Google has released the enterprise version of Gemini Code Assist. This latest version adds the ability to train on internal polices and source code. The product was announced at the Google Cloud Summit in London this week.

Gemini Code Assist offers AI-powered assistance for development teams building and deploying applications. It can be used in IDEs such as VS Code and JetBrains IDEs, with support for most popular languages.

Code Assist suggests code completions as developers write code, will generate full functions or code blocks from comments, generate unit tests, and get help with debugging and documenting your code.

Google initially launched Gemini (then called Bard) as an experiment in March 2023. Later incarnations added in a fine-tuned version of Google's Gemini Pro multimodal large language model, after which Bard was renamed as Gemini. 

The new enterprise edition of Code Assist adds features aimed at enterprise users, backed up by what Google calls "deep local codebase awareness."

Standard Gemini Code Assist provides contextualized responses to your prompts, including source citations regarding which documentation and code samples Gemini used to generate its responses. The Gemini large language models (LLMs) that are used by Gemini Code Assist are trained on datasets of publicly available code and Google Cloud-specific material.

In the Enterprise edition, Code Assist can analyze and understand your local codebase, and adds in code customization so you can have code suggestions based on your organization's best practices and internal libraries. Google says this allows Code Assist to generate customized code suggestions that are more accurate and relevant to your organization. The assistant can also help with complex tasks like upgrading a Java version in an entire repo.

Gemini Code Assist Enterprise code customization can index repos from GitHub.com and GitLab.com, and Google says support for additional source control systems and self-hosted, on-prem repos will be added in early 2025.

Organizations can specify if there are files that should not be indexed because of sensitive data, deprecated code, or other reasons such as auto-generated code. There's also the option to say which repositories specific developers should use based on IAM (Identity and Access Management) policies.

Data specific to an organization isn't used for general training of the Gemini model, according to Google, and source code stored for use in code customization is housed in a Google Cloud managed project and is isolated to each customer's organization. Customers also have full control over what source repos should be used for customization and can completely purge data at any time.

Alongside the additions to the general tool, Google has also announced it is building coding assistance into services across Google Cloud. Specifically, Code Assist Enterprise makes coding assistance available in Firebase, Databases, BigQuery, Colab Enterprise, Apigee, and Application Integration.

Gemini Code Assist Enterprise users of BigQuery can take advantage of SQL and Python code assistance, along with generation of pre-validated, ready-to-run queries, and a natural language-based interface for data exploration, curation, wrangling, analysis, and visualization.

In terms of cost, Gemini Code Assist Standard is available to try at no cost until November 8, 2024, limited to one user per billing account. Gemini Code Assist Enterprise is generally available and prices start at $45 per month per user, but is available for $19 per month per user on a 1-year subscription until March 31, 2025.

More Information

Gemini Code Assist Webpage

Related Articles

Google Adds Code Generation To Bard

Google Launches Developer Program

JetBrains Integrates Gemini Into AI Assistant

To be informed about new articles on I Programmer, sign up for our weekly newsletter, subscribe to the RSS feed and follow us on Twitter, Facebook or Linkedin.

Comments

Make a Comment or View Existing Comments Using Disqus

or email your comment to: comments@i-programmer.info