Listen to this story
|
At GitHub Universe 2023, CEO Thomas Dohmke introduced the world to GitHub CoPilot Workspace, which he believes has the ability to reimagine the very nature of the developer experience itself.
Within Copilot Workspace, developers can ideate, plan, develop, test, and execute code using natural language.
Introduced in 2022, GitHub Copilot has emerged as the world’s most popular AI developer tool. The Microsoft-owned company now anticipates Copilot Workspace to be the next evolutionary step.
“There are various ways in which Copilot Workspace can help a developer throughout the software development journey. One of the most important benefits is its ability to help developers get started on a task,” Jonathan Carter, head of GitHub Next, told AIM in an exclusive interview.
Research conducted by GitHub indicates that initiating a project is frequently one of the most daunting aspects of software development.
“Particularly deciding how to approach a task, which files to look through, and how to consider the pros and cons of various solutions. Copilot Workspace reduces that cognitive burden by meeting developers where a new task often begins–a GitHub issue–and synthesising all of the information in that issue to inform a sequenced plan for developers to iterate through,” Carter said.
GitHub Workspace vs Devin
Earlier this year Cognition Labs announced Devin, dubbed as the world’s first AI software engineer.
Upon its announcement, Devin had the developer community talking, as it effectively cleared multiple engineering interviews at top AI companies and also fulfilled actual tasks on the freelance platform Upwork.
However, Carter believes there are fundamental differences between Copilot Workspace and Devin, even though they both are designed to solve similar problems. “At a high-level, Devin and Copilot Workspace are working towards similar goals–reimagining the developer environment as an AI-native workflow.
“That said, we don’t view GitHub Copilot Workplace as an ‘AI engineer’; we view it as an AI assistant to help developers be more productive and happier,” Carter said.
Nonetheless, the biggest differentiator between the two AI tools is that Devin includes a build/test/run agent that attempts to self-repair errors.
“We initially built a similar capability, which you can see in the demo we gave at GitHub Universe in November, but ultimately decided to scope it out for the technical preview to focus on optimising the core user experience,” Carter pointed out.
“Our research has shown that developers value sequenced functions for AI-assistance, and we want to ensure Copilot Workspace meets developers’ needs to build confidence and trust in the tool before we invest in new features, including productising our build/run/test agent,” he said.
Opening GitHub Workspace for technical preview
Similar to Devin, GitHub now plans to give developers early access to test the newest AI tool for software development. Starting today, GitHub will begin the technical preview for GitHub Copilot Workspace.
“We’re looking forward to getting Copilot Workspace into the hands of a diverse group of developers so we can understand where they’re getting the most value from it today and where we can make adjustments to make it even more valuable in the future,” Carter said.
Developers who had access to Devin spoke profoundly of the AI tool. Yet Devin landed in troubled waters after a software developer claimed, in a YouTube video, that the demo video that Cognition Labs released earlier this year was staged.
Although the startup provided some clarification, assessing the pros and cons of an AI tool is challenging until it undergoes extensive testing. With the technical preview, GitHub aims to accomplish precisely that.
“It’s hard to say what limitations developers will find with Copilot Workspace until they use it at scale, and that’s exactly why we do technical previews.”
Mobile compatibility, an advantage?
Copilot Workspace encourages exploration by allowing developers to edit, regenerate, or undo every part of their plan as they iterate to find the exact solution they need.
It also increases their confidence by providing developers with integrated tools to test and validate that the AI-generated code performs as expected.
Copilot Workspace also “boosts collaboration with automatic saved versions and context of previous changes so developers can immediately pick up where their teammate left off,” Carter noted.
Moreover, the tool is also mobile-compatible, which GitHub believes is a huge advantage for developers.
“Personally, I love taking walks in between meetings and I often find myself thinking through a new idea while I’m on-the-go. With Copilot Workspace on mobile, I can easily create a plan for bringing that idea to life, and even test and implement it, all from my phone.
“We’re also excited about Copilot Workspace on mobile because it allows developers to collaborate from wherever they may be. If a colleague sends me a link to their Workspace, I can explore and review it from my phone just as easily as I could from my computer,” Carter added.
Making developers efficient
Carter expects Copilot Workspace to provide immediate improvements for developers in terms of efficiency when “you consider how long it typically takes to read through an issue, explore related files, and put together an implementation plan without Copilot Workspace. What has historically taken hours can now be done in seconds.”
However, he considers the productivity improvements to be an incidental outcome of the broader advantages of Copilot Workspace in terms of enhancing clarity of thought, fostering exploration, and boosting confidence.
“For example, on my team at GitHub Next, I have front-end developers doing back-end work with the help of Copilot Workspace and vice-versa.
“Being able to tackle projects outside of your specific area of expertise confidently is a huge benefit, and when you think of doing this at scale you can imagine how much more productive developer teams can be,” Carter said.