- The shift from generative AI (produces text) to agentic AI (takes actions) — what that distinction means in practice
- Levels of autonomy: AI that suggests, AI that executes with approval, AI that executes independently — and the governance implications of each
- What this looks like in real roles: agents that research, schedule, draft, run workflows, manage handoffs end-to-end
- What stays constant: you still verify, you still own the output, the confidentiality and disclosure rules from Episode 4 apply at higher stakes
- What changes: more of your job becomes directing and reviewing — the doing-to-directing shift accelerates dramatically
- The most durable professional skill in an agentic world is quality control — knowing what good looks like, catching what's wrong, holding the accountability that agents cannot hold themselves
Think about a job you wish you could hire someone to do — a role that would make your work better or your team more effective, but that you've never had the headcount or budget to fill. A researcher, a coordinator, an analyst, a writer. Now answer two questions:
First: could an agent do this job? What would you hand it, what would it hand back, and what would it need to know about your situation to do it well? Second: how would you evaluate its work? Not just "did it complete the task," but what would good look like? What would make you trust it enough to actually act on what it produced?
Discuss the imagined jobs we wish we could hire for — and what good would look like if an agent did some of that work.
The imagined jobs tell us where the team has been under-resourced — work that's been deferred, relationships that haven't been built, thinking that hasn't happened. The quality standards people articulate for agent work are equally important — being able to say precisely what good looks like is the skill that translates forward.
- Should I be worried about agents taking my job?
- Agents will change what's in every job, and some jobs will change more than others. The people best positioned are the ones who understand what the agent is doing well enough to evaluate it — which is what this series has been building toward.
- What should I be learning next?
- Three skills that travel: structured thinking and clear communication (prompting is the first version), quality evaluation (knowing what good looks like), and judgment under ambiguity. Those don't go obsolete.
Send us the imagined jobs your team came up with — and the quality standards they articulated. Aggregated across the organization, that tells us where unmet capacity needs are and how ready teams are to work alongside agents.