Google recently announced that its enterprise-level code assistant Code Assist has been upgraded to Gemini 2.0 and expanded the external data sources that can be connected, such as GitLab, GitHub, Google Docs, etc. This move aims to provide developers with more powerful code assistance functions, improve development efficiency and simplify workflow. Gemini 2.0 gives Code Assist a larger contextual window, allowing it to better understand large code bases and achieve seamless integration through direct integration into developers' commonly used IDEs.
Google recently announced that its enterprise-level code assistant Code Assist has been upgraded to Gemini2.0 and expanded the external data sources that can be connected. This new version will provide developers with a larger contextual window to better understand large code bases in the enterprise.
According to Ryan Salva, senior director of product management at Google Cloud, Code Assist can now connect to a variety of data sources, including GitLab, GitHub, Google Docs, Sentry.io, Atlassian and Snyk. Developers can request help from Code Assist directly within their integrated development environment (IDE) without interrupting their workflow. Previously, Code Assist only supported integration with VS Code and JetBrains.
Code Assist, formerly known as Duet AI, was first launched last October. With the increasing demand of enterprises to simplify coding projects, AI coding platforms such as GitHub C opilot have also gained widespread attention. Code Assist adds enterprise-grade security and legal protection with its Enterprise Edition release.
Salva emphasized that connecting Code Assist to other tools developers use daily can provide more context for their work without the need to frequently switch windows. "Developers may use multiple tools such as GitHub, Atlassian Jira, DataDog, Snyk, etc. throughout the day, and we want to enable them to introduce these additional contexts into the IDE," he said.
Developers can simply open Code Assist's chat window and ask for the latest comments about a specific issue or the latest pull request in the code base. Code Assist will automatically query data sources and bring relevant information back to the IDE, helping developers work more efficiently.
AI coding assistant is one of the early important applications of generative AI. Since software developers started using ChatGPT to assist with coding, a number of enterprise-oriented coding assistants have been launched on the market. GitHub released C opilot Enterprise in February this year, and Oracle also launched Java and SQL coding assistants. In addition, the coding assistant launched by Harness is also based on Gemini and can provide suggestions in real time.
It is worth mentioning that although Code Assist already supports Gemini2.0, it is still independent from Jules, a new tool launched by Google. Jules is one of several experiments launched by Google Labs teams to demonstrate how autonomous or semi-autonomous agents can be used to automate the coding process, Salva said. Although Code Assist is currently the only enterprise-level coding tool based on Gemini, Jules may incorporate similar functionality in the future.
Currently, feedback from early users on Code Assist and Jules shows that Gemini 2.0 has significantly improved response speed. Salva pointed out that during the coding process, rapid feedback is crucial for developers, and any delay will interrupt their train of thought.
Looking ahead, while coding assistants will remain critical to the growth of generative AI, Salva believes the way companies develop code generation models and applications is likely to change in the coming years. He mentioned that Google’s DevOps research and evaluation team’s 2024 Accelerated Development State Report showed that 39% of respondents expressed distrust in AI-generated code, while the quality of documentation and delivery has also declined.
Highlights:
Code Assist has now been upgraded to Gemini2.0 and has added connections to multiple data sources.
Developers can use Code Assist directly in the IDE to obtain relevant contextual information and improve work efficiency.
In the future, the development of AI coding assistants may focus more on the quality of code generation rather than simply improving work efficiency.
All in all, the upgrade of Code Assist is an important advancement for generative AI in the field of enterprise-level code assistance, but it also reminds us to pay attention to the improvement of AI code generation quality and avoid purely pursuing efficiency while ignoring code reliability and maintainability. In the future, the development direction of AI coding assistants will focus more on improving code quality and developer trust.