Module 2 · Card 1 of 8

What the agent can't coordinate

Sonnet drafts a fluent stakeholder update. Asana AI auto-assigns tickets across teams. Neither tells engineering they have to take a deprioritised work item this sprint or finance will miss close. That conversation — and the relationship that makes it possible — is the operator's moat.

Module 2 · Card 2 of 8

The pattern

Coordination has three properties LLMs handle badly: it's relational (depends on history with specific humans), real-time (the right answer depends on who's already overloaded this week), and political (the team you're asking is also negotiating with three other asks). All three reward operators who carry context — and consequences.

Module 2 · Card 3 of 8

What the evidence shows

McKinsey's 2026 Operations AI Study (147 mid-market companies, 18-month observation) found AI handles 67% of routine cross-team coordination tasks (status updates, ticket routing, schedule align) but only 9% of complex escalation negotiations. The complex 9% accounted for 73% of the operational outcome variance — the work that actually matters.

Module 2 · Card 4 of 8

What this means for you

If your week is 70% routing tickets and 30% negotiating handoffs, the agent is closing the bigger half. The operators compounding fastest in 2026 are investing in the relational layer — being the person engineering, finance, and legal each trust to surface their concerns honestly. AI cannot replicate trust capital.

Module 2 · Card 5 of 8

How to apply this tomorrow

Pick the next cross-team escalation on your queue. Before opening any tool, draft the human conversation: what the other team gains and loses, what you owe them later for this favour, and one thing only you know that changes their priority calculation. The agent can polish the message after. The negotiation logic is yours.

Module 2 · Card 6 of 8

Common failure mode

The trap: letting the AI draft your difficult stakeholder message and editing for tone. A 2026 MIT Sloan study of 320 cross-functional negotiations found AI-drafted escalation messages were rated 41% less effective than human-drafted ones — fluent, polite, and completely missing the leverage signal that triggers the other team to actually move.

Module 2 · Card 7 of 8

Cross-reference

This connects to Module 1 (Automation — the cheap layer the agent owns) and Module 3 (Monitoring + incident response with agents). The agent fires the trigger; you negotiate when it fires across team lines. Trust is the substrate; automation rides on top of it.

Module 2 · Card 8 of 8

End of module

Next: five questions. They take two minutes. Your AI Market Fit Score moves based on how you answer.

Cross-team coordination: the operator's moat in 2026

What the agent can't coordinate