Your AI Agent Org Chart: When to Split Roles, When to Combine Them

Most AI programmes start with a headcount question dressed up as a technology question: how many agents do we actually need? Get it wrong one way and you build a single assistant trying to be everything to everyone. Get it wrong the other way and you end up with a sprawl of single-task bots nobody can manage.

Kowalah runs its own go-to-market on a team of named AI agents, and we've made both calls in public. John McMahon (our CRO agent) and Brian (our AE agent) look at exactly the same deal data — but McMahon owns pipeline-level judgment and Brian owns deal-by-deal execution. Same source, different lens, so they stayed two agents, not one. Rancho, our delivery consultant, does the opposite: in a human team, strategy, functional consulting, and technical build would be three separate specialists. One agent handles all three, because there's no handoff cost between them the way there is between people.

What you'll see:

• The real system prompts and skill lists behind McMahon, Brian, and Rancho — not a diagram, the actual configuration
• The split test: same data, different lens means two agents
• The combine test: no coordination cost between the skills means one agent
• What a human manager actually does for an agent day to day, and why every agent needs one

Who should attend:

COOs and Heads of AI/Ops designing their first — or fifth — AI agent. Functional leaders about to write an agent's job description and unsure whether it should be one agent or three.

What you'll leave with:

• A working framework for deciding split vs. combine before you build anything
• A one-page agent org chart template you can fill in for your own team