Cognition Helps Devin Test Its Own Work With Astra

Cognition is teaming up with GPT-6 Astra to strengthen Devin, its AI software engineer, by giving it the ability to test its own work. The goal is to help engineers spend less time reviewing code and more time shipping it. Under the collaboration, Astra is used to improve Devin's capacity to verify that the software it writes actually functions as intended. Self-testing agents address one of the biggest bottlenecks in AI-assisted development: generated code still needs verification. Even when a model produces plausible-looking code, someone or something has to confirm that it runs correctly, handles edge cases, and does not introduce regressions. If an agent can validate its own output, the review burden shifts from line-by-line inspection to higher-level judgment. Engineers can focus on architecture, intent, and product decisions rather than manually tracing through every change. The collaboration reflects a broader trend of pairing coding agents with evaluation and testing capabilities. As AI-generated code becomes more common, the ability to automatically check that code becomes just as important as the ability to write it. Without reliable verification, faster code generation simply moves the bottleneck downstream, creating larger review queues and more opportunities for subtle bugs to slip through. For Cognition, the partnership is a way to differentiate Devin in a crowded field of coding assistants. For OpenAI, it is another demonstration of how Astra can be applied to agentic workflows that involve planning, execution, and self-checking. If the approach works, it could change how development teams think about code review, turning it from a mandatory manual step into a higher-level quality gate supported by automated testing agents.