The human + AI agent interaction layer

As AI agents scale, interaction matters as much as capability

As AI agents become more embedded in workflows, expectations shift. People care less about novelty and more about usability. They’re focused on how naturally agents communicate, how reliably they behave, and how much friction they remove from everyday work.

One clear shift is emerging: a growing expectation that AI agents communicate naturally across multiple conversational touchpoints (spanning voice, video, chat, and messaging), so work can continue seamlessly as interactions move between channels.

Respondents were asked to imagine interacting with an AI agent across customer-facing and employee contexts now and in two years. Here were their preferences:

Voice

Video

Chat

Conversation data captures the “why” behind behaviors, trends, and sentiment. Agentic Voice AI surfaces those insights and makes them actionable, equipping agents with the information they need to navigate everyday work.

Reliability matters more than personality

When asked which human traits they would give a digital worker, leaders overwhelmingly chose traits tied to correctness and clarity, not emotion.

The traits leaders want most in AI agents

Reliability

Creativity

Common sense

Empathy

Accountability

Humor

Patience

The message is clear: people want agents that are dependable, able to route context across workflows, and capable of leveraging common sense and creativity to handle ambiguity. This preference directly reinforces the need for orchestration. Reliable agents are trusted to move work forward without constant supervision.

Building trust to reduce day-to-day friction

Among organizations already using AI agents, the biggest barriers to broader adoption are confidence and connectivity.

Those using AI agents experienced the following barriers

Trust in outcomes

Employee resistance

Data integration issues

Cost concerns

Compliance or regulatory risk

Teams want AI agents that are predictable, well-governed, and clearly integrated into existing systems. Leveraging trustworthy agents that are easy to work with decreases resistance to agentic AI.

From siloed AI agents to coordinated, agentic systems

Chapter 3

The next phase of AI is system-level, not tool-level

Conclusion

RingCentral, the RingCentral logo, and all trademarks identified by the ® or ™ symbol are registered trademarks of RingCentral, Inc. Other third-party marks and logos displayed in this document are the trademarks of their respective owners.