Walks on West Cliff
Agentic AI – From Doing the Work to Directing It
Featuring: Mark Reimer, Agentic AI Consultant at PRG
While this conversation didn’t take place on West Cliff, the spirit was the same. Over a dinner at The Refuge in San Carlos, we explored one of the questions that has been coming up in recent conversations:
What happens to people when AI agents become part of the business?
Here is where Mark landed. What he sees in the field is an upshift in role, where the individual contributor becomes the manager of AI Agents. The work changes from doing every task manually to directing AI agents that handle the work, while you set the goal, judge the output, and own the decision.
That’s not an entirely new skill. Strong managers have always delegated effectively. The difference is that their newest team members happen to be AI agents.
Successful AI management means clearly defining:
– The desired outcome
– What “done” looks like
– Which decisions the agent can make independently
– When human approval is required
– What information and resources are available
To get the best results, instead of micromanaging, brief it like a new employee and then review the results.
Human approval is the product.
In real operational work, Mark has found that the value is trustworthy outputs that a human signs off on. Keeping experienced people in the decision loop is what makes it effective.
Manage outcomes.
The leverage lies in setting up a reliable system, where the work runs on a tight feedback loop. You define the outcome you want and the checkpoints that tell you it is on track. The agent works toward it, testing its own output against your criteria and flagging exceptions for you to evaluate. Done well, the same system runs again next week without you having to rebuild it.
The Bottom Line
The people who do well in the coming years will be the ones who learn to manage Agentic AI effectively, and who bring good judgment to the outputs that agents produce.