2025 marked the year when organizations began to adopt AI broadly. 2026 will be the year they confront a new reality that using AI and benefiting from it are not the same thing.

In partnership with Opinium Research, we recently surveyed 2,000 IT, HR, and CX decision-makers (manager level and above) from the US and UK, representing small and mid-size businesses (SMBs) and enterprise organizations across retail, technology, healthcare, legal, and financial services.

Our soon-to-be-released RingCentral 2026 Report on Agentic Voice AI Trends reveals that, across nearly every industry, AI is present in dozens of tools and workflows; yet, few organizations have successfully integrated those systems into a coherent whole. (Of decision-makers surveyed, 86% say they have an AI strategy, but only 39% have integrated AI company-wide, and only 16% have deployed AI at scale.)

The result is duplicative work, inconsistent customer experiences, and growing friction for employees asked to collaborate with intelligent tools that don’t collaborate with themselves.

That fragmentation has set the stage for what’s next in agentic AI: moving from AI features scattered across applications to orchestrated AI agents capable of carrying context and taking action across channels and systems.

Here are RingCentral’s top predictions for how AI (and the organizations deploying it) will evolve in 2026.

1. Agentic AI becomes the new architecture of work

Software has been structured around applications for two decades. What we’ll start to see next is that work organizing around agents instead.

AI agents, referred to as “digital workers” in our study, are emerging as the first operational layer capable of performing end-to-end work, moving tasks forward across platforms, workflows, and communication channels. Our research shows:

  • 97% of organizations are already piloting or deploying AI tools, and 83% have launched at least one AI initiative
  • 96% say AI agents will be essential to competitiveness
  • And, leaders across CX, IT, HR, and Operations expect agents to assume measurable workflow responsibilities.

The change is both conceptual and technical as organizations evolve their perspective on AI from “task automation” to “workflow orchestration.”

“AI is everywhere, yet nowhere. In 2026, the real shift will come from making AI truly operational. That means integrated systems, governed data, and intelligence that can understand human context at scale,” says Kira Makagon, RingCentral’s President and Chief Operating Officer.

We’re entering a moment when autonomous, context-carrying agents (and Agentic Voice AI) become the connective tissue of modern business operations.

2. AI strategy management shifts from scattered to truly governed

Right now, AI strategy ownership is split across functions, with 23% held by IT leadership, 23% by dedicated innovation teams, and 21% shared between business and technical leaders. Another 14% is owned by operations or CX leadership, and in 9% of organizations, no one owns it at all.

These function-specific ownership models may have worked in the past when AI lived in isolated pilots and automation projects, but they break down once AI starts impacting revenue, customer experience, risk, hiring, data architecture, and internal culture.

Companies are realizing that AI cannot be confined to a single department and must operate with clear, organization-wide rules. A more formal and coordinated approach is replacing the ad-hoc ownership model with:

  • Cross-functional oversight groups that bring CX, IT, HR, operations, legal, and finance into the same loop
  • Leadership roles focused on governance, data responsibility, and model oversight
  • Board-level involvement in risk, workforce impact, and data architecture
  • Direct guidelines for how AI agents operate, escalate, and interact across teams

“While AI adoption in enterprises seems inevitable, this research provides a more nuanced view of the realities. On one hand, it’s validating to see that CX leaders are seeing real ROI, and in less time than might be expected,” says Jon Arnold, Principal Analyst, J Arnold & Associates. “The data also shows a very high level of optimism towards AI, but a closer look counters some of that. Most deployments are more standalone for specific use, rather than being more fully integrated into workflows. For AI to deliver on its lofty promises, the latter will have to become the norm, where deployments have an impact at scale, not just across the organization, but with both customers and employees.”

A shared model turns governance into a stabilizer that eliminates rework, lowers project failure rates, and gives teams confidence to scale. Once decision rights and rules are clear, progress is steadier and less dependent on individual champions.

3. Human-AI teams become the new productivity engine

One of the most overlooked aspects of agentic AI is the friction that comes from learning to work alongside it. Teams are still figuring out what AI can actually do, where it fits, and how it changes the work they already know.

Our study shows that concerns about job impact haven’t completely vanished, and nearly 40% of organizations have paused or cancelled an AI project, mostly because expectations, workflows, or training weren’t aligned from the start. However, many (53%) now say they would prefer managing a digital agent.

This mix of reluctance and growing comfort suggests that people aren’t rejecting AI in general, but rather pushing back against AI that’s confusing, inconsistent, or bolted onto a workflow that wasn’t designed for it.

Most of this friction is practical. Take, for example, the frustration that results when an AI system behaves differently in chat and on the phone, fails to hand off work with gaps in context, or adds steps instead of removing them. Those are moments people remember and that shape opinions about whether AI is helping or getting in the way.

When AI starts behaving predictably with the same rules, logic, and expectations, tension decreases. Team members stop second-guessing the system and start focusing on their actual work. Then, the most significant productivity lift happens as AI lowers the coordination overhead that slows people down (like the follow-ups an AI agent handles automatically or the context it organizes before a call) and is a steady presence that’s deeply felt in the everyday flow of business.

4. CX becomes the proving ground for orchestrated AI agents

Customer experience (CX) is where disconnected systems cause the most pain, and the benefits of orchestration are most evident.

Fragmentation is a common experience in CX, and everyone feels its impact. Customers repeat the same information across multiple channels and receive inconsistent answers across touchpoints. Cases bounce between employees with sluggish handoffs that don’t provide the full details of the customer’s original inquiry or concern. As a result, the endless cycle of repetition and frustration continues.

“In CX, the gaps are impossible to hide. When information actually carries from one interaction to the next, whether it starts in voice, chat, or text, everything changes,” says Carson Hostetter, RingCentral’s EVP & GM of CX & AI. “Customers stop repeating themselves, agents get straight to the issue, and the whole experience becomes more streamlined. That’s where AI agents make a real difference.”

Organizations utilizing AI agents report significant improvements in resolution speed, satisfaction, and consistency across all channels:

  • 61% increased productivity
  • 58% faster workflows
  • 49% better customer experiences
  • 45% reduced operating costs
  • 45% improved customer satisfaction
  • 40% easier compliance/data security

These early gains likely stem from coordinated systems that are finally communicating with each other effectively, rather than requiring employees to manually bridge the gaps.

CX will be the proving ground for agentic AI because the stakes are high and the potential payoffs are tremendous: shorter wait times, fewer escalations, consistent responses, cleaner handoffs, and customer experiences that don’t differ by the channel used.

5. SMBs move faster today, but enterprises bridge the gap once they connect systems

SMBs have had a structural advantage in AI adoption. With fewer legacy systems and shorter decision cycles, they’ve been able to move quickly from experimentation to deployment. Our research reflects this speed, with SMBs reporting that they see AI ROI slightly faster than large enterprises.

Now, enterprises are positioned to catch up, supported by more robust architecture. When larger organizations adopt orchestrated systems (in which AI agents can pass context across workflows, applications, and teams), they’ll unlock efficiencies previously buried by their own complexity.

“When you look at what’s happening in the market, AI agents really are becoming the engine of growth,” says Akshay Srivastava, RingCentral’s EVP & GM of SME. “SMBs will keep moving fast as they can spin up agents quickly and see results right away. For larger companies, the turning point is different. Once they connect their systems so context actually travels with the work, their scale becomes a real advantage.”

2026 is the year AI stops operating in silos

After a year of rapid experimentation, organizations are realizing that the biggest challenge presented by AI is finding alignment.

Our research and predictions reinforce the same key takeaway: the advantage will shift to organizations that can effectively integrate AI as a cohesive system.

To achieve true integration, companies will need an agentic AI architecture that enables agents to carry context, coordinate tasks, and support people across workflows instead of operating in isolation. That’s when CX can become a real proving ground, SMBs and enterprises can unlock value through their respective AI paths, and governance can shift from a checkpoint to a growth function.

Conversational data, including Agentic Voice AI (AI that analyzes intent, sentiment, and context from voice and activates the appropriate agentic workflows), serves as one of the richest signals for intent and context within that system.

Our upcoming report goes deeper into these patterns, the data behind them, and where organizations are heading next. Sign up to get the report.

FAQs

What is an AI agent?

AI agents (defined as “digital workers” in our RingCentral 2026 Report on Agentic Voice AI Trends) are autonomous, software-based workers that can perform tasks, carry context, make decisions within guardrails, and coordinate work across systems, extending beyond generating outputs within an app.

What is Agentic AI?

Agentic AI refers to systems that take action. These systems can interpret intent, reason within context, and advance work across workflows, unlike traditional AI features that operate in isolation.

What is Agentic Voice AI?

Agentic Voice AI leverages agentic intelligence to analyze conversational data, particularly in spoken interactions. Voice provides one of the richest signals in the enterprise: intent, sentiment, urgency, nuance, and context. Agentic Voice AI captures that signal and allows AI agents to carry it forward through a workflow, triggering the right actions across systems. It’s not “the bridge” between tools, but it is a powerful upstream input that makes orchestration smarter, faster, and more aligned with how people communicate.

Are “digital workers” and “AI agents” the same thing?

Yes. The RingCentral 2026 Report on Agentic Voice AI Trends used the term “digital workers” in the fielding survey, but across RingCentral, we use the term “AI agents.”

+++

Survey methodology note: The RingCentral 2026 Report on Agentic Voice AI Trends is based on a Q4 2025 survey conducted by Opinium Research among 2,000 IT, HR, and CX decision-makers (manager level and above) from the US and UK, representing SMB and enterprise organizations across retail, technology, healthcare, legal, and financial services. In the study, “digital workers” refer to AI agents, defined as autonomous software-based workers capable of carrying out tasks and collaborating across workflows.

Originally published Dec 09, 2025