← Insights

Late 2025 / Early 2026: Voice, Infinite Capacity, and the Operating Model

November through January saw three big shifts: voice agents going mainstream, the "infinite capacity" framing becoming real, and the operating-model conversation maturing.

Three threads tied together the November 2025 - January 2026 window: voice agents going from demo to deployable, the "infinite capacity" framing becoming a real business reality, and the operating-model conversation moving from theory to architecture.

Infinite capacity is no longer a prediction

Since we started posting several months back, the predictions have become reality and normalized. Capacity constraints — historically the binding limit on what RevOps and CX teams could attempt — are now negotiable in ways they weren't a year ago.

That has a knock-on effect on every project prioritization conversation: "we don't have headcount" stops being a sentence-ender and becomes the start of a different question.

Agentforce Voice goes from demo to deployable

Voice agents had a quiet inflection point in this window. The Pat Dennis demo — watching Agentforce's Employee-Facing AI Agents handle real workflows via voice — was the clearest signal that voice has moved from "cool demo" to "we can actually deploy this."

The implication for service ops: the IVR-to-human routing pattern of the last 20 years is the next layer to get rewritten.

Service Cloud + automation patterns continue to compound

Multiple use cases in this window reinforced the same architecture: agent handles routine resolution, hands off cleanly to a human for judgment, logs everything automatically.

The win isn't fewer humans. The win is humans handling fewer routine cases — which means they handle the hard ones better.

The operating-model questions

If the infrastructure is mostly settled, the conversation moves to operations. The questions that came up repeatedly:

  • Who owns orchestration?
  • How do agents share context without leaking it?
  • How do we prevent agent sprawl before it becomes a governance nightmare?
  • How do we prove ROI beyond "cool demos"?

The orgs that have crisp answers to these questions before the next planning cycle are the ones that will scale agents past pilot. The rest will be explaining at QBR why the pilots never made it to production.