AI Communications
In our latest AI Real Talk session, Kristen Koenig, RingCentral’s GM of AI & Collaboration GTM, spoke with Robin Gareiss, CEO and Principal Analyst at Metrigy, and Andy Watson, Global Vice President of Product Marketing for CX Solutions at RingCentral, to unpack findings from Metrigy’s 2026 CX MetriCast research, a study of CX leaders across 10 countries.
Rather than focusing on AI hype, the discussion explored what separates successful AI deployments from those that stall. Watch the latest AI Real Talk session →
Below are six of the biggest lessons we took away and what they mean for businesses investing in CX technology today.
Key takeaways
-
- AI investment priorities have shifted from customer self-service toward agent enablement
- Fragmented data and weak governance remain the biggest obstacles to successful AI deployment
- Modern AI agents have evolved well beyond the limited chatbot experiences customers remember
- AI orchestration is improving, but many CX teams still lack end-to-end visibility across customer journeys
- Voice continues to be the richest customer interaction channel
- Strong AI business cases focus on revenue growth and cost avoidance
Why AI for agent enablement is growing faster than customer self-service
A year ago, Metrigy’s research showed that AI in customer experience investment was most heavily concentrated on the customer-facing side: chatbots, self-service portals, and automated resolution flows. In the latest research, however, those priorities have shifted.
As Gareiss shared during AI Real Talk, organizations now rank AI that helps employees serve customers more effectively ahead of AI built solely for customer self-service. That change reflects a growing realization: helping agents perform at their best often delivers faster, more measurable value than trying to automate every customer interaction.
“Where you can really start seeing huge benefits is internally. So take that AI, supercharge your agents, use AI-based analytics to find out what’s going right, what’s going wrong, and what you need to change. I think those are really important.”
We’re seeing that same shift across the industry. Businesses are investing in AI that helps agents work smarter through real-time assistance, automated summaries, quality management, coaching, and analytics because customers expect faster, more informed service when they do connect with someone.
This is also fueling rapid growth in workforce engagement management. AI is increasingly helping supervisors identify coaching opportunities, automate quality evaluations, forecast staffing needs, and improve agent performance at scale. As Gareiss noted, using AI to strengthen workforce engagement is becoming one of the fastest-growing areas in customer experience technology because it directly improves both employee effectiveness and customer outcomes.
For many organizations, this represents the most practical starting point for AI adoption. Internal AI deployments keep people in the loop, making it easier to validate recommendations, build trust in the technology, and demonstrate measurable business value before expanding AI into more customer-facing experiences.
Why most AI projects stall and what the data says about avoiding failure
The AI deployment success rate is not as strong as the volume of investment might suggest. Despite significant spending, Watson noted that roughly 85% of AI projects are stalling before reaching their intended outcomes.
Watson pointed to two common reasons why AI initiatives lose momentum: fragmented data and weak governance.
“85% of [AI projects] are stalling out for a couple of main reasons. Number one, the data is fragmented… then the other is just how do we govern [the data].”
Fragmented data has become one of the biggest obstacles to successful AI deployment. Customer sentiment, agent performance, self-service interactions, and other operational data often live in separate systems that don’t communicate with one another. Without a complete view of the customer journey, AI can only work with part of the picture, which limits the quality of recommendations, insights, and automation.
Governance is equally important. As Watson emphasized, AI isn’t a “set it and forget it” technology. Organizations still need people to validate responses, monitor performance, and continuously refine AI models to ensure they remain accurate, trustworthy, and aligned with business goals. Human oversight also helps reduce hallucinations and maintain confidence in AI-generated recommendations.
Successful teams treat governance not as a one-time implementation task, but as a built-in business practice. Combined with connected data, strong governance creates the foundation needed to scale AI confidently across customer interactions and internal operations.
How AI agents have moved beyond basic chatbots in customer experience
One of the biggest reasons customers were initially skeptical of AI in customer service was their experience with early chatbots. Customers were pushed into rigid conversation flows that handled only a handful of predefined requests. When their issue fell outside those boundaries, the only option was to ask for a human representative repeatedly.
Andy Watson likened those early chatbot experiences to having only three television channels: you could only choose from what was available, whether it solved your problem or not. As a result, customers quickly learned how to bypass chatbots altogether.
“We learned how to say ‘agent’ or ‘operator’ or ‘I need a human agent’” Watson said, “we’ve gotten so much more advanced now where there’s context.”
That context is what separates today’s AI agents from chatbots of the past. Instead of simply deflecting customer inquiries, modern conversational AI can authenticate users, troubleshoot issues, complete routine tasks, and determine when a live agent should step in. Just as importantly, it transfers the conversation with interaction history already intact, so customers don’t have to repeat themselves after every handoff.
This marks a clear shift in how success is measured across CX technology. Instead of focusing primarily on how many interactions AI can deflect, organizations are now evaluating how effectively AI resolves customer issues or how smoothly it hands interactions to human agents when needed.
What AI orchestration actually means for CX teams
AI orchestration is one of the most discussed concepts in enterprise CX right now, and also one of the least consistently understood. In practice, it refers to AI agents’ ability to manage customer interactions end-to-end across channels. This ensures that conversation details, history, and intent travel with the customer from one touchpoint to the next, regardless of whether they’re speaking with an AI system or a human agent.
The challenge is that most businesses have achieved this only in parts. Gareiss was candid about the current state of the market:
“So, I think that with orchestration, where we need to be heading is just continuously improving that.” Gareiss said, “Again, back to analytics. Once you have that, you can see here’s where I’m missing connections. Here’s where we’re losing customers.”
That visibility is essential because contact center teams have connected only parts of the customer journey. Customers may begin with an AI agent, continue with a contact center representative, and later interact with a frontline employee, yet each touchpoint operates in a different system. Without orchestration, those disconnected touchpoints create fragmented experiences, incomplete reporting, and missed opportunities to improve service.
Solving this challenge means giving customers seamless context across every touchpoint. Watson described this as the foundation of how RingCentral RingCX and RingEX work together on a single platform.
“It goes into a single platform. So you know when you’re contacting a business, there’s a front door, whatever that is. It’s AIR, AIR Pro, another AI agent. The same data is going to be used from beginning to end.”
Instead of treating customer-facing AI, contact center agents, and frontline employees as separate workflows, the platform shares context throughout the interaction. That means customer information, conversation history, and AI insights remain available regardless of who takes over the conversation.
As Watson continued:
“The customer journey doesn’t really change depending on who they’re talking to. And then the insights that you get, whether it’s an AI agent, a human agent, we’re going to have CSAT insights. We’re going to have quality insights.” Watson explained, “it’s bringing all of them together. So it’s a single journey for the customer.”
This unified approach gives CX leaders a more complete view of customer interactions while making it easier to identify operational gaps, improve handoffs, and optimize experiences across every stage of the journey.
Why voice is the most valuable data source for AI in CX
Voice remains the primary channel for many customer service interactions, particularly when customers need help with more complex issues. Gareiss explained that AI is increasing the value organizations gain from voice interactions rather than reducing voice’s role in customer experience.
“82% of all interactions go through voice either initially or as an escalation point… And that voice can be AI agent or human agent. The majority of it right now is human agent, but it’s shifting” Gareiss notes, “Consumers are actually surprisingly really willing to take a call from an AI voice agent and give it a chance…”
That finding challenges the assumption that digital-first customer service means text-first customer service.
Improvements in voice AI have changed how customers perceive AI-powered conversations. Natural speech, shorter response times, and smoother handoffs have made customers more willing to engage with AI voice agents, particularly when they know they can escalate to a human if needed.
But voice offers another advantage that extends beyond the customer experience itself: it provides richer data for AI. As Watson explained, organizations lose valuable insight when they rely only on transcripts.
“If you’re just looking at a raw transcript, you’re going to miss all that meaning. You’re going to miss the sentiment. You’re going to miss volume, speed. There’s so much more in voice.”
A transcript may capture what was said, but it often misses how customers express it. Tone, pacing, hesitation, emphasis, and even sarcasm can change the meaning of a conversation. Those nuances help businesses better understand customer sentiment, identify coaching opportunities, and uncover the root causes behind customer frustration.
How to build a credible AI ROI business case for CX leaders
The most common mistake CX leaders make when building an AI business case is presenting AI primarily as a way to reduce headcount. During the AI Real Talk session, Gareiss discussed that stronger business cases focus on revenue growth and cost avoidance and outlined the revenue framework.
“I like to look at more the revenue-generating ROI figures… customer lifetime value, churn rates, CSAT, upsell, cross-sell… if customers are happier, they’re going to write better reviews, they’re going to stay, and they’re going to refer us to somebody else.”
Satisfied customers are more receptive to additional offers, and that cycle builds on itself. When Gareiss described how better experiences lead to retention, referrals, and new revenue, Koenig put a name to it directly: “It’s the flywheel effect.”
On the operational efficiency side, AI adoption generates gains that translate to cost avoidance rather than immediate cost reductions. Gareiss addressed this directly:
“You’re not going to really cut people out necessarily,” Gareiss said. “You’re going to be not hiring as many new people moving forward in that area.”
The workforce engagement management numbers make this concrete. According to Gareiss, AI-assisted supervisor workflows are generating measurable time savings across key management functions.
-
- AI-assisted quality management saves supervisors roughly 12 hours per week
- AI micro-coaching saves approximately 7 hours per week per supervisor
- AI-driven scheduling and forecasting saves around 6 hours per week
That’s up to 25 hours per supervisor per week redirected from administrative tasks toward high-value work: career pathing, skills development, and meaningful agent coaching. As Koenig noted, supervisors aren’t being sidelined; they’re being made more dynamic, freed up to do the work they were originally hired to do.
For your finance teams, a stronger AI business case centers on executive metrics such as revenue growth, cost avoidance, workforce productivity, and customer outcomes rather than headcount reduction alone.
Conclusion
Throughout the AI Real Talk discussion, Robin Gareiss, Kristen Koenig, and Andy Watson returned to the same themes: connected data, strong governance, and measurable business outcomes. Together, these elements help organizations scale AI more effectively while keeping people involved where they add the most value.
Customer expectations, AI models, and business priorities will continue to shift. The companies that adapt successfully will continue refining how people, processes, and technology work together. That includes improving customer journeys, strengthening workforce performance, and using AI insights to make better business decisions over time.
Go beyond the top takeaways
Watch the full AI Real Talk session to hear Kristen Koenig, Robin Gareiss, and Andy Watson dig deeper into Metrigy's 2026 CX Metricast research, and discover how leading organizations are turning those findings into real business outcomes.
About AI Real Talk
AI Real Talk is RingCentral’s ongoing webinar series focused on the practical side of AI in business communications and customer experience. Each episode brings together RingCentral GTM and product marketing leaders with industry analysts and researchers to explore AI adoption trends, real-world deployment challenges, and strategies for building measurable business value.
Originally published Aug 03, 2026
