An AI agent should never be the Directly Responsible Individual for a project. That is the argument. The term DRI originates at Apple and names the single human accountable for the success or failure of a specific initiative. GitLab's public handbook offers the clearest written definition available.
The reasoning is not sentimental. Machines cannot take accountability. IBM said it plainly in a 1979 internal training slide: 'A computer can never be held accountable, therefore a computer must never make a management decision.' That slide is 46 years old and the principle has not expired.
The full post is short, but the question it raises is not. As LLM-powered agents absorb more operational work inside real organizations, the DRI framework forces a concrete answer to who owns the outcome when something goes wrong. Read it for the framing, not just the conclusion.
[READ ORIGINAL →]