Praveen Ravela:
Hi everyone. Thanks, David. Hi, Ward. I’m Praveen Ravela. I’ve been with AMC for over 15 years in various roles and currently serve as Chief Technology Officer and Head of Product, Services, and Solutions. I’m excited to have this conversation with David and share what we’re hearing from customers—what they’re excited about and how they’re using conversational AI and IVA to serve their end customers.
David Grant:
This conversation is really about conversational AI—something we’re all interacting with in some way today. Whether it’s Google giving you a Gemini response or Apple AI on your iPhone, it’s becoming part of daily life. I even use it to find a dairy-free pancake recipe for my wife!
Praveen, based on your experience and some of the customer calls I’ve pulled you into, what’s the best way for a company to determine if conversational AI is right for their contact center? As this slide points out, it’s not exactly “plug and play.” There’s development and work involved. What questions should organizations be asking themselves when evaluating whether to bring generative AI into their IVA today?
Praveen Ravela:
Great question, David. When we talk to customers, nearly every contact center or business already has an IVR in place. The goal has always been to address customer questions and issues as quickly and automatically as possible—so customers don’t have to wait in long queues to reach a human agent.
But as we all know, customers are often frustrated with traditional IVRs. They’re stuck answering prompts or keying in digits, and many try to bypass the system just to reach an agent directly. Contact centers tell us the same thing: no one wants to go through that process anymore. People want natural conversations that quickly resolve their issues.
Businesses recognize this—they want to deliver an experience where every customer interaction feels natural and conversational while minimizing wait times. The challenge is figuring out, based on their current contact center and technology stack, what it will take to enable that kind of conversational experience. What’s the effort? The complexity? The cost?
David Grant:
It’s also worth noting that what customers are asking for goes beyond just IVA. We’re seeing interest across the broader spectrum of generative AI in the contact center. One capability feeds into another. For example, many customers ask us about transcription—not just for reviewing agent conversations, but to power other functions like sentiment analysis.
They want to know: Are the scripts and training materials we give our agents creating positive customer interactions? Transcription can also help automate quality assurance, reducing the need to listen to every call by providing summaries or insights into how the call went.
Praveen, would you agree that an interaction flow could look like this: a customer calls in and speaks to a generative AI IVA. If the IVA can’t solve the issue—say it’s a retention call or needs escalation—it passes the call to an agent, equipped with context from the AI interaction. Then, after the call, a supervisor could review the transcription, sentiment, and summary, feeding that data into reporting and improvements. The IVA is just one piece of a larger ecosystem. Does that align with what you’re seeing?
Praveen Ravela:
Yes, absolutely. That’s exactly what we’re aiming for—looking at the entire customer journey. Every CX leader wants to deliver the best end-to-end experience. It starts with automating as many interactions and use cases as possible through conversational AI.
And for use cases the IVA can’t handle, the interaction escalates to a human agent. Even then, we’re working with customers on bringing generative AI assist tools to agents to help reduce average handle time—so they’re able to resolve calls faster than they could without AI support.
David Grant:
I also think it’s important to differentiate between hype and reality. What’s actually working today? What can companies realistically implement now?
Praveen, do you have an example of a current use case where generative AI is already making a tangible impact?
Praveen Ravela:
That’s a great point, David. One of our customers, a major U.S. car dealership, asked the same question: “What’s hype, and what’s real?”
We worked with them on a project to explore that. Their current setup was an on-prem contact center with a standard IVR. They weren’t satisfied with it—customers had to key in responses for every prompt, and most just tried to skip to an agent, only to end up waiting in a queue.
They launched a project to understand their options. Could they transition to conversational IVA without completely moving to the cloud? Could they layer conversational IVA on top of their existing on-prem infrastructure without disruption?
Their goal was to determine what steps, transformations, and investments would be required to deliver a conversational experience—and how much change their business could accommodate while still providing better service to their customers.
David Grant
Yeah, on that topic—
We engaged in a sandboxing project with this customer. We used what’s currently available in the market and tested all the different use cases they gave us. We leveraged a variety of technologies to do this.
Since that engagement, we’ve had several other customers come to us with similar requests. We’ve been able to share a lot of insights from that initial engagement and others we’ve done, especially regarding the technologies available.
We’re going to dive a little deeper into those technologies and explore what they’re offering—or, as you’ll see in the slides, what they’re “cooking up” for us. But we’re not quite there yet, Praveen.
I do have a question, though. We’re seeing a lot of interest in this space, and it’s really ramped up—especially in the last month. Why do you think customer experience leaders and contact center managers are feeling the pressure to act now? What’s driving that urgency?
Praveen Ravela
The way I see it, everything is being driven by what’s enabling us as individuals publicly.
In the past, the phone channel was king—that’s how everyone contacted a business to get issues resolved. But as technology evolved, and as individuals have become more tech-savvy, more communication channels have been enabled right on our handheld devices.
Now there are so many social channels and ways to communicate. That’s pushing businesses to transform. They see they need to evolve with the technology, just as individuals are adopting it. That’s shaping the business need—they have to meet customers where they are today.
If we look at today’s world, everyone expects answers through their mobile devices. People aren’t even using PCs the way they used to; they want everything handled on their phones.
That desire for comfort and convenience is key. And from a business perspective, if they’re not providing that comfort and convenience, customers will look for another business that does.
David Grant
Yeah, that makes a lot of sense.
Let’s take a look at these different technologies, Jess, if you wouldn’t mind pulling that up—awesome.
One thing we believe customers and partners should consider when thinking about their next steps and readiness with Gen AI is what we call the “three C’s”: Comfort, Complexity, and Cost.
Cost is pretty straightforward—how much time and effort is it going to take to implement something like this?
Comfort and Complexity are a little more subjective. Everyone has their own comfort level with AI and willingness to interact with it.
In my experience, I like to talk to friends and family about this because they’re not in the contact center world—they just interact with contact centers in their everyday lives. For example, I’ve been in the living room while my mom was on the phone with an internet provider, and I didn’t even have to ask her opinion—she was very vocal about her experience (I hope she doesn’t see this!).
One thing I hear a lot—and Praveen, you mentioned this earlier—is when I ask people what they think about these tools, they usually say, “What’s out there right now doesn’t work. I tell it what I need, it doesn’t understand, and then it just transfers me to an agent.”
And I’m always quick to clarify: no, that’s not what we’re working on. These Gen AI tools we’re talking about didn’t even exist five years ago. This is much newer, much more capable technology than what’s been around in many contact centers up until now.
Internally, companies have their own experiences with these tools too. Maybe they’ve called their bank and had a frustrating interaction with an IVA. That makes them question: “Why would we want something like that?”
It’s important to understand these are not the same systems.
So when we talk about Comfort, it’s really about resetting expectations.
How would you approach that, Praveen? How do we help people reset expectations around what’s possible and what’s reasonable to expect in terms of comfort?
Praveen Ravela
Yeah, with comfort and convenience, as we discussed earlier—
From the end user’s perspective, they’re calling because they have an issue, a concern, or a question. They just want to get it resolved quickly and get back to their day. No one’s calling a business just for fun.
From the business side, it’s really about how quickly they can deliver that resolution. That speed and ease are how businesses are evaluating their customer experience success moving forward.
David Grant
Exactly.
Looking at these options in front of us—Salesforce Agentforce, Copilot, Amazon Bedrock, Lex, and Chime—
I’d say, based on the interactions we’ve had with these tools, the conversations they generate really are conversational.
Praveen, I assume you’ve had similar experiences? When I ask these tools questions, the responses I get back—whether it’s voice with Copilot and Amazon or chat with Salesforce Agentforce (since voice isn’t available yet there)—they’re very conversational and, I’d say, pretty comfortable. Would you agree?
Praveen Ravela
Yes, I’d say they’ve improved significantly.
If you look at the evolution:
It started with traditional IVRs, where everything was scripted. You asked a question, you got a specific answer, then another question—it was a rigid path. The customer had to follow exactly what was scripted. That had major limitations.
Then the next phase of technology allowed you to say responses instead of keying them in—because no one wants to sit there typing every digit of their mailing address, for example.
So speech recognition came in, converting spoken responses into text and trying to process those. It addressed some challenges, but it was still pretty scripted—every use case had to be predefined to direct the customer down a specific path.
Praveen Ravela:
One of the major challenges businesses face today is the need to constantly update their systems to accommodate evolving use cases. As markets shift, businesses must adapt their offerings, leading to continuous re-scripting of processes. This ongoing investment often lacks a clear ROI, as the work merely shifts from human agents to engineers without eliminating the underlying tasks.
David Grant:
Exactly. For instance, with the tools we’ve been exploring, I can interact via voice or chat, saying something like, “I’m free for an oil change next Friday between 8:30 AM and 10:30 AM. Will it take more than an hour?” The system can respond, “It’ll take about 45 minutes. We have a slot from 9:00 AM to 9:45 AM. Does that work for you?” Upon confirmation, it schedules the appointment and updates my customer record in the CRM. Traditional systems require extensive scripting for such interactions, which is labor-intensive.
Praveen Ravela:
That’s the goal. Businesses are looking for Gen AI’s ability to naturally understand customer intents and determine the necessary operations without predefined scripts. The AI should handle conversations organically, like a customer saying, “I want to bring my car in for service,” without needing scripted responses for each possible query.
David Grant:
Right. So, with current on-premise Gen AI solutions, there’s complexity and a need for specialized skills. Does this mean enterprises must hire new talent for AI-specific roles, or can existing teams adapt?
Praveen Ravela:
Great question. That’s why we’ve evaluated current platforms to identify options that minimize the learning curve. For example, if a company uses Salesforce, integrating Agentforce is straightforward since their data and processes are already within that ecosystem. Similarly, Microsoft shops can leverage Copilot without significant additional training, and AWS users can utilize their existing tools and expertise.
David Grant:
So, for Amazon, the strengths lie in automation, voice-driven interactions, and integration with AWS tools. For Microsoft, Copilot’s integration across their suite makes it ideal for collaborative environments. And for Salesforce, Agentforce is best suited for those already embedded in their ecosystem.
Praveen Ravela:
Exactly. AWS is a natural choice for enterprises deeply integrated into their platform. Microsoft Copilot is gaining traction among organizations seeking to automate business use cases, including government agencies exploring conversational IVAs. Salesforce’s Agentforce is particularly beneficial for small to mid-sized businesses looking to optimize resources and reduce customer wait times.
David Grant:
Reducing customer wait times is crucial. Many people express frustration with being put on hold. By implementing Gen AI tools to handle self-service use cases, we can enhance the customer experience significantly.
Praveen Ravela:
Absolutely. The key is to identify where the business stands, the technology stacks in use, and the best fit for their needs without causing disruptions. It’s about integrating solutions that align with existing systems and processes.
David Grant:
That makes sense. There are many considerations for businesses to ensure successful AI integration. When done right, the benefits are substantial, leading to improved customer experiences and operational efficiency.
