Tabnine Ltd., the maker of an artificial intelligence code completion tool for developers, today announced the general availability of two AI agents that integrate with Atlassian Corp.’s Jira, which can generate and validate code from issues.
Jira issues are a primary source for developers to receive communication about updates needed for code and are often provided in natural language. Issues represent individual work items passed between collaborators with information about the various work to be done. An issue could be a story, a bug, a task or a subtask necessary for doing the developer’s job.
For example, a project manager could generate a Jira issue for a developer to update an existing function to include additional features such as adding the ability to display Celsius in addition to Fahrenheit for a temperature display on a weather app. The issue would be written up as a simple request to add the temperature.
First previewed in May, Tabnine’s Jira Implementation Agent provides a full “issue to code” capability that allows the developer to now cite the above Jira issue and have the agent generate the code changes. Once the code changes have been suggested, the developer can review the modifications to make sure they look good and okay the changes.
The agent also uses context from both the title and the content of the issue, as well as the local project code, to understand how to implement the changes to the developer’s code to provide the most accurate modifications, the company said.
After this is completed, developers can use the second AI agent, the Jira Validation Agent, to make sure that the generated code matches the requirements outlined in the issue. This agent will read the specifications written up in the issue and compare them to the code changes. If they match, it will explain how well they fit. If they don’t, it will make a code-change recommendation that the developer can accept or reject.
The Validation Agent will work any code written by a human or AI, making it a great way to receive instant feedback to reduce back-and-forth with peers. It can be a real-time saver if it can catch even a minor missed detail by making certain the code matches all the requirements.
“By offering developers not just best-in-class AI software development tools, but a platform that is CIO- and CISO-approved for its privacy and company-specific personalization, Tabnine is helping mature engineering teams build better apps faster and with more confidence,” said Brandon Jung, vice president of ecosystem at Tabnine.
These new AI tools join a growing stable of agents that Tabnine has released, including an AI test agent that can generate comprehensive test plans with detailed test cases, functions and methods, and an Onboarding Agent that can help developers get up to speed with a new project faster.