The First AI Software Engineer Devin Emerges, Officially Joining the Programming Ranks
-
A startup named Cognition recently launched an AI assistant called Devin, designed to assist software engineering teams with coding and other development tasks. Unlike existing AI coding assistants, Devin can be programmed to execute end-to-end software projects, including deploying applications, fixing bugs, and learning new technologies, while humans take on supervisory and guiding roles. Devin can execute multi-step workflows based on user requirements while maintaining progress. Engineers can monitor its progress in real-time and intervene to correct errors when necessary. This allows teams to delegate some tasks to the AI assistant and focus on more creative work.
Devin's performance in SWE benchmark tests According to the demonstration, Devin can handle multiple tasks including deploying websites, debugging code, generating steganographic images, and training computer vision models. In software engineering benchmark tests, it independently completed 13.86% of cases, significantly outperforming other large language models.
Although technical details haven't been disclosed, Cognition stated that Devin stems from their long-term research advancements in reasoning and planning. Currently in closed beta, interested users can apply for early access. Wider availability may follow in the future.
Cognition hinted that coding is just the beginning, suggesting their AI assistant could expand into more domains. The company plans to leverage AI's cross-domain reasoning advantages to empower multiple industries. The emergence of Devin brings a new artificial intelligence collaboration experience to software developers. By delegating tedious tasks to the AI system, engineers can focus on innovative work, potentially enhancing productivity. However, this technology is still in its early stages, and its maturity and effectiveness require further evaluation by the industry.