Home > Columns > Executive Interviews

Call Simulator Executive Interview

Click image to view the ebook

 In today’s customer service landscape, training and development are undergoing a radical transformation. As organizations continue to adopt advanced technologies like real-time agent assist and AI-driven quality management tools, one critical element is often overlooked: foundational, skill-based training. In this interview, Sheri Greenhaus, Managing Partner of CrmXchange, sits down with David Lawson, Co-Founder and CEO of Call Simulator, to discuss the shifting dynamics of contact center training and the growing role of simulation in preparing agents for real-world scenarios.

Lawson points out the risks of over-reliance on technology without preparing human agents to fully utilize it. The conversation explores how scalable simulation, combined with AI-powered coaching, is helping organizations train agents more effectively across a variety of roles. They also explore the ethical and practical concerns surrounding generative AI, with Lawson offering a thoughtful perspective shaped by his early work with IBM Watson.

Sheri Greenhaus: It’s been a little over two years since we last spoke, can you bring me up to speed on what Call Simulator is up to?

David Lawson: We’re very proud of our progress. We’re now approaching over 600 911 emergency dispatch centers training emergency dispatchers in the critical work they do. As I’ve said before, that market was always intended to be our launch point. From the beginning, we planned to expand into the broader corporate space—and that turned out to be a great decision.

Now, we’re not only working with Fortune 500 companies but also with several Fortune 50 organizations. And we’re doing it at scale.

What’s really exciting is how this applies across industries—everything from airlines and shipping and logistics to health and life insurance.

Sheri Greenhaus: When we last spoke, you described how your system provides scenario-based training—it can run at scale, monitor quality at scale, and even conduct evaluations at scale.

Looking at your website now, I’m not sure if this was available before, but it looks like you’ve added chat functionality. Is that new?

David Lawson: Yes, that’s right. We’ve added chat capabilities because we’re increasingly being used as a platform for communication training.

We started out more focused on call centers, and while they remain a core use case, our platform is now being applied to a wide range of communication types. That includes text-based support like live chat and social media—anywhere text is being used to communicate.

One of the important features we added was multi-chat simulation. That means we can simulate multiple chat conversations happening at the same time, which reflects the real-world experience of someone working in a chat support role.

People often ask, “Does it actually work?” And the answer is yes, as long as the practice is relevant and realistic, it absolutely works. If you can simulate the back-and-forth of real conversations such as handling curveballs like an angry customer or a complex situation, then, just like with anything else, you’ll start off slow, get better with practice, and eventually become much faster and more effective when you go live.

What’s interesting is that companies are starting to realize that when a real human gets involved, someone who can add empathy, nuance, or even identify upsell opportunities, the experience improves beyond what AI alone can provide.

Even if handle time doesn’t drop, or even increases slightly, you often see a big payoff in terms of higher Net Promoter Scores, improved CSAT, more revenue, and longer customer relationships.

Initially, the expectation was that bots would take over service. And at first, it looked great from a financial perspective—no salaries to pay. But the result was frustrated customers who didn’t feel heard or helped. Over time, companies realized that whatever they were saving with automation, they were losing in customer satisfaction, and worse, in long-term revenue.

Sheri Greenhaus: When we run webinars featuring companies talking about AI, there’s a consistent message, especially from those in sensitive industries like financial services and insurance. A bot can easily answer simple questions like “When is my payment due? and that’s fine. But when it comes to more sensitive or complex situations, you still need a human agent.

More and more companies, and end users, are realizing that bots can’t handle everything. So now, they’re more strategic; reserving their live agents for calls that are emotional, complex, or require deeper engagement.

David Lawson: Absolutely, 100%. And what that also means is that the so-called “easy” calls are now handled by AI, which leaves the human agents dealing mostly with the difficult ones.

Often, by the time a customer reaches a human, they’ve already gone through AI prompts, possibly feeling frustrated. They may have even had to convince the AI to escalate the issue. When the agent picks up, the clock has already been ticking from the customer’s point of view, even if the human just joined the conversation.

That’s why many advanced platforms are emphasizing their ability to transition customers seamlessly from AI to a human agent, providing the agent with a full summary so the customer doesn’t have to repeat themselves. That makes a big difference.

But with this shift comes a greater need for simulation-based training. One of the worst things companies can do is throw a bunch of advanced tech at the agent and then not train them on how to use it. There’s this flawed idea that with all this smart technology, training becomes less necessary. It’s actually the opposite.

Flight simulators exist not because pilots aren’t smart, but because the cockpit is complex, with countless scenarios that could unfold. The same logic applies here: the more tech we introduce to assist humans, the more essential it becomes to let them experience it in a safe, simulated environment, where it’s okay to make mistakes and learn from them without real consequences.

That’s another challenge we see: some organizations expect new hires to perform at full capacity on day one, without acknowledging the need for time to learn the tools, terminology, and unique aspects of the job.

We’ve learned a lot from training 911 dispatchers, where mistakes can mean life or death. Their work requires intense active listening, communication, and multitasking and lives depend on how well they perform. In the corporate world, the stakes may be different, but the principle is the same. It’s not about literal life or death but brand damage, or lost revenue.

No matter how intelligent or experienced someone is, they still need structured practice to succeed.

Sheri Greenhaus: It’s interesting, we’re hosting a webinar at the end of this month, and I asked the participating vendors to send over some bullet points on what they planned to cover. Not one mentioned simulation or training.

David Lawson: That doesn’t surprise me. There’s often a disconnect. A lot of focus is placed on things like real-time agent assist tools that support agents during live calls. And yes, those tools are impressive. Companies also invest heavily in quality systems that evaluate performance after the fact.

The typical process ends with something like, “Watch this video and try to improve,” which really isn’t effective. That’s what happens when you place quality at the end of the process and hope it somehow influences performance at the start. It doesn’t work.

Real training means giving agents the opportunity to actually do what’s expected of them, using the same metrics they’ll eventually be judged on. It means giving them the chance to build experience in a safe, simulated environment before they go live.

Then, when quality systems detect issues, you can say, “Looks like empathy is a challenge, go practice with these scenarios designed to improve your empathetic communication.”

A lot of CCaaS (Contact Center as a Service) companies have poured money into end-of-line quality measurement and agent assist tools, thinking that would be enough. They overlooked training.

When I first entered the call center space, I was surprised by how many of W. Edwards Deming’s classic management principles had been forgotten, especially things like:

• Principle 6: Institute training on the job.
• Principle 13: Institute a vigorous program of education and self-improvement. 

Sheri Greenhaus: I started CrmXchange nearly 30 years ago. Back then, training, and ongoing training, was like gospel. It was just something every company did.

Years ago, I managed an outsourcing business and we were constantly training our staff as the technology evolved. I think you’re right: somewhere along the way, people started believing they didn’t need as much training because agents could just press a button and get the answer.

David Lawson: Exactly, and it’s unfortunate. That shift makes sense in context. When I entered the industry, it was right around the time these advanced tech tools such as real-time assist, automated quality monitoring, etc., started coming into the picture.

But the irony is, the more technology you layer in, the more skilled the human agent needs to be. They’re no longer just answering simple questions. They have to operate at a higher level to make the most of these tools.

Sheri Greenhaus: How are you partnering with CX vendors?

David Lawson: We can integrate our simulations into any quality system. From the beginning, we designed our platform to be open, so it’s fully built with APIs and deep links to support those integrations. That flexibility is key. You don’t want your learning systems to be isolated silos, especially in large-scale organizations.

Sheri Greenhaus: Are you expanding within end-user organizations?

David Lawson: Absolutely. We're now being brought into communication training well beyond the traditional call center environment. In many cases, there's no texting or phone involved. It's live, in-person interaction. That could include leadership development, customer service roles like gate agents, or other face-to-face scenarios.

This has really expanded our use cases. Call centers will always be a key part of what we do, but we’re seeing broader adoption across departments. And for that reason, one of the most significant features we’ve introduced since we last spoke is AI coaching.

We’ve incorporated generative AI to create skills-based, rubric-driven scoring, in addition to the compliance scoring we’ve always offered. In call centers, we aim to integrate with whatever quality system they already use, but there are many departments and employees across large organizations who don’t use, or don’t have access to those systems.

By having our own built-in AI coaching, we can support both ends of the spectrum: the entire organization or just specific parts that lack an existing quality infrastructure.

We’re really proud of that development. More and more organizations are moving toward skills-based evaluation, and rubrics offer a far better measurement of effectiveness than a simple binary: “Did you say the right phrase or not?” The rubric-based approach allows for more nuance, context, and, ultimately, a more accurate view of performance.

Sheri Greenhaus: And I suppose when you're working in the contact center and then extending into other areas, simulations can help ensure consistency. For example, a gate agent wouldn’t be saying something completely different from what a customer just heard from the contact center.

David Lawson: Exactly. Maintaining a consistent corporate message across all touchpoints is crucial. In fact, human interaction is becoming one of the most important ways for companies to differentiate themselves. As AI becomes more common, many customer experiences will start to feel the same. When a real person enters the equation, that moment becomes a brand-defining opportunity.

That human contact is where companies can reinforce what makes their brand unique; whether it's quality, price, comfort, or something else, and also where customers can feel truly heard. And that’s one of the biggest limitations of AI. Even in industries where AI typically works well, if a customer is upset and just wants to be heard, AI will never fully meet that need. No one believes AI is truly listening. That’s a gap, and it opens the door for competitors who do offer real human connection.

Sheri Greenhaus: Yes, although this is a bit of an aside—but what about AI “boyfriends” and “girlfriends” that make people feel heard?

David Lawson: (Laughs) Right! That’s a whole other issue. But honestly, it worries me. When people start to believe those interactions are real, it becomes dangerous. AI is, by nature, a sociopath. It’s not trying to help or hurt, it’s simply responding based on data. It has no values, no soul. It will do the same task flawlessly at 2 AM or 4 PM, and it won't complain. But that’s also the danger. It may do something amazing one moment, and something completely inappropriate the next.

That unpredictability is like dealing with someone who is kind to one person and cruel to another. Psychologists would say that's because there's no consistent core, no person behind the actions. As someone who was at IBM Watson in 2013 and 2014, at the start of commercial AI, I’ve seen what it can do and what it can’t. Back then, we were one of the first companies allowed to use Watson, and I was there at the headquarters when it launched. That was the beginning of commercial AI.

Since then, we've moved toward more flexible platforms like Google’s, because we needed to scale. But those early experiences taught me a lot about what AI is and what it isn’t.

The real shift came with generative AI. It’s a complete game changer. But I think it’s already being oversold, especially in more nuanced human situations. And we’re beginning to see the backlash.

What could truly change everything is quantum computing. I’m old enough to remember when the Pentium chip revolutionized computing and gaming. Quantum will be that kind of leap. Its ability to hold both “yes” and “no” simultaneously, something current computers can’t do and is deeply human. It’s how we can have two favorite restaurants or hold conflicting emotions at once.

If AI is ever going to feel more human, more capable of empathy and nuance, I believe quantum computing will be the spark that gets us there.

Sheri Greenhaus: Do you worry more about AI becoming too human, starting to “think”, or that it stays essentially a giant database, just doing what it thinks you want it to do?

David Lawson: What worries me is something we’ve seen before in history: the technology advancing faster than the understanding of the people who created it. When that happens, bad actors can take advantage.

In the past, tech mostly helped humans do tasks faster, but the human was still in control, pressing a button, giving a command. Now, with generative agents, we say, “Go do this,” and it just does it. I was joking recently that maybe the Roomba was our first warning. It was a vacuum, yes, but it actually did something without us there. It was tech that took action on its own. Gen AI is that on a whole new level.

And that’s the concern. With a Roomba, maybe it bumps into your cat. But with a generative agent, you’re handing over access to your systems, your passwords, your sensitive data. What might it do with that? We’re already seeing stories where agents are being infiltrated and manipulated into giving up data to third parties. That’s why I love being in the training business. We work with simulated environments and fake data. It keeps the risk low while we build skills and understanding.

Sheri Greenhaus: In the last 30 seconds, any final thoughts?

David Lawson: If I think back two and a half years ago, most of the market didn’t even believe scalable role-play training was possible. What’s changed is the level of understanding. Today’s customers are much more informed. They already know what can be done. Now, they’re focused on how we’re executing it.

I’m excited. We’re seeing demand across industries and use cases. We’re past the novelty stage. It’s no longer, “What is this?” Now it’s, “How can we integrate this into our operations?” And that’s a great place to be.