Agentic Software Engineering — AI × software engineering, made practical.
Code review works because the correction sticks between iterations — an agent forgets everything the second the session ends, unless you know exactly where to put the feedback instead.