See It Work — Capability is the agent's; accountability is always a human's | Book 03 · Leading AI Agents

It's tempting to divide human-vs-agent roles by capability — give the agent whatever it can do. Wrong axis. The real division is accountability: an agent might be fully capable of a task, but a human still owns the result. Capability is the agent's; accountability stays human, always.

Confuse capability with accountability and you end up with outcomes no one owns — the exact gap where things go wrong and no one's responsible. Anchoring accountability to a human closes it.

What this means for you

Roles divide by accountability, not capability — a human always owns the outcome. What this means for you: you never end up with results no one is responsible for; no matter how capable the agent, a named human answers for the outcome, which is what keeps an agent-augmented team accountable instead of diffuse.

The dividing line is accountability, not capability:

Role Definition

not aboutcapability

aboutaccountability

the agentmay be capable

a humanowns the outcome

The categories are not about capability. They are about accountability. An agent might be capable — but accountability rests with a human.

For the technical reader — the command, and how to verify it yourself

# one line · you do not need to run this
see walkthrough

see walkthrough
# -> work divided by accountability, with a human owning every outcome

Full step-by-step is in Appendix RX: Hands-On Demonstrations in the book.