The first person to work side-by-side with her own AI clone says the hardest part isn't the existential dread — it's deciding what to delegate.
The Summary
- Kristi Edleson is chief of staff at Yutori, an AI startup that built an agent specifically to replicate her role — and she uses it daily to handle her workload
- The challenge isn't job security, it's cognitive: knowing which tasks stay in her brain and which get offloaded to the agent
- She draws one hard boundary: the AI doesn't touch finances, despite handling "numerous tasks" across context-switching work
The Signal
Edleson isn't an engineer. She came from 10 years in recruiting, pivoted to operations, and landed at an AI company in spring 2025 as the sole non-technical employee. Her job was to be a catch-all operator. The twist: her company's product goal was to build an AI chief of staff. She accepted the role knowing she was building her own replacement.
Except that's not what happened. The agent launched. She still has her job. And the most interesting part isn't that she survived — it's what she learned about the boundary between human judgment and AI execution.
"The hardest part is knowing when to do the work myself or outsource it to the AI agent."
This is the first real field report from inside the delegation layer of Web4. Not a demo. Not a pilot. A person doing knowledge work who now has an AI agent trained specifically on her role, and the cognitive load isn't "will I be replaced" — it's "which 40% of my day should I keep?"
Chiefs of staff live in constant context switching: scheduling, briefings, cross-functional coordination, fire drills, strategy memos. Edleson says the AI agent augments rather than replaces, but she refuses to let it handle finances. That's the tell. The line isn't technical capability. It's trust, liability, and the kind of judgment that comes from skin in the game.
Key tensions emerging:
- Cognitive overhead of delegation itself: deciding what to hand off takes mental energy
- The "numerous tasks" the agent handles aren't specified — likely the repetitive, low-stakes work
- Finances stay human because mistakes there have consequences agents can't own
The Implication
If you're in a role that involves coordination, synthesis, or high-frequency low-stakes decisions, you're about to face Edleson's problem: not whether an agent can do parts of your job, but which parts you're willing to let go. The bottleneck isn't the AI. It's you learning to manage a digital version of yourself.
Watch for more field reports like this. The companies building agents are starting to use them internally, and the people working alongside them will tell you more about Web4's actual shape than any product launch deck. Edleson's finance boundary is instructive: agents will handle breadth before they handle liability.