How a GitHub engineer built an AI Productivity hub with Copilot CLI

GitHub| 00:05:47|Apr 15, 2026
Chapters6
Explains creating a visual home for Copilot CLI tools to fit her workflow instead of relying on a plain CLI.

Britney Elick shows how she built a visual Copilot-powered Command Center with Marvin AI, 11 Labs, and Copilot CLI.

Summary

Britney Elick, a GitHub software engineer, shares her passion project: a visual Command Center built with GitHub Copilot CLI, the Copilot SDK, and Marvin AI. She explains she’s not a CLI person, so she created a visual home for Copilot CLI features to fit how she works. Marvin, an AI chat agent inspired by The Hitchhiker’s Guide to the Galaxy, is embedded in the app to provide context-aware planning, daily and weekly reflections, and prioritized task management without accessing sensitive data like calendars or emails. The system is integrated with 11 Labs, enabling briefing audio playback to aid context switching after meetings. Elick emphasizes personal and work tasks unified in a single place to prevent things from slipping through the cracks, with automatic task additions driven by Marvin. A calendar integration stores daily meeting context and surfaces meetings via a Copilot CLI-powered Work IQ MCP, so she can see times and join links without tab hunting. She’s publicly released a lighter version called Command Center Light to inspire others, while encouraging building from scratch to solve real-life problems. Finally, she argues that best practices for AI are still being written and that the most valuable players are those who build, share, and iterate now.”

Key Takeaways

  • Copilot CLI can be extended with a visual UI using the Copilot SDK to match a user’s workflow.
  • Marvin AI is integrated to provide planning, reflections, and prioritization while restricting access to sensitive data like calendars and email.
  • 11 Labs enables voice briefing of the AI assistant, aiding context switching between meetings and tasks.
  • The project includes a calendar integration and a Work IQ MCP view that surfaces meetings with time and join links in one place.
  • A public repo called Command Center Light offers a starting point for others to fork and adapt, but real value lies in building from scratch for personal needs.
  • User-driven, personalized AI tools can unify personal and work tasks to prevent context loss and improve productivity.
  • The speaker emphasizes that AI best practices are still evolving and invites others to experiment, share, and define the era.”],

Who Is This For?

Essential viewing for developers and product builders who want to understand how to tailor AI tooling to their own workflows, especially those curious about Copilot CLI, Marvin AI, and personal productivity hubs.

Notable Quotes

"This is also integrated with 11 Labs, so it can read my briefing out loud to me, which is fantastic for context switching."
Highlighting the audio briefing capability that aids task switching.
"Copilot did the heavy lifting and I was able to make the thing in my brain come to life."
Emphasizing how Copilot powered the construction of the hub.
"Anyone can be a developer now. Go build something today, not someday."
Motivational takeaway about starting quickly with AI projects.

Questions This Video Answers

  • How can I build a visual UI around Copilot CLI using the Copilot SDK?
  • What is Marvin AI and how does it integrate with Copilot CLI for productivity apps?
  • How does 11 Labs integrate with AI assistants to aid context switching in daily workflows?
  • What is Work IQ MCP and how does it surface meetings and join links in a centralized view?
  • What is Command Center Light and how can I start building my own AI productivity hub from scratch?
GitHub Copilot CLICopilot SDKMarvin AI11 LabsWork IQ MCPCommand Center LightAI productivityPersonal knowledge managementCalendar integration
Full Transcript
[music] Hello, I am Britney Elick and I'm a software engineer at GitHub in the billing org. Today I'm going to show you something that I built as a passion project for myself using GitHub Copilot, the Copilot SDK, and the Copilot CLI. Now, I'm a visual person, not a CLI person, and CLIs are getting a lot of love these days in the world of AI, which is awesome, but they're just not for me. I built something to take all of the parts of the Copilot CLI and give it a visual home that fits the way that I work and think. This is one of the things that I think is the absolute coolest in the world of AI right now and what it's unlocking for folks. Something unique just for you that solves real problems in your life and helps you make sense of the world. So, I'm going to show you exactly what I did related to that with my command center. Now, this is my briefing component and underlying this is Marvin, which is an AI chat agent that is built into this application based on Marvin from The Hitchhiker's Guide to the Galaxy. It's built on top of the Copilot CLI and shown in this app using the Copilot SDK, so I can see all of the things that I like but have it be in a more visual setting. Marvin has context into everything within this app and not all the things he shouldn't have access to like my actual calendar or email or Slack. He helps me plan my week, walk through a daily and weekly reflection, and stay on top of and prioritize what I've actually got going on, which tends to be a lot these days. This is also integrated with 11 Labs, so it can read my briefing out loud to me, which is fantastic for context switching as I'm returning from a meeting or moving between tasks. This is the kind of thing that you just can't find off the shelf because it's built exactly for my own brain. Also in this is I have a task list with the main priority highlighted so that I remember what it is I'm actually working on. I have personal and work tasks unified in a single spot because I am a parent, which often means that my work day and personal life tend to blend together. When something comes up midday, Marvin adds it directly to the list so that nothing ever falls through the cracks. I also have a calendar integration here where I store my meeting context for the day and have a separate callout through the Copilot CLI to the Work IQ MCP, which surfaces the meetings that I have, the time, and the join links here so that I no longer have to go hunting through tabs in between meetings. And then there's just everything else. Um as I have thought of things that I think would be helpful, I have added them to this thing to build what it has become. So, for example, I have a spot for blog posts. I have my uh inbox where I just drop things as I think of them to triage later. I have my annual and monthly and weekly goals. And then I also have a list of contacts that I should reach out to and have them set up on some sort of cadence so that I know, "Hey, you know what? It's been a month or two since you reached out to this person, you should probably do that." It has This is just one example of like things that people are building right now. Um and like I want to see all of it because all of these ideas came from folks sharing things that they are doing and then me incorporating it in my own application. I've back-coded this entirely with AI. Uh Copilot did the heavy lifting and I was able to make the thing in my brain come to life. Uh I built this also to sort of stretch my understanding of how to build with AI. Uh and I think that starting from scratch like this with a new problem is a really great way to improve your own AI skills as well. I made a public version of this repo called the Command Center Light with just a few of the features. So, if you like the catpuchine pastel vibes, you're welcome to fork it and use it as a starting point. But the real value isn't in forking the thing that I made and adapting it to yourself. It's in building something from scratch just for you. Like everything else in engineering, you're never going to get really good at using AI by following someone else's tutorial. You have to build something that solves a real problem in your own life. Start with a to-do list, your calendar, whatever it is you are having trouble keeping context on right now, and build from there. What I really want folks to take away from this type of a demo is that the best practices for building with AI haven't been written yet. These are still things that we have to experiment with and try them out. Developers like you and me get to be the ones that write them. The people who define how we build with AI are going to be the ones that are leaning in right now and figuring it out. The people that are experimenting, building, and sharing what they learn. Now, I've heard it over and over again, this type of industry shift isn't is a moment that does not come around very often. You don't have to wait to get the perfect idea or to have it all figured out anymore. You just have to start and let the momentum of these AI tools that are very good at building carry you to something new. Anyone can be a developer now. Go build something today, not someday. Do it when you're thinking about it. And when you do, share it. Share with your colleagues, share what you built, share what you're reading, share what you're learning, share the things that you found really interesting at 11:00 p.m. and now can't stop thinking about. The more that developers share, the more we get to be the ones who learn how to use this tool and get to define what this era of software looks like. We're all building incredibly cool things right now. Share what yours is so that we can learn from you, too. Thank you so much.

Get daily recaps from
GitHub

AI-powered summaries delivered to your inbox. Save hours every week while staying fully informed.