AI & RoboticsNews

Kite expands its AI code completions from 2 to 13 programming languages

Kite, which suggests code snippets for developers in real time, today added support for 11 more programming languages, bringing its total to 13. In addition to Python and JavaScript, Kite’s AI-powered code completions now support TypeScript, Java, HTML, CSS, Go, C, C#, C++, Objective C, Kotlin, and Scala. (The team chose the 11 languages by triangulating the StackOverflow developer survey, Redmonk’s language rankings, and its own developer submissions.)

AI that helps developers is growing in popularity, with startups like DeepCode offering AI-powered code reviews and tech giants like Microsoft trying to apply AI to the entire application developer cycle. Kite stands out from the pack with 350,000 monthly developers using its AI developer tool.

Kite debuted privately in April 2016 before publicly launching a cloud-powered developer sidekick in March 2017. The company raised $17 million in January 2019 and ditched the cloud to run its free offering locally. In May, Kite added JavaScript support, launched a Pro plan with advanced line-of-code completions for Python, and updated its engine to use deep learning, a type of machine learning.

Scaling ML for more programming languages

Kite CEO Adam Smith previously founded Xobni, an email service launched in September 2007 that Yahoo acquired in July 2013. We asked Smith how his team, which has now grown to 18 employees, expanded the number of supported languages so quickly. After all, it took years for Kite to add a second language, and then only five months to add 11 more.

“When we built support for JavaScript earlier this year, we focused on creating a scalable way to add more programming languages,” Smith told VentureBeat. “We’ve been iterating on our JavaScript models and ranking/sorting algorithms since then. This learning has allowed us to release each new language relatively easily, with the same accuracy and intelligence as JavaScript.”

Kite’s Python machine learning model is trained on 25 million open source code files, and its JavaScript machine model is trained on 30 million files. Each of the 11 new languages, however, is trained on just 12 million files.

“For this new release, we built a more sophisticated crawling engine that orders code by popularity,” Smith told VentureBeat. “We no longer include less popular code in our model training runs, which makes them less noisy. These models are even more optimized for efficient usage of users’ CPU and memory than our previous models. Note that Python was built in a different way from the other languages. Python has our richest feature set, including documentation and function signatures, which is one reason why we’ve chosen to only monetize our Python product.”

What’s next for Kite

For Python, Kite Free includes completions ranked by relevance, local code processing, documentation as you type, and function signatures as you type. While the free version had previously included Line-of-Code Completions, that feature, along with Multi-Line Completions, became part of Kite Pro with its launch in May.

Something similar is likely to happen to the other languages, but “for now” it’s just Python. Smith hired Kite’s first salesperson a few weeks ago. The goal? To sell Kite Team Server, which offers GPU-powered completions personalized to a company’s codebase, directly to enterprises. In other words, developers shouldn’t worry about the free version of Kite going away anytime soon.

Furthermore, Kite isn’t done adding languages. “We’re hoping to release a few more languages in the next couple weeks, including PHP, Ruby, and Shell,” Smith told VentureBeat.

In addition to the growing number of programming languages, Kite is also available in 16 editors. It supports Android Studio, Atom, JupyterLab, Spyder, Sublime Text, VS Code, and Vim. Kite also works with the IntelliJ family: IntelliJ, PyCharm, WebStorm, Goland, CLion, PHPStorm, Rider, RubyMine, and AppCode.

Here too, Kite plans to add support for more editors and IDEs “over the coming months,” Smith said. “For example, we added C++ support but don’t have a Visual Studio integration yet. The goal is to allow as many developers as possible to take advantage of Kite, no matter their language or IDE.”

Finally, Smith teased that Kite is building features “that can be used by all developers, regardless of programming language.” The next such feature debuts in November.


The audio problem:

Learn how new cloud-based API solutions are solving imperfect, frustrating audio in video conferences. Access here



Author: Emil Protalinski
Source: Venturebeat

Related posts
AI & RoboticsNews

H2O.ai improves AI agent accuracy with predictive models

AI & RoboticsNews

Microsoft’s AI agents: 4 insights that could reshape the enterprise landscape

AI & RoboticsNews

Nvidia accelerates Google quantum AI design with quantum physics simulation

DefenseNews

Marine Corps F-35C notches first overseas combat strike

Sign up for our Newsletter and
stay informed!