This post was originally published on the Microsoft for Developers blog.
At Microsoft, we strive to build the world’s most beloved developer tools and services. Our vision is to empower every developer to transform their concepts into reality, from idea to code to cloud, at lightning speed. The Visual Studio family, Azure, GitHub, and GitHub Copilot empower developers around the world to do just that.
Since GitHub joined Microsoft in 2018, we’ve been working hard to bring our tools and services together to deliver a productive and delightful experience for developers so they can focus on driving innovation through their code. We started by integrating features and services developers love and use every day from GitHub into our code editors. This includes creating and managing pull requests, issues, and repositories, and making it easier to deploy apps and services to Azure with GitHub Actions. We continue to innovate and collaborate to deliver an unparalleled end-to-end experience for developers and their teams.
GitHub Copilot is the most adopted AI pair programming tool today with more than 1.8 million paid developers in addition to the 1 million students, teachers, and open-source maintainers that use GitHub for free. Copilot now makes it much easier to learn a new technology or programming language, ramp up on a codebase that’s new to you, or learn how to best configure your developer workstation. Let’s take a deeper look at how we’re building, innovating, and integrating GitHub Copilot throughout the entire development lifecycle into the tools and services you love.
GitHub Copilot: Code faster with the world’s most widely adopted AI developer tool
Developing Copilot has been a cross-team effort and we’ve been relentlessly focused on building and releasing innovations to enhance the developer experience. Since the early days of code completions and contextual awareness, we’ve aimed to get to the heart of what developers can use to reduce toil and distraction and replace it with joy.
In 2021, we started by bringing Copilot’s code completions into the Visual Studio family. Today, Copilot is capable of so much more than completion, which makes it an essential part of any developer’s setup. Use Copilot Chat to ask Copilot to explain, refactor, optimize, debug, document, and test your code. As you type, inline suggestions come to life, enabling you to convert comments to code, get help fixing errors, or let it write the test for the function you just wrote.
As we continue to innovate together—we’re creating new ways to interact with chat. In addition to submitting prompts, developers can use keywords to help Copilot better understand your prompt, leverage slash commands to avoid writing complex prompts, and incorporate chat variables to specify context. Recently, we have added new support for multimodal modals, and the ability to speak directly to Copilot.
Perhaps most excitingly, AI model selection in Copilot unlocks a new world of possibilities giving developers the freedom to explore new models and select workable options. In Visual Studio Code, developers can access models from the Azure AI model catalog like Anthropic’s Claude 3.5 Sonnet, Google’s Gemini 1.5 Pro, and OpenAI’s o1-preview and o1-mini. OpenAI and Claude 3.5 Sonnet models are available now; Google’s Gemini 1.5 Pro is coming soon.
GitHub Copilot continues to evolve with new capabilities for developers. Developers now can secure code using Copilot Autofix in GitHub Advanced Security to help them discover and generate fixes for vulnerabilities in pull requests before they merge to production. AI-powered code review makes feedback from pull requests immediately actionable and easy to see in the context of your code editor. With Copilot Extensibility, extensions can interface with services like Jira, Sentry, and more from the comfort of Copilot Chat in your favorite code editor, without having to disrupt your flow or context-switch. Best yet, you can create your very own extensions that integrate with your own data, services, and workflows into Copilot.
Finally, for task-focused scenarios, we’ve built Visual Studio Code extensions for the new GitHub Copilot Workspace so developers can get straight to the task at hand, jumpstarting the process by describing what they want in natural language.
Alongside Visual Studio Code, GitHub Copilot is deeply integrated into our entire developer tools ecosystem with handcrafted experiences and features based on the work that developers need to do within the tool of their choice.
Visual Studio and GitHub Copilot
Visual Studio is our premier IDE for .NET, C++, and game developers. In Visual Studio, developers build large-scale applications with sometimes a hundred projects for their distributed system. With GitHub Copilot, developers can index and search their entire codebase for deep insight, make code edits across multiple files and projects with the upcoming Edits feature, and get advanced visualizations and debugging insights to help them solve problems.
For example, GitHub Copilot can be used with the Visual Studio profiler to identify and suggest fixes for a performance issue in the Visual Studio profiler codebase itself!
GitHub Copilot has been infused throughout Visual Studio with new features, including the ability to set breakpoints, suggest code fixes, debug exceptions and tests, migrate projects, and much more. With Visual Studio, .NET, and GitHub Copilot, solutions for some of the hardest challenges in building software at scale are now within reach of every developer.
Working across GitHub Codespaces and Visual Studio Code
Setting up a developer workstation is an arduous task, but we’ve built a seamless connection between Visual Studio Code and GitHub Codespaces to make it a breeze. You can use your local install of Visual Studio Code or one hosted on the web to create, manage, work in, and delete Codespaces all via the GitHub Codespaces extension. Dev containers are the common format for setting up both, which helped the GitHub team reduce their dev box setup time from 45 minutes to under a minute to onboard developers.
We love how developers around the globe are using Codespaces for help with everyday development and how they’re being used to teach others to code with samples, workshops, and documentation. The near-instant coding environment has been integrated through all our training modules on Microsoft Learn and has become our standardized way to get started with learning any Microsoft technology.
Better collaboration for dev teams
Software is a team sport. In the last few years, dev teams using Azure DevOps have been seeking better integration with GitHub. Thousands of companies using Azure DevOps want to leverage the strength of both platforms and we’ve been working to make that possible. Developers and dev teams can now use GitHub Advanced Security for Azure DevOps and GitHub Copilot for Business to secure your Azure Repos with advanced scanning features and harness the AI-driven coding assistance of GitHub Copilot, all without leaving your current workflow.
Streamlined integration and migration
We’re also focused on improving the integration between Azure Boards, Azure Pipelines, and GitHub Repositories. This work will empower dev teams to fully leverage the advanced features of Copilot and GitHub Advanced Security, while leveraging enterprise-focused project and tracking capabilities in Azure Boards, and the robust continuous delivery capabilities in Azure Pipelines. Developers can now create GitHub branches directly from Azure Boards, find Azure Boards work item links in GitHub pull requests, trigger Azure Pipeline jobs at pull request time or after changes are merged, and more.
Our goal is to provide developers using all of these products with the same end-to-end traceability they’ve come to expect from Azure DevOps. With the GitHub Enterprise Importer, developers can transition their repos from Azure repos to GitHub repos all while continuing to use Azure Boards and Azure Pipelines, and access GitHub’s powerful capabilities in AI and security.
Simplifying cloud deployments
We want to make the path to Azure easier and more intuitive for developers. To do that, we’re building better integration with GitHub.
Learn how to build and deploy to the cloud with GitHub Copilot for Azure
GitHub Copilot makes it easier to learn how to build and deploy to the cloud and how to do it without friction safely from the cockpit of where your code lives. We recently launched the public preview of GitHub Copilot for Azure, which builds upon the Copilot Chat capabilities in Visual Studio Code to help developers manage resources and deploy applications. Using GitHub Copilot for Azure, developers can get personalized guidance to learn, provision, deploy, diagnose, and estimate costs for Azure services and resources without needing to leave the comfort of their personalized code editor.
Seamless Azure CI/CD integration with GitHub Actions
GitHub Actions for Azure helps you create workflows that build, test, package, release, and deploy to Azure for CI/CD integration. GitHub Actions also includes support for key utilities that can streamline testing and deployment of intelligent apps with a variety of AI models, including Azure Resource Manager templates, Azure CLI, and Azure Policy.
The Azure Developer CLI (azd) gives developers a direct path in the dev loop to provision Azure resources, deploy, and set up a CI/CD pipeline for continuous deployments. GitHub Actions workflows are now automatically generated when creating apps with azd templates and infrastructure for popular services, including Azure App Service and Static Web Apps.
To help start you on your AI app development journey, we’ve launched the AI app templates gallery with a curated collection of templates that can be deployed to Azure using Visual Studio Code or GitHub Codespaces. These templates are built for reliability, repeatability, and security and have a wide choice of programming languages, frameworks, models, and services to choose from, including from popular AI toolchain vendors such as Pinecone, Arize, LangChain, and LlamaIndex.
Tools for building your own intelligent apps
The AI field is evolving rapidly, and it can be hard to keep up. GitHub Models supports secure experimentation with Azure AI models, providing QuickStart code for various languages and frameworks to simplify model exploration. You can seamlessly transition from GitHub Models to Codespaces, Visual Studio Code, and Azure as you go from experiment to prototype to production deployment. If you’re not sure where to start or want to test drive, you can experiment in the GitHub playground for free and scale your AI apps to paid endpoints using Azure for secure, enterprise-level deployment and monitoring.
You can now also scale your AI applications with key insights leveraging Azure AI evaluation and online experimentation Actions (both in preview). These capabilities can be fully integrated into your CI/CD development workflows with pre-production evaluation and online experimentation results posted directly back to GitHub for analysis.
Self-provisioning GitHub from the Azure portal
Procurement of tools and infrastructure should be simple and easily accessible. You can now purchase and self-provision GitHub directly from the Azure portal, including setting up an Enterprise Managed Users configuration. Speaking of Azure, the new GitHub Enterprise Cloud with data residency is built on top of Azure, which means that you get the security, business continuity, and disaster recovery capabilities of the cloud.
Developers build our future
As we look ahead, the future of software development has never been brighter. Microsoft and GitHub are committed to empowering developers around the world to innovate, collaborate, and create solutions that’ll shape the next generation of technology. By combining our strengths, we’re paving the way for a more connected and dynamic digital landscape. We’re excited to continue this journey with you and we can’t wait to see what amazing things we’ll build together.
The post Software is a team sport: Building the future of software development together appeared first on The GitHub Blog.
Source: Read MoreÂ
