- Google introduces Code Assist, an AI-powered code completion tool, at its Cloud Next conference.
- Code Assist, previously known as Duet AI, offers enhanced features and compatibility with popular editors.
- It competes directly with GitHub’s Copilot Enterprise, boasting a million-token context window for more accurate suggestions.
- The tool can be customized for internal code bases and supports various hosting platforms, including on-premises and cloud-based repositories.
- Google partners with developer-centric companies to integrate knowledge bases into Code Assist.
- Alongside Code Assist, Google launches CodeGemma for code generation and assistance and Gemini Cloud Assist for optimizing cloud application lifecycles.
Main AI News:
During its Cloud Next event, Google introduced Gemini Code Assist, an AI-powered code completion and assistance tool tailored for enterprises.
If this sounds familiar, it’s because Google had previously offered a similar service under the now-defunct Duet AI branding. While that iteration became widely available in late 2023, Google had hinted at transitioning from its Codey model to Gemini in the near future. Code Assist marks both a rebranding of the older service and a significant upgrade.
Google Cloud showcased Code Assist at its 30,000-attendee conference in Las Vegas, presenting it as available through plug-ins for popular editors like VS Code and JetBrains.
Code Assist not only competes directly with GitHub’s Copilot Enterprise but also offers unique Google-specific features.
One such feature is its support for Gemini 1.5 Pro, renowned for its million-token context window, enabling Google’s tool to gather significantly more context than its competitors. According to Google, this results in more accurate code suggestions and the capability to analyze and modify substantial code segments.
Brad Calder, Google’s VP and GM for its cloud platform and technical infrastructure, elaborated on the significance of this upgrade in a press conference prior to Tuesday’s announcement: “This upgrade brings a massive 1 million-token context window, which is the largest in the industry. This allows customers to perform large-scale changes across your entire code base, enabling AI-assisted code transformations that were not possible before.”
Similar to GitHub Enterprise, Code Assist can be customized based on a company’s internal code base.
Kai Du, Director of Engineering and Head of Generative AI at Turing, noted the impact of code customization: “Code customization using RAG with Gemini Code Assist significantly increased the quality of Gemini’s assistance for our developers in terms of code completion and generation. With code customization in place, we are expecting a big increase in the overall code-acceptance rate.” This functionality is currently in preview.
Another distinguishing feature of Code Assist is its ability to support on-premises codebases, as well as those hosted on platforms like GitLab, GitHub, and Atlassian’s BitBucket, and even those distributed across different services—an offering not currently provided by its major competitors.
Google is further enhancing Code Assist by partnering with developer-centric companies to integrate their knowledge bases into Gemini. Stack Overflow and a host of other industry players are already part of this initiative.
However, the true measure of Code Assist’s success lies in its reception by developers and the utility of its suggestions. While Google’s support for various code repositories and its expansive context window are promising, ultimate success hinges on factors like latency and the quality of results compared to Copilot. Should it fail to surpass its competitors significantly, it risks meeting a fate similar to AWS’ CodeWhisperer, which struggled to gain traction.
In addition to Code Assist, Google also unveiled CodeGemma, a new open model within its Gemma family, specialized in code generation and assistance. CodeGemma is now accessible through Vertex AI.
Furthermore, Google announced Gemini Cloud Assist, which is designed to aid cloud teams in designing, operating, and optimizing their application lifecycle. This tool can generate architecture configurations tailored to a company’s requirements, diagnose issues, and optimize cloud usage for cost reduction or performance enhancement.
Cloud Assist will be accessible through a chat interface and integrated directly into several Google Cloud products.
Conclusion:
Google’s introduction of Code Assist signifies a formidable challenge to GitHub’s Copilot Enterprise in the code completion market. With enhanced features, including extensive context capabilities and support for diverse hosting platforms, Google aims to attract developers and enterprises seeking advanced AI-driven code assistance. This move not only intensifies competition but also underscores the growing importance of AI in software development tools. Companies operating in this market need to innovate continually to meet the evolving demands of developers and maintain their competitive edge.
