Tabnine has recently announced the beta of Tabnine Chat to interact with Tabnine's AI models using natural language. The chat application works inside the IDE, allows organizations to train on permissive code only, and can run on isolated environment deployment.
The new chat functionality uses Tabnine's large language models and expands the AI assistant beyond code generation, explaining existing code, searching code repositories, and generating new code according to natural language specifications.
According to Tabnite, the chat application supports strict security and compliance enterprise requirements, managing private codebases, permissible open-source code, and Stack Overflow questions and answers. Dror Weiss, founder and CEO at Tabnine, explains:
Tabnine Enterprise also allows isolated environment deployment using Virtual Private Cloud or on-premises deployment, ensuring total code privacy and security. In addition, Tabnine Chat was only trained on open-source code with permissive licenses, guaranteeing that our models aren't trained on GPL or other copyleft code.
By connecting internal repositories to Tabnine Chat, organizations with a large set of internal APIs, libraries, services, and best practices can improve productivity. Weiss highlights how the new chat can help knowledge proliferation:
Large organizations can leverage Tabnine Chat to disseminate coding expertise, best practices, and lessons learned from the extensive codebase as well as data sources like Jira, Notion, documentation, etc. This makes knowledge more easily accessible to all developers and enables new team members to ramp up far more quickly.
The chat application can also help developers by providing insights into the code's structure, intent, and performance. Other expected use cases are refactoring and extending code with more capabilities, for example, logging or generating documentation (docstrings) for specific sections of code. The team released a video to show some of the new chat features.
The frontend application is a React application embedded in a webview, currently available in VS Code and JetBrains IDEs, and supports all the popular programming languages. Amir Bilu, software engineer at Tabnine, shows instead in an article how to integrate Tabnine Chat for Neovim using Tauri.
Tabnine is a code completion assistant that utilizes generative AI to predict and suggest the next lines of code. In the last few months, the AI company introduced unit test generation, support for Eclipse, and Single Sign-On (SSO) for self-hosted deployments.
Tabnine Chat will be available to Tabnine Enterprise and Pro users only, with Pro plans currently starting at 12 USD per month per user. To join the waiting list for the private beta of Tabnine Chat, developers have to fill out a form.
Tabnine is not the only company offering an AI "pair programming" tool, with GitHub Copilot and Amazon CodeWhisperer other popular IDE-based options.