Gemini Code Assist, Google’s AI coding assistant, has gained a new “agent” feature in preview.
At its next cloud meeting on Wednesday, Google said Code Assist will be able to deploy new AI “agents” that allow Code Assist to take multiple steps to accomplish complex programming tasks. These agents can create applications from product specifications in Google Docs, for example, and perform code conversions from one language to another. Code Assist is now available in Android Studio, in addition to other coding environments.
Code assist upgrades are likely to occur in response to competitive pressures from rivals such as Github Copilot, Cursor and Cognition Labs, who are the creators of the virus programming tool Devin. The AI Coding Assistant Market has expanded sharply by the moon and has a lot of money. That’s not surprising, considering the potential to increase technology productivity.
Code Assist agents can be managed from the new Gemini Code Assist Kanban board, but they can generate work plans and report gradual progress to job requests. Beyond software generation and migration code, agents can implement new app features, perform code reviews, and generate unit tests and documentation, the company claims.
However, it is unclear whether Code Assist can do all this. Research finds that even today’s best code generation AI tends to introduce security vulnerabilities and bugs due to weaknesses in areas such as the ability to understand programming logic. One of Devin’s recent reviews found that only three of the 20 tasks completed successfully.
So, if you tap Code Assist to create or refactor your app, it’s not harmful to check your code yourself just to be safe.