Yesterday we hired for IQ and EQ.
Every insurance role — underwriter, claims analyst, agent, actuary — was scored on intelligence and empathy. Judgment plus care. That framing held for a century. It stops holding this decade.
AQ is the third quotient.
Agentic Quotient — the fluency to design, direct and trust AI collaborators. It's not prompt engineering. It's operating literacy: knowing which decisions belong to a human, which to an agent, and where the handoff lives.
RPA asked the wrong question.
RPA asked how do we speed this up? Often, the answer was you shouldn't have been doing it at all. We accelerated broken processes and called it transformation.
Agentic AI asks the right one.
Now we ask three questions of every task: what should be done by bots, what should be done by humans, and what ought to be done by both — the human in the loop.
Volume plus repetition equals bot.
You cannot compete with agents on tasks that are repeatable, scalable and high in volume. That is not a threat. It's a boundary. Below the line, deploy machines. Above the line, deploy people.
Document ingestion is the wedge.
Claims. Medical records. Submissions. This is where agentic AI compounds fastest, and where humans have suffered longest. Start here.
RPA and agents are not evolutionary.
They are complementary. RPA is a rail — deterministic, brittle, cheap at scale. Agentic AI is a brain — probabilistic, contextual, resilient. Mature enterprises run both, deliberately.
HR is now central, not consulted.
Every role gets rewritten because every workflow does. The job description is the product. Culture change is the deliverable. If your HR team isn't in the room from day one, you're building an org chart for a company that no longer exists.