In the latest March AI Real Talk session, Mike Egli, CX Transformation Practice Leader at RingCentral, and Melody Brue, VP and Principal Analyst at Moor Insights, brought into focus a critical issue enterprises are facing today, highlighted in the RingCentral Agentic AI Trends 2026 report:

Adoption has surged across industries. In fact, 97% of organizations are now using at least one form of AI, and 86% report having an AI strategy in place. Yet the expected shift in performance has not materialized at the same pace.

Organizations aren’t questioning whether AI works—they’re questioning why, despite near-universal adoption, it hasn’t fundamentally changed how their business operates.

What’s driving the disconnect beneath the surface? Here’s how Mike Egli explains the gap between AI adoption and impact:

AI adoption’s reached scale, but it’s not embedded in the workflows that drive the business

AI is present across organizations, but it is rarely positioned where work actually moves.

“When we see those projects go forward, they didn’t really integrate. Integrate means a lot of things. There are systems integrations, like did it tie into your CRMs and your EHRs, but did it integrate into your business processes, too?”

In most cases, AI is introduced at the point of interaction. It assists employees or surfaces information, but the surrounding workflow remains intact. Work still relies on manual handoffs, delayed decisions, and follow-ups that depend on someone taking action. AI sits on top of the workflow instead of inside it, limiting its ability to influence how work progresses.

Organizations are overestimating their AI maturity because outcomes are unclear

The visibility of AI has created a perception of advancement.

“AI is phenomenal, but there’s a bit of an illusion around the maturity of organizations actually getting to outcomes with AI.”

The absence of clearly defined outcomes complicates that assumption. Many organizations can’t trace AI deployments to specific improvements in performance. There’s no consistent framework for measuring whether AI has reduced friction, improved responsiveness, or increased revenue.

This ambiguity allows activity to be interpreted as progress. It becomes difficult to distinguish between experimentation and execution. Without outcome-based evaluation, maturity is inferred rather than demonstrated.

Lack of Agentic AI implementation

Earlier automation models were built around containment. Systems were designed to perform specific functions within predefined limits. Once a task was completed, responsibility shifted back to a human operator.

“Agentic AI offers something really unique that we haven’t necessarily had before. Before, things had to be very prescriptive and very defined, and the guardrails you had to put around them were fairly strict. With agentic AI, it’s very different. Even after conversations at certain points in a customer’s journey, we can have awareness and drive action and workflows.”

Agentic AI allows systems to respond to context as it emerges and to initiate actions that extend beyond a single task. During a customer interaction, multiple processes can be set in motion across systems, with each action informed by the same underlying context.

The significance lies in continuity. Work no longer pauses at the boundary of a task. It progresses across functions without requiring manual coordination at every step. This transforms AI from a tool that assists work into a mechanism that carries it forward.

The scale of improvement reveals how inefficient legacy processes have been

Organizations are reporting improvements that would have previously been considered outliers. Productivity gains and efficiency improvements are no longer incremental.

“We’re seeing this massive acceleration of impact that we’ve never seen before. We were happy five or ten years ago, if you could make a 2 to 4 percent improvement in business efficiency and bottom line. You see that agentic, for the first time, can hold the ball and run with it. It can build better customer experiences and drive immense change.”

These results expose the structural limitations of legacy processes. Many workflows were designed around constraints, such as fragmented systems, delayed information flow, and reliance on human intervention for coordination.

The data already shows what is possible. By deploying AI agents,

  • 61% of organizations reported productivity benefits
  • 58% reported faster workflows
  • 49% reported a better customer experience
  • 45% reported reduced operating cost

Siloed AI deployments are preventing organizations from achieving enterprise impact

The way AI is introduced often determines its impact. Many initiatives begin within individual teams, deployed quickly to address immediate needs. Today, AI can be implemented with minimal setup, lowering the barrier to entry and accelerating adoption.

But speed creates a new problem. Organizations struggle to translate rapid AI deployment into sustained execution.

“A bunch of siloed AI projects and products that have their own data is not functional. We’re at the point now where we need to identify where we want to make an improvement, what the workflows and business processes are that need to be improved, and then how we measure that impact.”

Without a broader framework, each team defines its own objectives, selects its own tools, and measures success independently. This fragmentation is visible in ownership structures, where 27% of AI is owned by IT, 23% by dedicated AI teams, 21% is shared, and 9% has no clear owner.

Orchestration is required to connect systems, data, and teams

As AI deployments multiply, the complexity of managing them increases. Organizations often attempt to connect systems after they have already been implemented, creating layers of integration that are difficult to maintain.

“I see so many points of failure in organizations that typically point to the lack of organizational orchestration between all the individual teams that are going to own a piece of it. We see communications platforms rolled out for the contact center, but no one else is aware of the acquisition.”

Orchestration introduces a different approach. It aligns systems, data, and workflows from the beginning, ensuring that each component operates as part of a cohesive whole. This requires coordination across teams and a clear governance model.

Organizations overlook voice data as the richest source of customer intent

Voice interactions capture dimensions of communication that structured data cannot fully represent. Tone, pacing, and emotional nuance provide insight into the customer’s intent and state of mind.

“Voice is ultimately the most powerful form of communication. Context is the modern currency for customer experience. Because to provide an excellent customer experience, we have to understand what’s happening within the conversation and around the journey that a customer is going through.”

As voice data becomes structured and accessible, it can inform both real-time decisions and long-term strategy. Our RingCentral Agentic AI Trends 2026 report shows an anticipation that voice will be preferred in the next two years, increasing from 14% today to 23%, while chat potentially could be dropping from 42% to 25%.

Proactive engagement represents the next phase of customer experience

Customer experience has long been reactive, with organizations responding only after a customer initiates contact, often once issues have escalated.

“Customer experience is largely revolving around ‘I’m a customer, I have a problem, so I reach out to a brand.’ Inevitably, we’re already starting on the wrong foot because we’ve allowed the customer to get to the point where they have a problem.”

The next phase of customer experience is defined by anticipation. Using data and context, organizations can identify when a customer is likely to need support and act earlier.

The experience becomes more seamless and timely. Success depends on capturing and handling every interaction, and AI enables consistent coverage so that every opportunity is addressed.

RingCentral’s approach: a unified agentic AI layer

RingCentral’s approach is built on a deliberate decision to avoid fragmented AI deployments. Instead of releasing multiple standalone tools, the company embedded AI as a single agentic layer across its cloud platform.

This architecture ensures that all interactions—across voice, messaging, and digital channels—operate on the same data and context. It eliminates the need to reconcile multiple systems and enables consistent execution across workflows.

On top of this foundation, RingCentral structures its capabilities through an agentic voice AI communications suite:

  • AI Representative (AIR Pro) introduces agentic execution into customer interactions. It understands intent, determines next best actions, and completes tasks in real time. Beyond answering inquiries, it can orchestrate multi-step workflows, trigger systems, update records, and resolve requests across voice and digital channels.
  • AI Receptionist (AIR) addresses a fundamental operational problem: unanswered interactions. By ensuring that every call is handled and basic needs such as scheduling are met, it creates a consistent entry point for customers and eliminates missed opportunities.
  • AI Virtual Assistant (AVA) focuses on employees. It provides real-time access to context, past interactions, and next steps, removing the need to search across systems. This changes how agents operate during and between interactions.
  • AI Conversation Expert (ACE) extends beyond individual interactions. It analyzes conversations at scale, identifying patterns, surfacing coaching opportunities, and enabling insights that influence decisions across marketing, product, and operations.

Because all are built on the same underlying layer, they operate with shared context. This allows organizations to move from reactive responses to proactive engagement, where the system can anticipate needs, initiate outreach, and escalate to human agents when appropriate.

What separates adoption from impact

The conversation has moved past tools, pilots, and proof of concept. AI is no longer judged by what it can generate, but by what it can complete. Many organizations are stuck at a stall point, where AI identifies tasks, but execution still depends on manual follow-through.
Agentic AI closes that gap by turning insight into action, advancing workflows across systems and teams. This transition, from insight to completed action, is what turns AI from a capability into a driver of results.

If you’re ready to move from AI experimentation to real transformation, watch the full AI Real Talk session.

Originally published Apr 03, 2026