AI researchers unveil an AI agent that can program better than GPT-4, putting the open-source project in competition with offerings like the recently launched Devin AI programmer.
The Princeton NLP team has developed SWE-agent, an open-source system that converts language models like GPT-4 into software engineering agents. These agents can fix bugs and problems in real-world GitHub repositories. With 12.29% of issues solved in the SWE-Bench test set, SWE-agent nearly matches the recently released commercial Devin (13.86%).
The result is a special Agent Computer Interface (ACI) that allows the language model to browse the repository, view, edit, and execute code files. The ACI development includes features such as a linter for syntax checking, a special file viewer, and directory browsing.
Cognition AI raised 21 million US dollars for Devin
SWE-agent plays a similar role to Devin, an AI software developer developed by AI startup Cognition AI. Devin can collaborate with human developers, perform tasks independently, and submit them for review. Devin is also designed to handle new, unknown libraries, program entire applications, find bugs in code bases, and process bug reports and feature requests in open-source repositories.
Unlike SWE-agent, Devin is not yet publicly available and has only been made available to select developers via a waiting list. Cognition AI recently closed a $21 million Series A funding round.