Tabnine is updating an artificial intelligence-based coding assistant for developers

TabninIntroduced a new, time-saving version of the artificial intelligence assistant for software developers who can predict how code will be written for specific programs and fill in the blanks with the appropriate fragments. software programming interface June 15.

The company, which uses the Freemium-plus-service provisioning business model, also said that by 2022, the API will have surpassed the 1 million user mark.

Tel Aviv-based Tabnine’s software allows Python, JavaScript, Java, C #, TypeScript, and other languages ​​available today in both Visual Studio Code and IntelliJ IDEs to find reusable code in 31 languages. has, CEO Dror Weiss told VentureBeat.

“Our product is very simple to use,” Weiss said. “Your developer writes code in the IDE and has a top line installed as part of its editor. We look at the generated code in real time, and in context we query the AI ​​engine in real time; then we offer parts such as large next lines or fragments or a block of code needed by the compiler. The application makes a really sudden decision about whether they intend to do it, and they accept the code. This happens mostly in the process with mostly zero changes. ”

“Ninety percent of all codes are already written”

Ninety percent of all code is already written, and there’s no need to create it from scratch, Weiss said.

Weiss said developers now see that 30% -40% of their code is automatically generated using the Tabnine API without the need for deviations. “In my favorite editor, no code is changed from the standard workflow of coding, and it works for all popular ideas or code editors and all popular programming languages,” he said.

Over the past year, Weiss said, Tabnine has updated both AI models and an approach to the integration of new models into the platform. Weiss said that unlike the monolithic approach, which requires a single AI model, Tabnine is now able to support any suitable AI model to guide developers.

From infrastructure-code to sustainable integration, companies that have implemented devops in recent years have seen improved productivity in software development and release processes. Weiss said these productivity benefits usually stopped at the developer’s keyboard, as the explosion of frames, tools, languages, and APIs made the work of just one developer difficult.

Artificial intelligence has great potential to help the designer’s productivity

Weiss said CI has the potential to do what it did for the graduation process for developmental productivity.

“Every day, we receive amazing feedback from developers and teams that Tabnine’s artificial intelligence is now part of their flow and gives them time to focus on more valuable activities,” Weiss told the media consultant. “With longer and better code completions, our next-generation AI technology is a big step forward for developer productivity, establishing Tabnine as a leader in the development of artificial intelligence software.”

New generation platform a with them new AI models, each taught from scratch based on the code (compared to models pre-trained on the text and re-equipped to learn the code). Weiss said each model is optimized for a specific language or domain, which allows Tabnine to complete complete lines and pieces of code in context with unprecedented accuracy.

Launched in 2017, Tabnine has also raised $ 15.5 million from Qualcomm Ventures, OurCrowd and Samsung NEXT Ventures with the participation of existing investors, including Khosla Ventures, Headline Ventures, Hetz Ventures and TPY Capital. With this latest capital inflow, Tabnine has raised a total of $ 32 million so far.

Tabnine claims to be the only software development tool on the market that works with this type of artificial intelligence, Weiss said.

The latest funding will be used to add support to additional AI models and further accelerate the development of the Tabnine platform. programming languages Weiss said later this year.