India, Aug. 9 -- Artificial intelligence is becoming part of the software development lifecycle at a much deeper level. Developers are no longer using AI only to generate code or answer technical questions. Increasingly, AI agents can break a larger task into subtasks, use tools, make decisions within defined boundaries, and execute those steps with limited human intervention.

That changes the security equation, according to Naresh Agarwal, SVP, Engineering, India, at Harness. Once an agent can take action inside a software environment, it can no longer be treated as just a code generator. It has to be seen as an identifiable, controlled actor. The question is no longer simply whether the generated code is correct, but which agent perfor...