Dream it, Build it, Ship it! | GitHub Copilot Day

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: Dream it, Build it, Ship it! | GitHub Copilot Day · Channel: GitHub · 29:05 · Watch on YouTube

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

Summary

James Montemagno and Pierce Boggan showcase GitHub Copilot’s tools for moving from an idea to working software, including integrations with Slack, the Copilot app, VS Code, and the CLI. They demonstrate agent sessions, automations, issue-based workflows, pull request review and merging, and building and testing web and mobile apps. The presentation also covers customization through MCP servers, plugins, skills, and canvases, plus remote development in WSL and the HydraFusion workflow. They close with links to the Copilot app, community events, and a survey to name the VS Code pet.

Key points

GitHub Copilot is available across the app, CLI, SDK, IDEs, Slack, Teams, and other integrations.
The Copilot app offers interactive, plan, and autopilot modes, configurable tool permissions, and model selection or optimization.
A Slack integration can use conversation context to launch an agent session and show a live application preview.
Automations can handle recurring work such as issue triage, accessibility audits, and dead-code checks.
The app connects agent sessions with GitHub issues, pull requests, worktrees, and VS Code.
Copilot can review pull requests, address review comments and CI failures, resolve conflicts, and support agent merge.
Canvases and other customizations—including MCP servers, plugins, and skills—extend the app for web, iOS, and Android workflows.
The Copilot app can run sessions locally, in the cloud, or in remote environments such as WSL.
HydraFusion coordinates model-assisted work, including a primary pass, critique, review, and revision.

Chapters

Start	Chapter	Summary
00:04	Copilot and the agent-native engineering workflow	The presenters introduce Copilot’s end-to-end role in software development and its growing integrations.
02:05	Copilot integration in Slack	James demonstrates connecting a repository to Slack, viewing a live preview, and starting an agent from conversation context.
05:08	Building an app and setting up automations	The Copilot app’s modes, permissions, and models are shown as an agent builds a Seattle planner, followed by a recurring dead-code automation.
10:16	Managing work, code sessions, and pull requests	The presenters show issue-based sessions, worktrees, VS Code integration, pull request review, and agent merge.
15:57	Validating web and mobile apps	They review the Seattle planner and demonstrate agent-driven testing and screenshot capture for iOS and Android apps.
20:34	Customizing the Copilot app with canvases and plugins	The presenters explain how to add MCP servers, plugins, skills, and canvases, and demonstrate custom dashboards.
23:07	WSL and HydraFusion in the CLI	They demonstrate running sessions in WSL and describe HydraFusion’s multi-step approach to completing a coding task.
25:41	Wrap-up and community events	The presenters recap how to get the Copilot app, highlight HydraFusion and integrations, and share community event and VS Code pet-name information.

Quotes

I wake up every day now and I say, how many bugs am I gonna fix? — 00:35

I want to build out a beautiful website — 07:12

it allows me to focus on the outcome that I’m trying to achieve — 19:02

the best model is not having to pick a model at all. — 24:09

Concepts and entities

GitHub Copilot, GitHub Copilot app, HydraFusion, MCP, canvases, VS Code, Slack, GitHub, James Montemagno, Pierce Boggan

Full transcript (original language)

00:04 Hello, I’m James Montemagno, and with me, Pierce Boggan, over here. Hello. And we’re super jazzed just to be bringing you everything that Kyle just talked about. So let’s just honestly get right into it. Everything that Kyle’s been talking about is how GitHub, GitHub Copilot, the amazing tools from Microsoft and our amazing partners are really helping build an agent native engineering system, right? From idea to production to security to deployment and so much more in that entire iterative loop. Every single day, Pierce, I am changing how I build software.

00:35 What’s unique is we used to wake up and say, what bug am I gonna start working on? What feature am I gonna get 25 % done today? I wake up every day now and I say, how many bugs am I gonna fix? How many features am I gonna add? How much value do I get to add to my software for my customers? It’s really amazing. Yeah, our jobs have fundamentally changed a lot in the last year, especially since the advent of agents, definitely since the beginning of AI. It’s awesome that not only can I use something like GitHub Copilot to do

01:05 these things faster, but I have that entire end -to -end where I can start with an issue, I can complete fixing that bug with GitHub Copilot, I can deliver that software all with the same platform. Pretty amazing that that’s all in one. Honestly, GitHub Copilot is everywhere, right? Everything that Kyle talked about from the app to the CLI to the SDK enabling you to build your own agents, leveraging the GitHub Copilot agent harness, but the deep integrations, VS, VS Code, Teams, Slack that we’re gonna talk about, and our amazing integrations into

01:35 Xcode, IntelliJ, Neovim, and integrations with Raycast, and so many more places where you can use your Copilot subscription. And what’s crazy is that these are all iterating and moving so fast. And honestly, the team has been cooking. And shipping so many new features, new models, enhancements, performance, everything every single day, basically. When you wake up, brand new updates are ready to integrate, ready to cook. Yeah, we’re moving super fast. It’s not just like hypothetical that

02:05 AI is making us move faster. I see that every day on our product and engineering teams across all of our products that James just said. I know it’s a lot to keep up with, and so we have the GitHub changelog where you can follow all the latest news. So just keep an eye on that for what the team’s been shipping. And let’s get in the code and let’s cook. Let’s do it. Okay, so I’m gonna start over here directly in a place that maybe some of us are in every single day, other than teams or Slack, some sort of channel. And actually, I’m gonna open this up, a little scary, but here I am at the top. Nothing like sharing your actual

02:36 Slack in a live stream. It’s real. So you, me, and Burke have been working on a little app. We call it threeup, so we can kind of stay connected with each other. And this is the GitHub Copilot integration directly into Slack. And it looks like Burke’s been switching up last night some new layouts here. I can take a look at what he’s got cooking here. That looks really nice. It’s a way for us to stay connected, basically, throughout the day. And we can see directly inside of there, we have these code channels. So you can add and connect your GitHub repository directly to it.

03:07 Here it is, the threeup repository. It’s on a branch. It has a PR already open, and it right here is using Sol. And if I tap over here on the artifacts, I can actually see a live preview of the running application. So I can see that you and I have been committing some code, and Burke, classic, doing it. So we can do our check-ins, everything like that. Now this is live, so I can say, hey, we forgot to add Wes Bos, add him and call it fourup, and

03:37 his GitHub is Wes Bos. He’s gonna be with us coding live later today. How can we forget Wes? You can’t forget Wes. So now what’s gonna happen here is that the agent is gonna start to pick up the conversation that’s in the loop. And in fact, if I go over to my agent, we can actually see the exact agent session in real time, behind the scenes, if you will. So we can see all those messages that Burke was posting directly here. We can see all the different calls that were going on along the way, right there.

04:08 Okay, we got some live mics possibly, so we’ll see. So this will start to cook, and then it will start to update our live preview, which is super awesome. Yeah, and the thing I love about this integration is GitHub Copilot’s coming in. You said it’s where you work. And for a lot of us, like our collaboration tools, whether it’s Slack or Teams, right, that is where we’re having a lot of conversations. On the GitHub Copilot app team, we have a channel. People just pop in and be like, hey, I love this in the app, but yeah, I’m struggling with this. This is confusing, I have this bug. And like, you know, pre-Slack

04:38 integration, we would copy pasta, write that into the Copilot app, VS Code, something like that. Now immediately inside of Slack, where all that context is, they’ve already uploaded the logs, some screenshots, a video. I just say @GitHub. And you go investigate this, propose a fix. There you go, and then it’ll click right in and it’ll be good to go. Yep, all right. Now let’s go over and actually focus on, one of the things that we’ve been and the team’s been working really hard on is the GitHub Copilot app. It’s your hub for all things code, really, and all things Copilot and all things GitHub.

05:08 So we actually start right here on the new tab. What do you wanna build today? So I figured I’d just kind of walk through a little bit here, because there’s a lot of different options here. First, you see on the left-hand side, new, work, automations, customizations, we’ll get to it, and all my projects. These can be local folders, remote Git repositories, projects on GitHub, anything you want. So I have been working on a bunch of things here and I’ll actually show some of the active sessions I have going on. So here, I’m able to go in and pick what mode I want. Interactive, plan, autopilot mode, and

05:40 I have tool permission calls. You’re in control. You can always ask, you can approve all, and there’s the brand new assisted, which is experimental, but basically it automatically approves tool requests that pass an LLM safety check. So like, if you wanna do like a npm lint, like that’s probably gonna be pretty safe, you’re good to go, it’ll handle it for you. Also, I wanna call out, there’s a brand new just launched today, new auto optimizations. So before you have auto mode, which based on your prompt, we’ll take a model for you, a nice little discount, you can now optimize

06:10 it for your request. So you can give it a little hint. So you can do it for efficiency, balanced, or you can do intelligent. It’ll pick different models based on your needs. Now, of course, we have all sorts of models, right? So here we have all the different models from Anthropic, OpenAI, MAI, Gemini, xAI, and I’ve even connected my open router to actually get access to different Qwen models too. I could even run them with Ollama or Foundry local on my machine. You can see all sorts of models, all the models, right? So I’m actually gonna, for this one, let’s do Luna,

06:41 and I’m gonna put it on extra high. You’re in control of reasoning, context window, your accessibility. Now I wanna build a brand new application from scratch. So let’s see if we can cook. I’m here in my Seattle day planner, and Pierce is always visiting me in the Pacific Northwest, and I wanna optimize and build a beautiful website for the greatest day. So I’m gonna hit this little dictate button so I can actually go ahead and talk to it. Hey, okay, I wanna build out a beautiful website, and I want it to be for my friends that visit Seattle, Washington, and I want it to be a day planner for them. And specifically, I want it to

07:12 be for all the best coffee shops and awesome attractions, like the Space Needle, Kerry Park, and Pike Place, for example. And what I really want it to be is a really visually rich. In fact, I’ve given you a folder here called Assets, so take a look at the photos there and use those, but make it so they can really kind of plan out their single day, show it on a map. And then what I want you to do is after you get that built, go ahead, there’s no GitHub repository, so create the GitHub repository, push all the code up, and then go ahead and configure GitHub Pages for me and

07:43 deploy it automatically for me with a CI/CD pipeline. Thank you. And I’m just gonna go ahead and fire it off. Perfect. Now, you got an idea in your brain, you just tell it from idea to production. Now, this is gonna start cooking. Now, a few things are gonna happen here. This is gonna take a look at my skills, my MCP servers, anything like that, and start to get to work. So we can see it’s renamed my session, it’s moved it into my Seattle Planner folder here. I have some things available to me, like VS Code, the run, I can

08:13 take a look here. I’m just gonna think it through it. It’s loading impeccable, which is a skill that we have, we’ll talk about it in a bit, and it’s gonna get to work. It’s gonna take some time, Pierce. So let’s keep moving on, because this is the boundary, you got to keep moving as this is going. So the thing that I want to show off next is actually some automations that I run. So here is really cool, and this is kind of a nice little hidden feature here, is, by hidden feature, I mean it’s directly right there for you, is that these are all my automations for my personal work, but also for my code. So I have some things here, like doing a daily

08:44 issue triage for my application, TinyClips, that reviews all my GitHub issues, looks at priorities and enriches the issues. Here I have an accessibility audit, and every week I run cost tips, using a chronicle, because I want to make sure that I’m being cost efficient. I also have one that’s connected to WorkIQ, so it looks at all of my teams and emails, it gives me a daily report, so I can go ahead and kick it off, and I can look at all my runs, I was shutting down the app, so those I stopped them abruptly, but I can see all the ones that were running here. And you can get inspiration from

09:15 the templates, they’re all right here, so you can click on one, get it set up. I’m just going to go ahead and do a new one here, and I’m going to call this, this is a prompt about removing dead code that Burke gave me, I call it dead code. Burke’s taking a lot of strays today. He was like, I want a dead code. So here I’m just going to say dead code, verify everything, look for obsolete flags, open a PR, and I’m going to go ahead and do this on this Pickleball app I’ve been creating for the .NET Conf keynote, and here I could just auto say auto, you have all the options, and I’m just going to

09:45 run it in the cloud. Perfect. And I could say all tools, I have full control, anything I want, and I can go ahead and just fire that off and select it here. So there’s the kitchen sink, create it, boom, and now that’s going to run every single week in the cloud and get a report, and the PR is going to be generated automatically. Yeah, automations are super cool, like you said, we see a lot of people building personal automations for toil in their day. I have some that run telemetry queries and do things like that, help me prepare for my day. Triage feedback we have on the

10:16 repo from social media, tell me what’s important, but also we’re seeing a lot of engineering teams doing things like issue triage, auto investigation, things like that with these automations. So it’s super cool to see what people are doing with them. And we can see that the Seattle planner is asking me some questions. That’s because I forgot to put it in autopilot mode, so let’s go ahead and switch it, and now it’s just going to keep going. But it did ask me questions because it was in that interactive mode, it wanted to understand a little bit more. So I’m going to go ahead and let it cook here a little bit more, and I’m going to get to other work. So you click on my work, these are all of the projects that I have here. So these are all of my

10:47 issues, PRs that are open, closed. I can look at pull requests, reviews, I can create a custom dashboard with any filter I want, which is really nifty. I use this active one all the time because if I just go to GitHub, I have 550 repos, a little bit much. So I can create my filters and I can look at my work here. And I can see anything assigned to me specifically. So for example, if I wanted to get to work on this big feature request that a lot of my users of my stream timer, which I use for countdown clocks, here, it only

11:20 does a system beep. I want to have custom like MP3s and waves. If I want to create a new session, I have it here. Let’s start a new session. I’m gonna do it on this machine and just - Perfect. Starts it up. It’s gonna automatically attach this issue and I can say issue fix and go. It automatically creates the worktree for me. So I don’t have to worry about conflicts automatically or anything like that with any of my branches from main. It’s gonna pull everything local and then it’s gonna get to work. And just like that, I’m already starting to do multi

11:51 -agent processing across multiple projects, which is super cool. Yeah, it’s great. And I love having the native like GitHub collaboration built in where I not only have my agents working but also all my planned work. I feel like now with the agents, like the temptation is like, it’s very easy to start things, but it’s actually very difficult to discern what should I be doing, right? And so having that super close to where my agents are working has been super helpful for me. Totally. So let’s go into one that I’ve been doing earlier. I have an application called Tiny Clips, which allows me to take

12:21 screenshots on Mac and Windows. And I’ve been working specifically on multi -capture editing because when I would take a screenshot, and if I did another one, it would close the old one. So I wanted multiple. So I was working here. I kind of did something very similar. I basically came up over here and I said, hey, here’s an issue. Let’s trace through this and get to work. And it puts together a nice little proposal for me. I can see everything here. Here’s the plan. The great part here is I wanna get closer to the code. I can look at the changes or I could just open it in VS Code or

12:51 my favorite code editor here. If I bring up VS code, we can see that I have the same exact worktree down here. So here’s that worktree. It opens it automatically when you click that button. I have the same exact session. You can see it’s running here in the Copilot agent harness. I can see all of the steps. I can chat back and forth directly there. I can create the PR here. It’ll sync between the two. I can click on it and get really deep into the application, run it, do anything I want, which is super nice. So that deep, tight integration

13:22 between your favorite developer tools. Now, one thing that is missing here though, is my slash VS Code buddy up there. Oh, there he is. Hey, how’s it going? He’s like, hey, what’s going on? Yeah, woof. There you go. And you can kind of move him around, just like we kind of have these nice interactive elements in the GitHub Copilot app. Justin and the team have been working on our VS code pet, which doesn’t have a name yet. We’ll talk about a little bit more here, how you can help name him too. I get achievements. I got this little crown on. So the more I’m using the agents and investigating and learning new capabilities, it’s all there. But I want to point out, right? All the same models, everything’s

13:54 available here for us. We have different reasoning levels, the agent modes, everything like that, permission checks, all the things that we kind of just saw. So that nice agent harness running across, which is rad. Now, if I’m back over in the app, right? I’ve reviewed the code, I’m good to go. I can just one click, create a PR. This is now going to do a few things. It’s going to check any requirements I might have in the repo. I was going to fire off right here to go ahead and open that pull request, which runs a skill. And it will then go ahead and fire that off and start to run my

14:24 CI and CD checks. Now that’s going to take a second or so because I’m building Windows and Mac applications based on what I’m changing. So I did one earlier that for all intents and purposes is exactly the same. And I created the PR ahead of time. So we can see that I basically did the same exact thing for the multi edits. And then I have the full validation here. But I want to point out that I have the GitHub Copilot review agent running on GitHub.com. So when I fired this off, it ran all my checks and everything seemed to pass, which was good.

14:55 But it left me a comment here on some code improvements. So what’s great with this integration is that I can one click fix unresolved comments. Or what I like to do is from this dropdown, I can click ready to merge and merge and see all the checks. But I can also turn on agent merge. And agent merge allows the agent to automatically address review issues, CI failures, resolve conflicts and a lot more. So when you’re working on a lot of different branches, a lot of worktrees, this becomes super nice. And you’ll see agent merge is

15:26 going to get to work and it’s going to start to merge, load the skill and get to work and commit back that code automatically. Now I did not do the automatically merge pull request, but you could. If you’re really in that really tight integration, you feel good about it, you can go ahead and put that out there. Yeah, and I mean, even before AI, every single team struggles with like, okay, I did the work, I have the PR, now I have to spend a week babysitting this thing through to actually getting merged. That problem got worse with AI because we’re now sending more PRs than ever thanks to things like the GitHub Copilot app.

15:57 And so agent merge is awesome because now I can submit the PR, I turn on agent merge, I don’t have to babysit the thing. And when all the checks are addressed and I feel good about the state of the PR, I can merge it in. Exactly. Now let’s get back up and look at our hub. Let’s see if it is how it’s going. Okay, it looks like we have a canvas. So that’s good. Let’s go and see here. We have our Seattle canvas. Okay, this is looking pretty good. So we got a Seattle day planner, build a day worth wandering so we can start with a place, beautiful, a good Space Needle. We have a beautiful map here that we can outline our location.

16:28 That’s really cool. They hand drew that. That’s amazing. I can copy this route and I can see all the different locations here. So I did provide the photos, but I actually worked earlier and I had Copilot go grab all the different photos for me, which is cool. So we can see all the different places here, and it actually filtered and it’s testing the app in real time. So it’s actually been adding. this is actually a browser inside the GitHub Copilot app. This is directly here and you can pick and polish, you can pick elements and you can see that it’s been invoking clicking elements, reading elements and doing a lot more. So that built-in deep integration is

16:59 directly inside the Copilot app. So now this is gonna get to work validating because it kind of waited a few minutes for me to answer a question and then it’s gonna create our repo. So I figured while we wait there, I wanna also show you how I do a lot of my iOS and Android development on my Mac. So you have my Mac. let’s head over to my Mac over here and I know we’ve been working on this weight tracking application. We have, yes. You know, we’re entering winter so it’s time to bulk again. So maybe not the time for weight loss, but that’s okay. We’ve been building this iOS app

17:29 together in the GitHub Copilot app. And I think one really cool thing like, you know, the GitHub Copilot app can be used to build any sort of thing, right? We just saw how we can build web apps. We have a built -in browser. There’s an extensibility model called canvases which we’ll show later today inside the GitHub Copilot app. And here I actually have a canvas for testing and viewing the iOS simulator directly in this app. So I can say with natural language, here’s what I want in my app right here inside the GitHub Copilot app. I can validate right here.

18:00 Of course, this is like a real simulator. I can interact with it. I can come over here. I could change like the viewport if I wanted to try a different simulator. And it looks like from scrolling up, like this thing can actually go and drive the actual app from the agent. It can self-validate. Like, if I ask it to test things. And it looks like we’ve got to get this thing ready to ship. So we need some screenshots for this App Store. So I’m just going to fire it off and I’ll let the app take care of that. Like I’m not going to go manually capture these screenshots myself. It’s not 2025 anymore. Exactly. This is a canvas that ships and it has an MCP server.

18:31 So I can actually speak directly to it. So it can understand what to do. So this is great in that iteration loop just like we saw in the web that specifically it was off testing it. And we can see hands up, hands up. I’m not doing anything. Right? And it’s a native Swift application directly right here. Yep. And you can see everything it’s doing, including asking you questions and anything that you want to do. Yep. Yeah. And I mean, I think the really cool thing about the GitHub Copilot app is I feel like it allows me to focus on the outcome that

19:02 I’m trying to achieve, right? What is the feature I’m trying to build? Imparting my wisdom on architectural choices, things like that. And I’m not super close to the code, but that’s kind of intentional right here. And you can even do things like agentic self-verification, which of course is you’re firing off 10 things at once. Like the less things you actually have to do in any one of these threads, the more things you can do, right? And of course, like you said, like if I ever wanted to go and look at the code, I can do so in VS Code. So maybe let’s hop over to VS Code actually. Yeah. In fact, I was working on another app, a .NET MAUI application on Android, this little financial application. Yep.

19:32 And I think what was great here is that these same MCP servers, the same tools, the same agent harness runs in here. So we can see our custom MAUI agent and the same thing. We have all of our different devices. In this case, I’m deploying it here on Android and you can run it the same exact prompt to have it test everything. So no matter what you’re building, no matter what you want to do, it can drive it. I think what’s cool is like a lot of developers using VS Code are used to like the extensions, right? There’s tons of extensions, the mobile canvases extension from John Dick here at Microsoft. And we can see that now the agent is parsing, taking screenshots

20:03 and navigating fully through you and it’s gonna self validate. So I recently added an interest calculator. So I wanted to self validate it, take screenshots, add it to documentation. And now it’s just gonna kind of get to work doing its thing automatically. Scrolling around everything that we needed to do in real time, which is super powerful. So implement something, figure it out and it’s so cool to watch it. It’s literally taking screenshots and stuff right now, it’s super cool. So while that’s cooking and doing some screenshots, I think the thing that we should really talk about is the extensibility

20:34 of the GitHub Copilot app. That’s right. So in the GitHub Copilot app, if you go to this customize tab, you can go install MCP servers, plugins, skills, canvases, essentially any customization you want inside the GitHub Copilot app to make the Copilot app work just like you and your team, right? So I can come in, I can add MCP servers. So my team is using Azure DevOps, I’m using Figma. I wanna do device testing with Playwright. I can install that right here. Of course, plugins from marketplaces

21:05 like Awesome Copilot that our community has curated. So I can one click install things like WorkIQ to give the GitHub Copilot app context on my Microsoft 365 documents, conversations, team stuff, skills. We got Matt Pocock right after us. We got the grill me skill. He’s gonna tell us all about it, but you can go install and discover skills from your marketplaces. You can even actually add your own organizational marketplace. So we have that at Microsoft. We have a curated set of plugins for just people here at Microsoft.

21:35 You can add that marketplace registry directly to the GitHub Copilot app. So every single engineer on your team has access to those plugins. And of course canvases, right? Canvases are basically a visual way for you to drive the app. So we saw the iOS simulator canvas, but there’s other canvases like Azure DevOps, Foundry, Jira. We use the Sentry canvas a lot on our team. It doesn’t just show us like active issues we have for Sentry, but from that canvas, I can actually go in and I can say, oh, this is the top performance issue we have in

22:06 the GitHub Copilot app. Fix this bug. It’s literally one click away. And so then my interaction model shifts from just being so focused on chat to now I have the canvas and I’m using that to actually fire off agent sessions in the app. Let’s head back to my PC because I got two more things I wanna show. Perfect. The first is that you can build your own canvases. You can use slash create canvas. And in fact, I have my own. I have my own repo actions. So I never have to leave the Copilot app to see all of my actions running on GitHub.com. And for TinyClips, I actually

22:36 created my entire release dashboard here. I can create release notes, generate PRs, tag releases, and see everything all in real time, including what’s live for Mac and on Windows. And I can even see the status of what’s on winget today and what’s released, which is super cool. So you can just build your own and these can leverage the GitHub Copilot SDK to infuse the agent inside of these canvases, which is super duper cool. All right. I have two more things I wanna show. First is something down here, which is kind of interesting called, is this WSL? Is it WSL?

23:07 Yeah, I wanna point out here there’s this little icon here and it says it’s running in WSL on Ubuntu. This is a brand new feature. If you go under environment, you’ll see that I am connected to the Windows subsystem for Linux. So I built a little application that I can run here with this little run button. It’s supposed to be there. Where is it at? See, run the app. There we go. Okay, it was there. That should run it. Let’s see. I think I need to, there we go.

23:38 Perfect. And so the GitHub Copilot app, of course I can run sessions locally on my machine. I can also run sessions in the cloud and now I can run them on remotes like WSL as well. And there it is. So there’s my little app. Is this WSL? It is. It is indeed WSL. It’s right here. In fact, I’ll prove it to you really quick. I’m gonna go over to my terminal, I’m gonna type in btop over here. I’m gonna make this bigger and now I have btop running directly inside of it. So this is right directly inside of this and you can just run everything configured inside of WSL.

24:09 So you can configure all those environments. Last thing really quick, I’m gonna head over to the CLI. Over here, I wanna talk about HydraFusion. So the best model is not having to pick a model at all. So here I’ve gone ahead and I’ve created a new worktree with the slash worktree command and I’m gonna tell it to get to work on this. And HydraFusion here is gonna pick the best approach, best on the prompt. It’s picked critique here and it’s gonna do a few things. It’s gonna do a first pass with a primary worker. It’s gonna then understand that pass

24:40 and go back and critique it. And it’s gonna have a read only reviewer to provide notes on it. And then I’ll do a revision pass and then go ahead and implement it automatically. The cool part here is all this is happening. You can see my AI credits in real time go and I’ve used less than one so far to do these first pass and primary. So the team is working really hard on the output but the idea is you shouldn’t have to worry about the output until the end when it all sort of finishes, which is really nice. There’s gonna be a deep dive in HydraFusion so I’m not gonna go too deep onto it. But I also wanna point out

25:11 if I hit the left arrow, I got a bunch of things cooking over here too. I’m a big GUI person so I love a TUI and I love that I can just click on things and I can navigate around into all of my different sessions here. So I can go and I can see here all of my sessions and I wanna also point out that all of these sessions that I’ve been going on here, I’ve been using HydraFusion on and they also show up inside, if I want to, in the GitHub Copilot app because they’re all there. There’s an option to show your CLI sessions. So as these are working back and forth, they’re all

25:41 readily available for you and then this is gonna go ahead, finish this off. And if I wanted to create a new session, I can just click on new session, I can view all my issues for the repo, open pull requests, everything, deep integration. I love that I can click on things. I love a good mouse click. All right, let’s wrap this thing up as Copilot continues to work. Okay, first thing is you can get the GitHub Copilot app today, gh.io/app to turn your idea into reality and start to check out all of these amazing features

26:11 that we’ve talked about today. The rich ecosystem that continues to build out, publish your plugins, publish your MCP servers and go ahead and configure these for your team, your project or your enterprise. We love to see this continue to be built out. More models are added every single day and you have those readily available to you. Or if you’re like me, I’m Auto Mode and now I’m HydraFusion all day. Every day is the default inside the CLI, I simply love it. And of course you can bring your own key, configure anything you want. Now I do want to

26:41 talk here that the HydraFusion part is really unique. We saw it use critique, but we can also see that there’s other modes such as a single shot, cascade and these different are automatically selecting different models at the right time for the task and switching when it makes sense to cache bust efficiently and effectively. Deep dive into the blog post on HydraFusion, type in HydraFusion GitHub into your favorite search engine, read the deep dive blog post. It is awesome to just see what the team is working on. We showed the deep integration into Slack, also available in Teams.

27:13 If you haven’t gotten that configured, you got to do it. It’s mind -blowing to work on stuff. A bunch of great things to help you dream, build and ship. Last thing really, really quick that I want to point out while you take a screenshot really quick is that there’s tons of great stuff here. Matt’s going to be right after us, grilling us hard with his skills, deep dive on HydraFusion, folks from the VS Code, the Copilot team and you all be back live coding with Wes, hopefully an updated website, automatically there. But fun doesn’t stop today.

27:43 The community is rallying behind the GitHub Copilot Dev Days, and there’s hundreds of events around the globe. You can go to this URL or scan that QR code to take you to the Luma calendar where you’ll see a hundred events. I’m doing one in Portland in October. So if you’re in the Portland area, hope to see you Portland, Oregon, not Maine. So hope to see you there. And these are really awesome to see the community come together. Also, I want to point out that our little buddy does need a name. He does. And we have a whole survey that the team’s been putting on for a chance to win a whole month of GitHub Copilot Max

28:14 if the name is selected. Now you can go ahead and go to aka.ms/vscode-pet to run the same command inside of VS Code, but the team’s been cooking up a little trailer. So let’s go ahead and play that now.

29:05 That’s going to do it for this, but we have so much more in store. So as always happy coding. Pierce, thanks for being here and over to you Burke.