GitLab's AI-assisted Code Suggestions are a tool designed to improve coding productivity and accelerate software development. It provides AI-powered generative code suggestions, allowing you to complete entire lines of code with a single keystroke, quickly start functions, complete code templates, and generate tests, among other tasks.
The tool prioritizes privacy and data security, as it keeps proprietary source code protected within GitLab's enterprise cloud infrastructure and does not use code as training data.
Source code inference against the Code Suggestions model is not retained. Code Suggestions use pre-trained open-source models that are continuously fine-tuned with a custom open-source code dataset.
This allows for support of multi-language coding and additional use cases. Currently, it supports 13 programming languages, including C/C++, C#, Go, Java, JavaScript, Python, PHP, Ruby, Rust, Scala, Kotlin, and TypeScript. GitLab has future plans to expand the capabilities of the tool.
They are working to provide Code Suggestions for self-hosted instances through a secure connection to GitLab.com. Additionally, they intend to improve the user experience by offering support for additional IDEs, such as JetBrains IntelliJ-based IDEs and Visual Studio.
They also have plans to enhance how suggestions are presented and accepted within IDEs, giving developers more control over the feature. Overall, GitLab's AI-assisted Code Suggestions tool aims to empower developers by providing intelligent code suggestions, optimizing coding efficiency, and ensuring privacy and data security.
