In a groundbreaking leap for artificial intelligence, meet Devin, the world’s first fully autonomous AI software engineer. Developed by Cognition, a leading applied AI lab, Devin is not just a tool but a tireless and skilled teammate, capable of working independently or collaboratively with human engineers.

Devin represents a significant advancement in AI technology, equipped with state-of-the-art capabilities that redefine the boundaries of software engineering. Unlike traditional AI tools, Devin can plan and execute complex engineering tasks autonomously, making thousands of decisions along the way. With its unparalleled ability to recall relevant context, learn from experience, and fix mistakes, Devin enables engineers to focus on more challenging problems while driving engineering teams towards ambitious goals.

Capabilities of Devin

Devin comes equipped with a comprehensive set of developer tools, including a shell, code editor, and browser, all within a sandboxed compute environment. This ensures that Devin has everything it needs to perform tasks just like a human engineer would. Moreover, Devin actively collaborates with users, providing real-time progress updates, accepting feedback, and working together through design choices when necessary.

Here’s a glimpse of what Devin can do:

• Learn to use unfamiliar technologies.
• Build and deploy apps end to end.
• Find and fix bugs in codebases autonomously.
• Train and fine-tune its own AI models.
• Address bugs and feature requests in open-source repositories.
• Contribute to mature production repositories.

Performance Evaluation

Devin’s performance was evaluated on the SWE-bench benchmark, a challenging test that requires resolving real-world GitHub issues found in open-source projects like Django and scikit-learn. Impressively, Devin correctly resolves 13.86% of the issues end-to-end, far surpassing the previous state-of-the-art performance of 1.96%. Even when provided with the exact files to edit, previous models could only resolve 4.80% of issues.

About Cognition

Cognition is an applied AI lab focused on reasoning, aiming to build AI teammates with capabilities beyond existing tools. Their vision extends far beyond code, seeking to unlock new possibilities across various disciplines. Backed by significant funding, including a $21 million Series A led by Founders Fund, Cognition is poised to shape the future of AI and engineering.

Stay tuned for a more detailed technical report on Devin’s capabilities. To learn more about Cognition and their groundbreaking work, visit their website.