis built 40 automations not because she was trying to eliminate her job, but because she was trying to actually do her job.