Tabnine Ltd., the maker of an artificial intelligence code completion tool for developers, today unveiled an enterprise-grade code-centric chatbot powered by generative AI that will allow developers to write, update and understand their code in new ways.

The new tool, named “Tabnine Chat,” uses Tabnine’s own in-house large language models and runs inside the company’s integrated development environment that professional software developers use to write and prototype code, which means that they don’t need to go anywhere to use it. That means it has the full context of their projects and can understand the nature of the work.

Tabnine Enterprise customers can also connect their repositories to the AI, which will allow it to answer questions about their internal projects and even write documentation.

The company’s flagship product is an AI-powered code completion tool for professional software developers in a development environment, which saves developers time and energy. The addition of a generative AI code-producing tool puts a whole new level of power into their hands, which allows developers to use text prompts to explain code, search repositories and generate new code.

For example, using Tabnine Chat, developers can gain a better understanding of the code that they’re looking at. Upon visiting a new block of code, a library or an application programming interface or some other complex piece of software, the developer can simply type: “Tell me about the intent of this piece of code.” From there the interface can produce a conversational explanation of what the code does, including insights into the code’s structure, how it operates, its endpoints and interfaces.

Beyond that, the Chat AI can produce entire new code blocks using natural language prompts such as “Create an app that reads the weather in London.” For developers who hate writing documentation, it can do that too, so there will be no more excuses for poorly commented code.

Tabnine Chat can be used to refactor code and identify potential bugs such as performance bottlenecks, areas that could use improvement or address other issues in overall software quality based on training the model has received. Since it’s an AI model, it also constantly learns from the company’s codebase and provides personalized recommendations that reinforce standardized coding practices across teams to ensure consistency and maintainability.

“As the tools developers use continue to rapidly evolve, organizations must ensure the solutions their teams rely on not only allow them to be more productive, but operate in a secure way,” said Tabnine Chief Technology Officer Eran Yahav. “Innovation at speed requires trust in the tools developers use, especially in enterprises.”

Most notably, Tabnine Chat is a secure and managed solution that is compatible with security and compliance requirements. Tabnine Enterprise customers can also deploy the coding chatbot in an isolated environment using a Virtual Private Cloud or on-premises. That will ensure an experience that’s entirely within their own security perimeter. Equally important, Tabnine Chat was trained on open-source code with permissive licenses, which means that it will not use code with restrictive licenses.

Without such a tool at their fingertips, developers might turn to a tool such as ChatGPT or Google Bard, which are also capable of generating and refactoring code. However, some organizations have banned their employees from using these tools because inputs given to them are integrated into their training models and proprietary code given to them can be leaked.

Tabnine Chat is available for free use in an early access beta release today and will be generally available to customers of the Pro and Enterprise plans beginning in Fall 2023.

Image: Pixabay