October 5, 2022

Robotic Notes

All technology News

Tabnine enhances developer collaboration on GitLab using custom AI code models

3 min read



Tabnine Ltd., the maker of an artificial intelligence code completion tool, today announced that it has integrated with DevOps lifecycle platform GitLab Inc. to bring enhanced collaboration for developer teams using custom AI models.

Using AI models, Tabnine provides predictive suggestions as developers type code based on current projects that can greatly speed up development sessions. By connecting to a private GitLab repository, Tabnine can automate away many of the repetitive coding tasks that developers do on a daily basis when coding, such as producing functions, writing common statements and forming syntax.

Nima Badiey, global vice president of alliances at GitLab, and Brandon Jung, vice president of ecosystems and business development at Tabnine, told SiliconANGLE the integration would help GitLab customers save time.

“Just think about the amount of time you waste on your phone correcting autocorrect, or how quickly you can write in Gmail with the suggestions on, the code is pretty much repeatable patterns and people tend to write from scratch what they could mostly automate,” he said. Badiey.

Most developers have one screen that’s their editor and one screen that’s the Git repository, and they really don’t want another screen. Nobody wants to have to switch out of their current task just to have to look up a function or a library or figure out what to type next. It’s useful to have an AI agent to automatically suggest the correct code that will come next and be able to trust it. Also, doing it all in one place really boosts developer productivity.

Those suggestions need to be powerful and they’re only as good as their model, Jung said. That’s where Tabnine’s AI model comes in. It’s trained off the private GitLab repository, using the best practices of the team, and learns from them to provide consistent automated code that helps speed productivity.

“Over the past six to eight months we’ve been working on custom models, which takes the strength of the big universal model but we can take your private code and build you a model that dramatically ups the security and quality of your code,” Jung said. “This is because, of course, you already know it’s a solid codebase. That’s hugely valuable, you’re not just taking some random open-source code and integrating it. So, you can bring new developers into your team and they’re getting the best suggestions from your own codebase. “

These AI custom models have other use cases as well, Jung said. For example, there are senior developers and junior coders throughout the DevOps pipeline who will be using the AI ​​suggestions at different levels. Senior coders will probably code extremely quickly based on their own knowledge and skillset. Junior coders may need to lean on the AI ​​suggestions a lot more.

“This integration is this: Use Tabnine, connect to your repository, we’ll build the model, we upload it, and every developer is off and running. Then they can use it as much or as little as they want, ”Jung said. “Your most senior developers are going to code really fast. They may not need to accept as much from Tabnine. However, Tabnine may still learn a lot and help the junior developers. ”

With that in mind, Jung said that AI models could be trained with codebases produced by senior developers, which would provide junior developers a leg up in coding consistency and skill development. Since the suggestions would come from code written by the senior developers, it would be very similar to working alongside the senior developers.

The integration between Tabnine and GitLab requires zero overhead from developers. Most repositories with enough data will have a working custom model within 24 hours.

The solution is available right now and users who are interested in connecting their GitLab repository can learn how to do it now. Tabnine supports more than 30 of the most popular programming languages ​​across 21 code editing environments as well.

Image: Tabnine

Show your support for our mission by joining our Cube Club and Cube Event Community of experts. Join the community that includes Amazon Web Services and Amazon.com CEO Andy Jassy, ​​Dell Technologies founder and CEO Michael Dell, Intel CEO Pat Gelsinger and many more luminaries and experts.



Source link