Native and Cross-System Support Agents

This points to support AI splitting into two layers, native agents that make the help desk itself more useful, and cross system agents that sit above the help desk and automate a broader workflow. Zendesk can bundle a small set of default agents because it already owns the ticket queue, macros, knowledge base, and agent desktop. But once an agent needs Gmail, Slack, Gong, or an industry system like ServiceTitan, the job moves outside Zendesk’s natural product boundary.

• Incumbents have strong reasons to ship native agents first. They already control the system of record, can write new AI generated data back into it, and can use tighter bundling to make the core product stickier and harder to replace.
  1 sacra
• The adjacent layer is where fragmentation starts. New support agents and workflow tools run on top of Intercom or Zendesk for escalations, docs, and human handoff, while competing to own the customer conversation and more of the automation budget.
  2 sacra 4 sacra
• The practical dividing line is workflow breadth. A native Zendesk agent can answer tickets, classify issues, suggest replies, and trigger support actions inside Zendesk. A vertical or marketplace agent can combine support data with call transcripts, email, chat, and back office systems to do a fuller job.
  1 sacra 2 sacra

From here, the winners are likely to be a few deeply embedded native agents inside each major system of record, plus a long tail of specialized agents that span multiple systems and industries. That creates more demand for the integration layer, because the more the market fragments at the agent layer, the more valuable the infrastructure becomes that moves fresh context between them.