GitHub becoming the AI-powered developer platform
Hello there 🖐️,
Broadcasting from GitHub Universe, I’m excited to share insights into the AI-infused features that GitHub has unveiled, aiming to transform it into the ultimate AI-powered developer platform.
Copilot’s new API horizon
The new Copilot API empowers you to build apps that interact directly with Copilot—think of developing a custom analytics tool that for example, queries for optimizing database queries or integrating with Postman to test API calls seamlessly.
This is a great extension point for me, as you can add your own systems to the chat interface. Besides using the existing addons, like Datastax, LaunchDarkly, Postman, Hashicorp, and Datadog, it opens up a world of possibilities.
I saw an interesting demo showing Copilot asking LaunchDarkly if the selected code was actually feature-enabled or not.
First-party addon support
Creating GitHub issues is now a breeze within the Copilot chat interface. As you discuss code, you can generate issues with all the necessary details included, streamlining project tracking and management. This shows one of the first-party integrations now possible with Copilot chat.
Copilot Enterprise
Imagine having an AI that not only reviews code but also generates documentation and summarizes pull requests tailored to your organization’s style—this is the promise of Copilot Enterprise. It’s like having an extra team member versed in your company’s codebase and can swiftly help onboard new developers.
The new Enterprise tier will be available in February, costing 39 dollars per user per month. I believe you will earn that back quickly.
Personalizing Copilot with your code
Training Copilot on your data means it can understand your proprietary frameworks and coding styles. It’s akin to customizing a high-performance car to your driving preferences—ensuring that Copilot drives your projects forward most efficiently.
By default, Copilot is trained on public open-source data, but if you use something more exotic, you might run into less ideal suggestions. Adding your own complete repository contents can give you better results.
You might have a repository containing coding standards or preferred framework setups, so it would be ideal to add these to the Copilot chat as well to get more relevant results.
JetBrains welcomes Copilot Chat
For those using JetBrains IDEs like Rider, Copilot Chat is now also available, ready to assist with coding queries directly in your preferred environment, bringing AI assistance into the comfort of your established workflow.
No longer are you confined to Visual Studio (Code) for the chat functionality. Copilot itself was already available in the JetBrains IDEs, but the chat part was not.
You can sign up for the private beta.
Copilot Chat anytime, anywhere
The introduction of Copilot Chat in the GitHub mobile apps means support is always at your fingertips, whether you’re reviewing code on the go or brainstorming solutions outside the office. Besides typing, you can also speak to the Copilot and interact with it.
GPT-4: Elevating Copilot’s intelligence
With GPT-4, interacting with Copilot becomes more intuitive. Select a code snippet, and you can start an in-depth conversation about it. It’s like discussing code with a colleague who has an encyclopedic knowledge of programming languages and patterns.
GitHub.com welcomes Copilot Chat
Initiate Copilot Chat directly from code views on GitHub.com. Just click on the top button called Chat with Copilot to start a new conversation.
The new Chat experience from the home page
Need to understand a complex function? Copilot Chat is there to offer explanations and insights right where you need them.
Ask Copilot about code
AI-Generated PR descriptions
Crafting detailed PR descriptions can be time-consuming. Copilot can now automate this, allowing you to focus on coding rather than documentation.
Generated AI summary of a Pull Request
Proactive security with Copilot
When GitHub Advanced Security flags a vulnerability, Copilot is ready to suggest fixes, acting as a vigilant guardian over your code’s security.
It does this already in your Pull Request, ensuring vulnerabilities are less likely to reach the main branch.
Secret Scanning: the AI sentinel
The AI-powered secret scanning tool is your lookout for unintentional sensitive data exposure, far beyond basic pattern matching—it’s the security net you always needed but never had.
Regex Made easy with AI
The AI Regex Assistant turns the often-dreaded task of crafting regular expressions into a conversation, simplifying one of the more intricate aspects of coding.
Copilot Workspace: A Visionary Leap
Imagine starting with a plan and ending with a fully implemented system, all orchestrated through Copilot Workspace. It’s a glimpse into a future where AI bridges the gap from concept to code, accelerating the realization of your creative visions.
For more information about Copilot Workspace, have a look at the video.
Links
To sign up for the previews: https://github.com/features/preview
To read the official blog post: https://github.blog/2023-11-08-universe-2023-copilot-transforms-github-into-the-ai-powered-developer-platform/
Access to the JetBrains Copilot chat beta: https://github.com/github-copilot/chat_jetbrains_waitlist_signup/join
Conclusion
As GitHub strides towards becoming an AI-powered hub, these features represent the building blocks of a more intuitive, efficient, and secure development process. Stay tuned as we continue to explore and leverage these advancements together.
Warm regards,
Michiel