Generative Data Intelligence

GitHub Copilot Enterprise reaches general availability

Date:

GitHub on Tuesday made Copilot Enterprise generally available, hoping to sell corporate developers on automated coding assistance.

GitHub Copilot, a generative software autocomplete and suggestion service built on OpenAI technology, saw commercial release in June 2022 as a $10 per month subscription. The next tier up, Copilot for Business, arrived in February 2023 for $19 per month.

Shortly thereafter, in September 2023, GitHub parent Microsoft said it would – under specific conditions – indemnify customers accused of copyright infringement as a result of using Copilot – a charge Microsoft and GitHub are presently fighting in court over claims that Copilot cribs code from humans.

Now, with Microsoft having subsequently doubled down and converted its Copilot Copyright Commitment into a broader Customer Copyright Commitment that extends to its AI services, large organizations looking to pay a bit more for all the bells and whistles can provide employees with AI-powered programming assistance, with some legal assurance, for just $39 per month.

Copilot Enterprise provides several features not available to lesser chatbot tiers. First, it supports repository-based semantic search, to make it easier to scour repos. Second, it can analyze pull request diffs – code differences – to help developers understand proposed changes.

Third, it can access internal knowledge bases which makes its suggestions more comprehensive. And fourth, Copilot Chat supports Bing integration, so developers can search more effectively for software-related information.

At some point, Copilot Enterprise will also support fine-tuned models, so organizations can train a private OpenAI GPT-4-based helper with the ins and outs of their own codebases to improve its predictive powers.

Apart from the fact that a GitHub Enterprise Cloud account is also required, GitHub Copilot Enterprise’s pricing is comparable to the Enterprise tier from rival AI assistant service Tabnine.

How it works: using public code as a model for AI-assisted development

How it works, according to GitHub: Using public code as a model for AI-assisted development … Click to enlarge

“Just by integrating generative AI into the editor, GitHub Copilot has quickly defined a new age of software development, resulting in clear gains of developer productivity and happiness,” said Thomas Dohmke, CEO of GitHub, in a statement.

“Today, we are bringing the next frontier of developer tools with the general availability of GitHub Copilot Enterprise – a companion that places the institutional knowledge of your organization at your developers fingertips.”

To support his claims about productivity, Dohmke cites a six-month study of 450 Accenture developers conducted in conjunction with GitHub. The study found, among other things, that 90 percent of developers reported writing better code with the help of Copilot.

What’s more, it’s claimed that 88 percent of the code suggested by Copilot was retained and 50 percent more builds were made.

Dohmke’s post also includes supportive quotes from the likes of Shopify and Figma to suggest it’s not just friendly consultancies that find generative code assistance helpful. Indeed, unaffiliated organizations, such as Australia and New Zealand Banking Group (ANZ Bank), have also reported generally positive experiences using GitHub Copilot. ®

spot_img

Latest Intelligence

spot_img

Chat with us

Hi there! How can I help you?