2
2 Comments

AI Employees are easy to build. Managing them is the hard part.

We're seeing more AI agents become capable of doing real business work.

But once you have multiple agents working across an organization, a new problem appears:

How do you actually manage them?

Who owns the work?
What should they be allowed to do?
When does a human need to approve something?
How much did the work cost?
What happened after the task was completed?

This is the problem we're building OneCo around.

Our model is simple:

AI Employee → Outcome → Approval → Governance → Audit

We're starting with the idea that an AI Employee shouldn't just be another chatbot or automation.

It should be a managed worker inside an organization.

We're still very early, and that's what makes this stage interesting.

I'd love to hear from other founders and builders:

If you could hire one AI Employee today, what job would you give them?

posted toAvatar for product Oneco
Oneco
  1. 1
    One insight that kept coming up during development: Building an AI agent is becoming easier every month. Building trust around AI work is not. Organizations don’t just need execution. They need ownership, approvals, visibility, cost accountability, and auditability. That’s the part we’re most interested in exploring.
    1. 1
      Approval is the part of this chain I think gets surprisingly difficult once the AI employee can take consequential actions. Suppose the employee receives valid approval, begins a task, and the underlying authority changes before the action actually reaches the system that performs it. The approval record is genuine. The agent may have followed its instructions perfectly. But neither necessarily proves it was still authorised at the moment of consequence. We've been testing this boundary with OpsWatch, and it keeps pushing the same question to the surface: should authority belong to the task, the approval, or be revalidated at the resource immediately before execution? I suspect managing AI employees eventually requires treating authority as something that can expire during the work, rather than something established once at the beginning.