GitHub Copilot everywhere: building code across Slack, Teams, mobile, and VS Code

Video authored by the channel shown, not by IANIA; all rights belong to the creator. Transcript and summary generated automatically; the original-language content is preserved.

Original language: English

Original video: GitHub Copilot everywhere: building code across Slack, Teams, mobile, and VS Code · Channel: GitHub · 12:04 · Watch on YouTube

Topics: AI agents, Task and workflow automation, Software engineering with AI, AI developer tools

Summary

The presenter demonstrates how GitHub Copilot can support development across Slack, Microsoft Teams, mobile, GitHub, and VS Code, using a fitness booking app as an example. In Slack, they launch a feature to show users their class history and describe how the agent provides progress updates, implementation details, and code diffs. They also show how work can continue across GitHub surfaces, including opening a pull request and getting code review. The demo concludes with the class-history feature rendering and a discussion of using the agent in Teams meetings.

Key points

The presenter has developed a fitness booking app primarily in Slack and shows a Slack code channel used to collaborate on trending classes.
Slack includes a live app renderer that supports rapid iteration on front-end workflows.
Copilot work can move between Slack, Microsoft Teams, mobile, GitHub, VS Code, and the Copilot app.
The live feature request is a visual history of the classes a user has attended.
The Slack channel displays information such as the repository, branch, session, model, and implementation plan.
Users can follow cloud sessions on mobile and inspect session progress and spending.
The presenter demonstrates generating promo concepts and reviewing options before continuing with the agent.
Work initiated in Slack can continue into a GitHub pull request and benefit from code review in GitHub.
In Teams meetings, the presenter uses the GitHub agent to create issues and investigate problems discussed in the meeting.

Chapters

Start	Chapter	Summary
00:04	Copilot across surfaces	The presenter introduces the fitness booking app and shows collaboration and live iteration in Slack.
02:09	Starting a class-history feature	The presenter launches a request for a visual log of attended classes and explains the Slack session setup.
04:46	Feature ideas and code review continuity	While the agent works, the presenter shares promo concepts and demonstrates how Slack work connects to GitHub diffs and pull requests.
08:57	Following the agent and using Teams	The presenter checks the agent’s progress and describes using the GitHub agent during Microsoft Teams meetings.
11:02	Reviewing the rendered result	The class-history feature appears, and the presenter reflects on the connected, cross-surface workflow.
11:32	Closing remarks	The presenter invites feedback and hands the session back to Burke.

Quotes

I’ve predominantly developed this app inside of Slack — 00:36

build wherever is natural to me — 02:09

The sessions just flow. — 08:26

that all happens inside of this team’s meeting. — 10:31

every class leaves a mark. — 11:02

Concepts and entities

GitHub Copilot, Slack, Microsoft Teams, VS Code, GitHub, Agent, GitHub Copilot CLI, GitHub Copilot SDK

Full transcript (original language)

00:04 Hello everyone. Thank you Burke for the intro. Uh as Burke said, today I’m going to talk a little bit more about uh how you can use Copilot from multiple surfaces uh and spend a little time on our Slack surface specifically. Uh the keynote today, James and Piers talked about and showed some of the value of using it in Slack, but I just want to spend a little bit more time building something together and uh so we’ll just we’ll bake the cake together. Uh, so what we’re looking at here is my fitness booking app. This is just like a fun app that I’ve been building that’s a way for

00:36 users to book local fitness classes. Um, and I want to show some of the history of like what I’ve done inside Slack. I’ve predominantly developed this app inside of Slack, which is very fun and it’s useful when I’m on the go and I can just kick things off. Uh, and so an example of the types of things I’ve built here in Slack. If you jump into this code channel, which is a new Slack feature in their uh, Slack code offering, you can see that I’m going back and forth with another PM on my team, Tim, and uh, we’re designing this new ability to see trending classes

01:07 inside of the app. And that way it kind of gives the user some signal on like what are some fun classes that, you know, people are really enjoying. And so, uh, I I kicked this off, but then I asked Tim for his feedback. Tim comes in and says like, okay, looking at this, like, can we hide the trending pills? It’s a bit too noisy. And then right here in Slack, you can see the live app renderer, which makes it really easy to iterate quickly on these kind of like more front-end workflows. Uh, and the GitHub copilot harness is powering all of this. So the harness is returning all

01:37 of the um all the work and it’s allowing us to see it and iterate on it directly. And what I love about this is like this experience is really nice uh inside Slack, especially if you’re working with a team member. We also offer it in Microsoft Teams, but it’s uh especially powerful when you can use it across all surfaces. So I can open this inside of GitHub and I’ll show that flow while we do the live uh the live building together. and I can open it on mobile. It I can pick it up, grab the PR and

02:09 changes and pick it up inside VS Code or the GitHub Copilot app. And so this just allows me to build wherever is natural to me and kind of start work and finish it wherever I happen to be at that time. Uh so one thing I wanted to do together while uh I have you all is build some sort of new feature. Uh I kind of wish this was like a birectional like situation where you guys could suggest features. I think that’d be very fun. But the one I just came up with the top of my head right now is let’s build a visual log uh of every class that I’ve

02:41 been to so that the user has this like self-personaliz personalization. Uh and so I’m going to kick that off and show you live just like how fast this is. I think that’s one of the things I’m enjoying the most is, you know, on the GitHub side, we’ve done a lot to make the experience really good in the cloud sandbox environment. And so you can kind of like feel that experience when you’re using it and developing on it. So I’ve kicked it off. GitHub’s kind of telling me what it’s doing. So the first thing it said to me is like, hey, I’m starting your cloud environment. And then it said, I’m reading the chat history to

03:12 get some context of what we’re doing. Uh, and then from there it opens this code channel which is a way to just kind of keep all of its metadata about this feature inside one place in Slack. You c you can obviously just have it like paste in the thread as well. This is kind of just like a a user choice thing. But then GitHub saying to me, okay, I opened this uh channel here for us to collaborate together. And if I come into this channel, you can kind you can see some new metadata that you probably aren’t used to seeing inside of Slack and Teams, which is now we see the like

03:44 repo it’s working on or the branch it’s working on, the repo it’s working in. And then there’s this open session which I will follow and show you all how I can see all of this inside github.com. And so it is nice to have it all in context. I’m already inside this repo. I can see the other work that’s being done in this repo. Um, I can also see it on mobile, which I wanted to attempt to demo live, though I know it won’t render great, but um, our mobile app is rendering my cloud sessions. And so I can see them all and am able to follow its progress on the

04:16 go. And then from here, I can kind of see what the history is. If someone else came in, it would show me their work as well. I can see what the spend is of the session, which I know is like very important um, for a lot of users to be able to calibrate their own spending and keep an eye on it. And then inside the slack channel, the last thing I can see is the model. Uh you can obviously change that model, but we have our default model here. And so the uh implementation plan is just walking through what the agent is working on, similar to what you would expect in

04:46 other GitHub copilot clients. Uh the runtime is all the same. So the experience that you get inside the CLI or inside the GitHub copilot app should be very similar to what you see inside Slack and Teams. So, right now it’s just working on, you know, building this out. And while it’s doing that, I have a couple of my already baked cakes that I wanted to walk through because I think this experience is fun. And this like use case of a um like a fitness booking app is I think a really fun one to just be able to like quickly visualize what

05:17 what’s going on, what is the agent doing. So, right before I jumped onto the stream, like literally right before I asked uh Patrick, who you’ll see later in the room, I said like, “Hey, what’s a feature you would want?” And he’s like, “I feel like a really good feature is like, what’s the promos? Like, what is a class that’s cheap this week?” Because that might influence, you know, me changing my routine. And so, I just quickly shot a note to GitHub like, “Hey, build this thing, but give me five examples so that I can just or sorry, five, give me three examples that I can just choose from.” Um, and I I said like

05:49 an example is like a $5 promo on Wednesdays for classes. Uh, and so it rendered a couple of options for me. Uh, the first one is this like campaign banner, which I think I like the most. Uh, and then the second option is this almost like this very salesy. This reminds me of like some sort of like Uber Eats, you know, type promo, which I feel like is maybe not the aesthetic I’m going for, but it’s really fun to have this level of iteration because then I can just add Patrick and be like, “Hey, you know, you asked for this feature. What are your thoughts? Is this what you

06:19 were expecting or not?” And uh, you know, from there I can pick it up with the agent. I can also see the code diff inside of um, oh, apparently I can’t right now, but I will show it in my other channel. um when uh the agent will by default show the code diff and then you can open a PR which I feel like is you know a very common workflow of where I would want to go from here. So let’s do that live and then I can kind of show the continuity throughout all of GitHub. Let’s check back in now on on the the

06:52 session we kicked off around doing the class history. So, I’m just checking in on it and it’s uh working on the complete history experience. The test data that I have in the app for this has like 12 attendance records. I’m going to quickly show you all just what it looks like today so we have a reference point when it makes this change. So, here’s our app today that I briefly showed earlier. Um the profile experience has this like classes remaining because it’s based on like a monthly quota, but there’s no real insight into like okay,

07:23 you know, what value did I get from this subscription? And so part of the the thing we’re trying to add is just the the history to kind of make it fun to be like here’s all the things you did, you know, since you’re paying for your monthly subscription. Um, and then I want to show one code div while we’re in this channel so you can see what I’m saying. Uh, very similar to the GitHub PRS view where it’s broken down by file and each change. Um, this obviously would match. Let me see if I have an open PR. Yeah, let’s look at the open PR for this. um they should be

07:55 identical because it’s just the diff inside the sandbox um being opened into a PR. And so you can imagine the continuity here in terms of, you know, in my repo I have copilot reviewing it by default. And so I am doing this work from Slack, but I’m getting all of the benefits of my like enterprise code review inside of GitHub. And all of these things have a great integration across them, which I feel like as a user is nice to not have to think about like, oh, I need to copy and paste context from here to this agent to this thing.

08:26 You know, the the sessions just flow. uh whether it’s from mobile to uh or or from Slack to my local you know desktop client uh or terminal and then into github.com. It’s just that continuity that I think as a user is really nice so that it just happens for me. It’s not something I have to manually think about. This is the the fun of live demos. Uh it should just magically appear with something shortly. Um there we go. That

08:57 was great timing. um as I was doing that. Uh okay. So it looks like we are opening some issues based on the history. Uh which isn’t exactly what I wanted. Uh but we can we can easily redirect here. Oh, okay. We’re doing a rendering. Awesome. So we’re thinking through what each of the issues are and how we need to address it. And then it’s it also appears that the agent is going to make the change. So hopefully that pops up here while we’re doing it live. Um, but luckily we had those uh backup

09:28 examples in case it doesn’t finish in time. And I’ll take one more look at this uh live session so you can just kind of see the experience of as it’s going through here. Uh I can see like the additional agent logs if I want more details as to what the agent’s doing. Obviously in Slack it doesn’t make a ton of sense to like render all of these things, but it’s really nice for me to have that option to be able to go and see its full chain of thought, see what commands it’s running on the sandbox, etc. So, we did talk about Microsoft Teams and I want to make sure that I at least

09:58 mention that. Uh, one of the things that I’ve been really enjoying since we released uh these new integrations in the last few weeks is that inside Microsoft Teams, we’re using it a lot in meetings and that’s been a really fun use case. Specifically, as we’re like talking about status updates, we just add the GitHub agent to our teams meetings and then we’re creating issues. we are like you know saying hey GitHub go investigate this problem that we are hearing about in this meeting. I’ve also been doing it uh with some of our customers just to like be able to show them how the integration works and how

10:31 they should leverage it. And it’s been a very fun like direct demo to be like hey look we can I can show you right here for this open source repo like like the copilot SDK or the copilot CLI what the major changes are and that all happens inside of this team’s meeting. Uh so that’s been a really cool surface for us to kind of explore. Okay, so demo gods have blessed us and just in time we have this rendering of the history class history and you can see that the the the model has made some creative sales pitch

11:02 here around every class leaves a mark. Um but it’s showing over time kind of like the history of what classes um where like you know yoga, pilates, cycling, and strength are called out. Um this is like kind of what I had in mind. think we could definitely iterate on it. This fake data looks very fake, but uh that’s not what the demo is about. The demo is about showing you how powerful uh GitHub Copilot is in multiple surfaces and meeting you where you are. And so I think this timing worked out perfectly. I could not have planned it

11:32 better where you can kind of see this endto-end flow and how fast it feels, how connected it is um to our existing github.com experience. Uh and I hope that you all try it out. I hope you give us lots of feedback. I’m available on Twitter. A lot of us are or X um as well as LinkedIn. Feel free to just like share all of your sentiment using the thing. We’re making it better every day and uh hope that regardless what client you’re using from GitHub Copilot that it’s doing the job and you’re loving the experience. So with that said, I will

12:04 hand it back to Burke. Thank you.