GitHub Copilot Workspace Moves Beyond Code Generation

The killer app of generative AI is getting better

GitHub announced Copilot Workspace last week and materially expanded the impact of generative AI in the software development lifecycle (SDLC). While GitHub Copilot has materially transformed software development via code generation and code reviews, Workspace extends features into planning, testing, and other SDLC steps. GitHub CEO Thomas Dohmke commented in a blog post:

In 2022, we launched GitHub Copilot as an autocomplete pair programmer in the editor, boosting developer productivity by up to 55%. Copilot is now the most widely adopted AI developer tool. In 2023, we released GitHub Copilot Chat—unlocking the power of natural language in coding, debugging, and testing—allowing developers to converse with their code in real time.

After sharing an early glimpse at GitHub Universe last year, today, we are reimagining the nature of the developer experience itself with the technical preview of GitHub Copilot Workspace: the Copilot-native developer environment. Within Copilot Workspace, developers can now brainstorm, plan, build, test, and run code in natural language…

Copilot Workspace represents a radically new way of building software with natural language, and is expressly designed to deliver–not replace–developer creativity

Copilot Chat

The core capability behind Workspace is Copilot Chat. It is the natural language interface introduced in beta in late 2023. Early users employed it for common coding activities such as code generation, debugging, and summarization of code functions. It is also central to Workspace as the natural language interface is employed in setting up new projects and executing other non-coding tasks.

As an aside, I recently spoke with a software development executive at a leading U.S. bank that has been using Copilot and Copilot Chat. His teams gave Chat rave reviews for its initial functionality, which predated the new Workspace features.

There has been talk from NVIDIA CEO Jensen Huang and others that natural language interfaces will enable anyone to become a coder. That’s not entirely true. The more complex the problem, the more necessary it is to have an experienced developer who knows what to ask for. The biggest impact of code generation tools is raising the floor of performance and efficiency among junior developers and making the best coders more productive by offloading simple tasks.

Workspace

Those simple tasks range from setting up a project to planning, coding, testing, debugging, pull request summarization, and collaborating, among others. The documentation also lays out the guiding principles:

Copilot Workspace is built on a set of principles that guide its design and development:

• Copilot Workspace is contextual. It is deeply integrated with GitHub, and is aware of the context of your task — the repository, the issue, the pull request.

• Copilot Workspace is assistive. It offers a canvas for you to navigate unfamiliar tasks, augmenting your development skills with a new kind of AI assistance.

• Copilot Workspace is pervasive. It is ready and waiting for you, available on every issue in every enabled repository on GitHub. And Copilot Workspace is even there for you when starting new code, available on every template repository, to create new software using natural language.

• Copilot Workspace is iterative. Copilot Workspace encourages you to check, review, refine and iterate on AI-generated outputs. You, the developer, are in control.

• Copilot Workspace is collaborative. You can share sessions with your team and publish links to your sessions on issues and pull requests. And, if you're a repository maintainer, we give you controls to help manage the use of AI-assisted development with your repositories.

• Copilot Workspace is configurable. You can integrate Copilot Workspace into your workflows via deep links to Copilot Workspace that specify common tasks.

Contextual and assistive principles are expected characteristics. However, pervasive, collaborative, and configurable are not. The pervasive principle leverages GitHub’s unique scope of features that reach back to code repositories, versioning, and design. Collaborative in this context also leans on GitHub features for sharing code, designs, issues, and plans. Configurable relates to integrating Copilot into workflows as opposed to simply being a tools that is used within workflow tasks.

GitHub is a widely used developer product. GitHub Copilot was the first generative AI coding assistant built upon OpenAI’s Codex foundation model. This also made it the first coding assistant many developers were exposed to. Copilot’s early market availability and attachment to a trusted developer software tool has surely reduced potentialy anxiety developers may have about adopting a new tool that is the self-driving car of coding.

However, Google’s Codey and Amazon CodeWhisperer were not just late to market. They are not integrated into key elements of the SDLC in the same way as Copilot. This creates an opportunity for GitHub. It has sales advantages because the company can sell to a large, loyal, cross-cloud, and cross-platform user base. It also natively serves in multiple SDLC steps and tasks. This is likely to make it harder for Google, Amazon, and the numerous startups looking to displace GitHub Copilot as the leading segment developer.

Generative AI’s First Killer App

Synthedia wrote in November 2023 that GitHub Copilot was generative AI’s first true killer app. While LLM-enabled text generators get most of the attention, code generation has already transformed software development across numerous enterprises and startups.

Software development is expensive and time-consuming. In addition, developers are often in short supply. It is easy to see why so many companies have adopted GitHub Copilot. A company executive said more than 50,000 organizations are now using Copilot and there are more than 1.8 million users. That latest paid user figure represents 80% growth over about two quarters.

GitHub Copilot Workspace is a logical feature expansion for the solution. It also shows more about what competitors are up against in battling against Copilot’s early market share lead.

• Dabble Lab: AI-powered bespoke software development services

• Loccus.ai: AI for protecting against voice fraud