Insight ON AI Didn't Replace These Workers — It Gave Them Their Mission Back

Audio transcript:

AI Didn't Replace These Workers — It Gave Them Their Mission Back

Jillian Viner (00:00):

Well, Chad, welcome.

Chad Brothers (00:01):

Thank you. It's

Jillian (00:01):

Nice to have you here.

Chad (00:02):

Pleasure.

Jillian (00:03):

Before we get started today, I just want to ground the conversation in what Viscommunications does and what your role is.

Chad (00:10):

So VizCommunications provides call center services for large carrier customers. We have two segments of the business. On one half, we do operator assistance and agents that provide information for toll numbers, so on and so forth. And the other side of the business, we do emergency services programs where we support carriers that need default routing or overflow routing to centers to help 911 callers reach their destination. And then we provide AI agents for non-emergent type calls.

Jillian (00:47):

I think your story is so fascinating because for a lot of us knowledge workers, desk workers, there is a pretty urgent sense of anxiety around AI coming after our

Chad (01:00):

Jobs. Sure.

Jillian (01:01):

And your story is unique because it kind of flips the script there where AI actually helped with those desk workers.

Chad (01:09):

Yeah, it does. I would say that I can't say that we're immune to that idea too. There's still the idea that our agents are going to be replaced by AI. They're human

Jillian (01:23):

Agents.

Chad (01:23):

Right. Yeah. Thank you for that clarification. Yes. But yet we find that really the pairing of the two tend to create such a better outcome. For many years as AI has started to creep into our industry, which tends to lag in technology, that's been one of the themes that we've had to really be careful about as we introduce the technology and saying, "We need to really focus on this as a supplemental support technology for our folks."

Jillian (02:02):

Help us understand the scale of this first. So when you're looking across the clients and really like the end user that you're serving, what is the problem that you're actually trying to solve and what is the impact on the communities that they serve?

Chad (02:17):

That's a great question. I'll use one of our most recent applications. So a lot of the work that we do is in public safety. And for the last several years, it's been a pretty big theme about staffing issues. And in the public safety telecommunication space, there's just been a struggle to maintain adequate staffing to support the need. And so to kind of tie that to the criticality, when you call a 911 center, you expect to get a response. And these centers are critically understaffed and have been for many years. And so when we started to look at ways that we could apply technology, AI really kind of came to the forefront and what we could do to help those centers, like bright size staff. Maybe they didn't need more staff, maybe they just needed the staff to be able to focus on their primary mission and let AI do something in the background to help support them in that role.

So it was finding the right technology at the right time to meet the need.

Jillian (03:31):

Do you have a sense of why this wasn't solved before? What had been tried before that didn't work?

Chad (03:39):

Previously, I think the focus within this space had been on policy, right? Policy, can we change process our way out of this? Can we just modify a couple of things? Can we loosen the standards for how we hire? But the reality was that wasn't changing. It wasn't moving the needle whatsoever. And it took, I think, technology to catch up. So I would say our industry started to see this problem in 2017 and start to raise it to standards and working groups to say, "Hey, something is not right here. We can't meet the need. Our volumes are increasing, but we can't keep people employed. We can't hire our way out of this. So what's next?" And I think it took a few years for AI applications to be ready and available to really to be almost plug and play in the space, if you will, to meet that need.

And so it took a couple of years of really soul searching, if you will, to find the right applications and for those applications to be mature enough to start to bring them into the space.

Jillian (05:00):

Humanize this for a moment with me because you talk about overstaffing. What is the impact or what was the impact of that for, let's say, a caller and then also the person who is on the other line receiving that emergency call or non-emergency call, depending on what the situation is.

Chad (05:17):

Yeah, let's dive into that. So on average annually, there are about 240 million 911 calls that happen in the United States. Those are 911 calls. That doesn't account for about 60% of the calls that show up in those same call centers. So if you think about your community and it has a 911 center, only 40% of the calls that they're answering are 911 calls. The other 60% are calls that are ostensibly services that were previously handled by agencies or directorates or departments within your community, but just no longer have staffing and resources to answer those questions. Think about- Like

Jillian (06:05):

There's something in the road.

Chad (06:06):

Dog catchers, road obstructions, public works issues. 911 centers are answering those calls around the nation 24 hours a day because there's no one else to take those calls. So we've shoved all of that work into this place where these highly trained people to deal with emergencies are now answering non-emergency calls. Yeah. So it's

Jillian (06:30):

A massive

Chad (06:33):

Mismatch of skills. So if you think about the mismatch of that, so now we've got folks, you've got race car drivers that are driving taxis around as an analogy and back and forth all day long. So there's a human toll on the telecommunicators that are answering those things, but there's a toll in your community. If you think about it, if I have to pull a telecommunicator off of the queue of answering calls to handle a dog catcher or a pothole or something of that nature, I'm taking them away from the queue of services. Now, I mean, obviously 911 calls are going to get priorities, but nevertheless, you're still taking a toll on the telecommunicator. You're still pulling them away from their primary mission, which is first response. And so what we and others in this space have proposed to do is let's use AI to address that 60%, right?

Jillian (07:36):

Yeah.

Chad (07:36):

So

Jillian (07:36):

Tell me, what did you build? What did you actually do?

Chad (07:38):

So we used Google Cast platform and conversational agents to start addressing those non-emergent calls, right? Flow the calls through an AI agent, use generative AI to create appropriate responses to caller inquiries based on information that you've trained this agent on and do your best to avoid getting those calls to human telecommunicators. So if I can provide a full-throated response with an AI agent on what someone needs to do to pay a parking ticket, I don't need a human to answer that conversation. Let that human answer the call about a heart attack or a injury traffic accident. Let's use AI to answer all of the other stuff. And I think that there's a ... I mean, we've seen great success in the space. We've seen adoption pick up in this space. And so that's just a fraction of what is capable of doing. And I think that was one of the things that opened up the floodgates for adoption of this technology, even greater adoption of the technology within the space.

Jillian (08:56):

So how does that work? If I call and I'm calling with a heart attack, who's answering that call? Is it an AI agent or is it human?

Chad (09:03):

Well, that's a great question. If you dial 911, almost exclusively, it's going to be a human, and it should be at this point in time. Those lines are routed and appropriately queued to human telecommunicators. Those that have, let's just say, the ability to do CPR on the phone or trained medical response on the phone, and that's where we want them to go right now. But if you, let's say you don't dial 911, but you dial an administrative or a non-emergency line at a 911 center, which I'm not aware of any center in the United States that doesn't have a non-emergency number, you may get an AI agent now and you may call about having a heart attack. And if you do that, that AI agent should be trained to say, "Wait, just a second. This isn't something I'm qualified to respond to. I need to get you to a human and therefore we'll do that transfer over." So there's safeguards in place to ensure that we keep the AI agent in the box that we want them to, only responding to things that we feel like they're technically capable of responding to, everything else should be escalated to a human agent.

Low risk

Jillian (10:22):

The agent can handle.

Chad (10:24):

Yes. Yeah. Let's talk

Jillian (10:24):

About the adoption piece because you said at the beginning that this is an industry where technology adoption change is historically slow- Very

Chad (10:33):

Slow. ...

Jillian (10:33):

Yet you've seen pretty good adoption. Explain that paradigm.

Chad (10:40):

I'm actually kind of shocked myself, to be honest with you. Within our space, let's just say for the last 15 years, we've been trying to adopt a new framework for technology that enables things like AI to become more ubiquitous. Yet AI came along and within two years, there were agencies across the nation that were adopting it in one form or fashion. Whether or not they were adopting conversational agents for non-emergence calls or whether they were adopting agents that were doing QA and training or agents that were performing psycharts and protocols, the level and the quickness at which folks realized that there was an actual tangible benefit for them was surprising to me given our typical apprehension to adopting technology. And quite frankly, the only thing that I can explain it is that the industry was so hungry for something that would help them, their daily lives, making their workload much more manageable, being able to actually meet the expected metrics.

When we think about being able to answer a certain amount of calls within a certain number of seconds, being able to do that and do it consistently, I think that they were so hungry for tools that would allow them to be successful, that they were willing to make slight concessions that they previously hadn't. And when I say concessions, what I really mean is that within the public safety communication space, there's almost zero tolerance for failure just because of the stakes being so high. When you think about someone calling with an emergency, your technology has to be bulletproof, right? You don't want to miss a call. You don't want to drop a call. You don't want to have a failure in that process. And so that and being willing to say, "Well, we know that we are not meeting the mark now without the technology. We know that the technology probably has room to grow and get better, but if we're not part of the conversation, then perhaps we're never going to bring it into our centers and we can do so much more now with it than we did if we didn't."

Jillian (13:07):

It's almost like the desperation created this space to be willing to try something different.

Chad (13:13):

I think that's a fair statement. Folks were hungry for change and change wasn't moving fast enough.

Jillian (13:21):

And we always joke in the business world that things are never really an emergency. We're not saving lives. I feel like your situation is kind of dramatizing what is true for the business world, which is you have people who are trained to do highly skilled work that are getting sucked into work that is below their skillset, meaning in your terms, you've got people who are trained to do emergencies. We have people who've trained to do really good strategy or what have you, relationships, but

Chad (13:50):

They're

Jillian (13:50):

Getting sucked into the mundane work. When you relieve that mundane work, it frees your head space. It frees your emotional space, your energy of focus on the work that really you should be doing.

Chad (14:03):

Yeah. Yeah. Well, for me, it's like context switching, right?

Jillian (14:07):

Yeah,

Chad (14:07):

That too. As humans, we're not really meant for that. Even though we shove ourselves into that box, it's difficult. We don't perform well. And to your point, when we say, "Hey, I'm going to take all of this mess off of your plate. I just want you to focus on the one thing that you do good and get better outcomes." I think that we are well on our way to providing space for folks in the public safety industry to do that. And I think that it's also acted as a springboard for the willingness to explore more. What more can we do with this? We're just scratching the surface on it, but I think that it enables folks to realize that they are also a part of the conversation. They're not just getting technology shoved at them. They're a part of the group that's helping find the solutions by leveraging it.

Jillian (15:04):

Final thoughts. For organizations that aren't truly saving lives, what have you learned through this experience that you think is a tangible, transferable lesson for others who recognize they need to do AI, but haven't quite figured out how we'll win?

Chad (15:20):

Well, one, I think the best way to do that is to literally step right in the deep end to some degree. I'm not talking about completely transforming your organization to an AI first company without really thorough and deep thought, but I do think that there are probably dozens of ways to step in and find operational improvements in your organizations. One of the things as we were developing some of our products, one of the things that I started doing within our own internal customer base was to start deploying small agentic tools that didn't require a lot of lifting, didn't have any risk and outcomes, but it also got the exposure to the stakeholders so that they knew that they were part of solving the problem that maybe that they had. So empowering the stakeholders within the organization is a way of introducing the technology so that they could see that there was something there for them that was going to make their lives easier.

And from there, it just really acted just like as a change agent within the organization itself. So even on our customer base, I empower them, I encourage them, go find an AI tool and experiment with it. You don't have to completely change your organizational process, but I can guarantee you, you will find something that will make your daily work easier and allow you to focus on something that maybe is more passionate for you.

Jillian (17:05):

Yeah. Can you give me one concrete example, one of those low stakes agents that you've made?

Chad (17:12):

Yeah. Yeah. So one of the things that we did was when I stepped in the door at Viz, the first question I asked was, how often do we do quality assurance for our human agents and what does that process look like? And we probably do, let's just say 150,000 calls in the emergency space a year. And I'm sure it's higher than that now. And then I asked, "Well, how many of those calls do we actually review? Did we meet the mark on them?" And the number was like one and a half percent. And we had four quality assurance people that were doing that. I saw your eyes go up and that's the same thing I did was how is that possible? And so the first thing that we did was,

Why aren't we transcribing all of these calls and why aren't we doing an initial cursory QA? Ask it five questions. What are the most important five questions that we have? And we set that up within a week and a half and 85% or more of our calls now get an initial AI cursory QA. But not only that, all of our QA personnel are able to do 2X, 3X of the work that they were doing before because they have transcripts and they already have an initial assessment on the call. Did we meet these five things and what the percentage was? And we can customize that for any of our customer protocols. So just in a matter of days, we 3Xed the output of one of our most important staff groups and they didn't even know it was possible.

Jillian (19:01):

And what was the impact of that? What did that actually result in?

Chad (19:03):

Well, one, our agents are better trained now. Two, we can say to our customers, we are paying attention to how well we're doing on your calls. These are the metrics and these are the things that we're improving. So we improved our training, we improved our throughput, we have a better story that we can tell when we go out to our customers and we want to expand our business. It's just been like a force multiplier across the board.

Jillian (19:31):

Amazing.

Chad (19:32):

Yeah.

Jillian (19:33):

Makes sense and so easy. You just looked at one little problem.

Chad (19:36):

Right. And that's how you step your way into this, in my opinion. Just start to find the areas within your processes that feel like they can be better. Find an agent that can do that work and you've kind of sprung board your way into a process automation, a Six Sigma process, if you will, operational excellence. It's all there. It's just, it's ready for the taking.

Jillian (20:05):

It sounds like a great piece of homework for anybody listening right now. Just go and identify one friction point in your operations.

Chad (20:12):

Yes, absolutely. And then find an agent that can do some of that work for you.

Jillian (20:18):

Build it.

Chad (20:18):

Go build it.

Jillian (20:20):

Chad, thank you so much for your conversation today. I learned so much. It was really helpful. Thank

Chad (20:24):

You. Thank you. Appreciate it. For me, having insight resources available and engaged with me on a constant basis is really, it's been a force multiplier for the work that we try and do. I have access to finite resources with inside my organization. And when I want to deliver something fast and I want to deliver something right, I pick up the phone, I get on a chat with my insight team and we find a way to get to the other side of that. And they've always been great in providing me creative solutions that are sustainable.