Design the operating model first
Agents inherit the quality of the process beneath them. Clarifying state, ownership and decision criteria comes before choosing an interface or model.
AI-native sales operations
A current operating practice that makes pipeline rules, territory logic and forecast inputs legible to both people and agents.
system · current
Senior Director, Sales Operations at Emergence AI · Updated Aug 2026
- Sales operations
- Agents
- Revenue systems
- Decision design
01 · Problem
Traditional sales operations assumes a person will notice every signal, interpret every rule and move every record between systems.
02 · System
The operating model restructures data, rules and handoffs so agents can support pipeline inspection, territory decisions and forecast preparation without obscuring human accountability.
03 · Outcome
This is active work at Emergence AI. Public detail is intentionally limited to the operating principles that can be shared responsibly.
Context
Revenue teams have accumulated dashboards, fields and workflows designed for manual interpretation. Adding an agent on top of that stack does not make the underlying operating model agent-ready.
The work starts below the interface: definitions, decision rights, context quality, exception handling and feedback.
Architecture
- 01Canonical revenue definitions and pipeline state
- 02Rules expressed as inspectable operating logic
- 03Context assembly for agent-supported decisions
- 04Human approval at consequential decision points
- 05Feedback captured as structured operating data
Decisions & trade-offs
Keep judgement visible
Recommendations should show the evidence and rule path that produced them, especially where a human remains accountable for the outcome.
Learn from corrections
A corrected recommendation is captured as a signal for improving definitions, rules or context—not discarded as a one-off exception.
Evidence
- Current operating work; confidential metrics and implementation detail are not published
- Public case study scope limited to transferable design principles
What this taught me
- AI-native is an operating-model property, not a feature label.
- The most valuable automation targets the movement and interpretation between systems, not isolated tasks.
Follow the system
Connected work is resolved from the shared content graph.