Home > Columns > Executive Interviews

Talkdesk Executive Interview

How Talkdesk Is Transforming CX Through AI and Vertical Expertise

Sheri Greenhaus, Managing Partner of CrmXchange, sat down with Talkdesk CMO Neville Letzerich at CCW Las Vegas. Together, they explored how AI is reshaping customer experience across industries. Their conversation reveals where the market is headed — and what enterprises must do next.

Sheri Greenhaus: Let’s catch up and talk about what’s new at Talkdesk. I’ve known Talkdesk forever, of course, but I really want to focus on the newest developments.

Neville Letzerich: Absolutely. Talkdesk has always been a disruptor in this space. CCaaS is still a strong market, but it’s changing quickly. You’ve got CRM vendors coming in, conversational AI vendors, traditional CCaaS and UCaaS players. All these markets are converging. Everyone is trying to figure out how AI can automate parts of the customer experience without breaking it.

We’ve all dealt with robotic voices for years, and no one loves them. People still hit zero to get to a human. IVRs aren’t going away anytime soon, but nobody enjoys pressing “2” or “3” to navigate menus.

We saw an opportunity: bring humans and AI agents together as a team to deliver exceptional customer experiences. What makes us unique is that we have deep CCaaS expertise and a natively built AI platform. We didn’t bolt on acquisitions or stitch together multiple systems. Our code base is unified, and that gives us a very different advantage.

One of our biggest recent announcements is our AI Agent Builder. We don’t think customers should have to become prompt engineers. Sure, anyone can write a prompt, but who wants to spend hours refining and testing them?

Instead, we let users describe in plain language what they need: “I need an agent that does this, checks that, and brings this information back.”

Behind the scenes, we generate the prompts, test them, look for errors, run scenarios, and then allow deployment. We’re making AI accessible to business users, not just technical teams. And we’re listening closely to what customers actually need.

Another major advantage is that we sit on top of all the CCaaS interactions. We understand what customers are trying to do — and we’ve built deep industry‑specific solutions that automate those top use cases. 

We’ve been focused on industries for a long time. Take healthcare: we know the top 50 patient use cases a provider typically needs. So instead of starting from scratch, we can sit down with a healthcare organization and say:

“Let’s pick one use case. Let’s make that successful. Based on our experience, here’s the order in which customers succeed. We’ve prebuilt the workflows, agents, and templates to get you there faster.”

That approach has been incredibly successful and is a major differentiator, especially with integrations into core systems. The same applies to financial services and other verticals.

The AI market is changing fast. It’s our job to stay ahead of the technology, but it’s also our job to make it digestible and usable for customers. They don’t need a blank slate; they need a real starting point, not a marketing demo.

Over the last 12–18 months, there was a lot of pressure to “just use AI.” Everyone rushed in. Then they realized: “Wait: if we’re just generating tokens, we’re getting big bills and not seeing value.”

So now C‑suites are saying: “Hold on. We’re going to use AI, but we need to use it the right way.”

Some organizations had false starts and now feel hesitant. Others are succeeding because they’re focused on value‑add use cases. They ask: “What’s the problem we’re solving? 

Did we achieve it? What did we learn?”

You have to start small. What’s your biggest issue? What’s the pain point? Solve that first.

Sheri Greenhaus: During our webinars, we always ask attendees: “What’s making you crazy?” That’s where the real issues come out.

Neville Letzerich: Exactly. One of the challenges we saw early on is that AI projects can become massive. You get to the end and realize the customer didn’t actually know what they wanted. We made that mistake ourselves at times by trying to say “yes” to everything and letting the scope get out of control.

Now we start by containing it. Let’s get a win. Because you have to understand:

● How does your organization adapt to change?

● How do your employees adapt to change?

● What systems will be touched?

● What needs to be learned along the way?

If you don’t narrow the scope, you won’t get the best outcome. You won’t look like an instant hero… but you will look like someone who delivered a real win, learned from it, and built a foundation for the next step.

There’s a balance: you need to go fast, but you also need patience. You have to run both tracks at the same time. That’s why clarity on the desired outcome is so important.

There’s also fear. People hear “just do it” and think, “Great, now AI is coming for my job.” There’s pressure from the top, and in many cases, executives haven’t given their teams enough direction. So, teams are left trying to figure it out on their own, and that’s tough.

When you start from scratch without knowing the application whether healthcare, financial services, whatever it is, you’re flying blind.

Sheri Greenhaus: You’re giving them a base to start with.

Neville Letzerich: Exactly. And that’s been a huge advantage for us. There are a lot of vendors out there, especially AI newcomers, trying to be everything to everyone. It confuses the market.

Some of these companies appear, make noise, and then disappear. And one of the biggest issues they overlook is data structure. It’s probably the least glamorous part of AI, but it’s critical. Your knowledge management and knowledge interfaces must be clean. If they’re not, it’s garbage in, garbage out. AI will make bad decisions based on bad or conflicting data, and then you’re running your business on top of that.

No one loves knowledge management projects. They’ve failed many times in the past. Data gets scattered everywhere. That’s why we go back to the idea of a pilot.

Start small. Clean the data for one use case. Learn what it takes. Keep it contained and tied to a specific outcome. If you have the patience to do that, you can move forward with confidence.

For many companies, it’s the first time they’ve really looked at how dirty their data is. We’ve had painful lessons where customers thought their data was clean. It wasn’t. When you scratch the surface, you find agents have been working around conflicting information for years.

But once companies take the time to clean it up, both human agents and AI agents perform better. They build mechanisms to keep the data current. It improves the entire organization.

If you look at customer experience, the contact center is critical, but it’s not the only touchpoint. Many other parts of the business rely on that same data.

One of the solutions we’re bringing in is a branch banking solution. Think about someone walking into a branch and having a conversation. Why wouldn’t that branch employee have access to the same clean, consistent data as the contact center?

The goal is for every employee, whether in a branch, a contact center, or anywhere else, to see the exact same data. If you and I are having a conversation in person, that information should be entered into the system correctly so that when you later call the contact center, the agent can say:

“I see the conversation you had earlier. Let me help you pick up where you left off.”

We’re looking at all the different places where conversations happen and bringing them together. Think about an emergency room visit:

● Did you schedule the MRI?

● Did you pick up your medication?

● Did you complete the follow‑up steps?

When you look at the full end‑to‑end experience, it’s not just: “You called the contact center, we solved your issue, check the box.” It’s about the entire patient journey. And that’s where AI agents become incredibly valuable.

If I talk to Susan in the contact center, she might be wonderful. If I get her again, she might remember me, or review notes about me in the system. But an AI agent will always know who I am, what I’ve done, what I asked for, and what I still need. There’s real value in that continuity.

This also changes how we think about cross‑sell and upsell. It’s irritating when you call a telco provider just to understand your bill and they immediately start pitching internet, TV, or mobile plans. You just want help, not a sales script.

But imagine if the system actually knew you:

● What you already have

● What you recently asked about

● What you’re trying to solve

● What might genuinely help you

Then cross‑sell becomes relevant instead of annoying. It becomes: “I see you recently upgraded this service, here’s something that complements it.”

That’s a completely different experience.