From crisis to control: Rethinking 9-1-1 dispatch strategies

Good morning, afternoon, and good evening. Thank you for joining TTEC Digital for our webinar from Crisis to Control, How to Rethink nine one one dispatch strategies. In today's webinar, we will explore the increasing challenges faced by nine one one emergency contact centers and the technology solutions that aid the call centers in handling non emergency, calls more efficiently, ultimately bringing improvement to nine one one wait times. We will take a look at the main topic. But before we do, we'll do a quick introduction of TTEC Digital. After the introductions, we'll take a closer look at the challenges, how AI is making an impact for nonemergency services, and automation tactics. We'll also take a look at the next steps depending on where you are in your journey. Partnership. So TTEC Digital collaborates with top tier technology partners to enhance our capabilities portfolio. This positions us uniquely to provide our customers with best in class c x services. Through our partnerships with AWS, TTEC Digital delivers reliable, scalable, and cost effective cloud based contact center solutions to drive the ultimate in customer experience. So we are a global systems integrator with established practices in North America, Europe, and Asia Pacific with over a million customer interactions supported by our deployed solutions, including our client, Amazon dot com. We are an advanced AWS partner and a part of a bigger team with over forty years of experience in the contact center industry. TTEC Digital's AWS practice, formerly known as VoiceFoundry, was the first signed Amazon Connect partner, reseller, and systems integrator in the AWS ecosystem. With over two hundred dedicated professionals focused on AWS contact center solutions, we have more engagements on Amazon Connect than any other partner in the market with expertise in Lex based automation and end to end customer experience with more in the CX realm. I would like to introduce you to our expert per presenters, Curt Hamm, senior solution architect for public sector, and Kelly Jacob, vice president of public sector sales from TTEC Digital. And we're excited to be joined by John Persano, emergency communications lead for AWS Justice and Public Safety from Amazon Web Services. I'm Lisa Colbert, partner marketing manager for AWS at TTEC Digital and your emcee for today's webinar. Now I will hand it over to Kurt and our panelists to lead the discussion on how AI is making an impact in nine one one emergency management. Take it over, Curt. Thank you, Lisa. In the mission to enhance emergency response, we face critical challenges. A key issue is the overwhelming number of non emergency calls estimated at sixty to eighty percent of total volume. This influx strains resources and diminishes service quality for real emergency, affecting the emergency call standards. We are moving toward quickly toward next generation nine one one to address these issues with advanced technology. However, this transition brings new challenges, such as information load for dispatchers, emphasizing the importance of not missing critical information. These challenges underline the urgent need for continuous investment in training technology and process enhancements to improve nine one one contact centers' efficiency and effectiveness. Facing these challenges head on is crucial to developing emergency response systems that are robust, resilient, and responsible to evolving public needs. I think you hit it right on right on the head. And really that, you know, that staffing issue, all these challenges, you know, have to be addressed by right now, they're being addressed by human beings. And how much of this can we take off that load of of the human being and still provide, obviously, the critical support needed on the nine one one side. So we'll talk a lot more about it, but, yeah, you hit them right on the head, Kirk. And then as we examine the landscape of emergency response sector, it's imperative to understand that there is a complex interplay between legislative developments and public pressures. In two thousand twenty one alone, we witnessed a significant legislative push with eighty four bills enacted across thirty seven states, target targeting critical aspects of the nine one one system, such as funding, telecommunicator training, and the integration of mental health crisis hotlines. The journey towards next generation nine one one is also not without its challenges. Mission critical partners reports that integrating these advanced technologies is being hampered by staffing shortages, highlighting the operational difficulties emergency call centers face. Public driven initiatives, such as the District of Columbia's implementation of a performance dashboard, stem from growing concerns over dispatch call handling, efficiency, and transparency. So this just piles on to the already stressful environment in the in the nine one one public sector. Okay. At the risk of turning this webinar into an all out AI discussion, we felt it was necessary to speak a little bit about the nature of AI solutions. AI encompasses a broad range of technology, including the latest in generative AI, such as OpenAI's ChatTBT, Google's Gemini, and, of course, Amazon's Bedrock. Beyond generative AI, our solutions also harness the power of machine learning and natural language processing. My advice is to understand what specific AI technologies are integrated into any solution you're considering. If you're expecting generative AI and the solution is geared more toward machine learning, you may be disappointed with the results. John, I know you know a lot about this. Yeah. So, you know, I we could turn this into an entire, you know, webinar just on these topics and these terms. I'm gonna give it very short and very broad brush, and we're gonna answer some questions later on. So the first thing to understand is all artificial intelligence is machine learning. You can't have AI without machine learning. K? It starts as machine learning, building blocks. And then basically, what you're getting is, more and more autonomous, you know, outcomes by whatever, machine you're working with. So a lot of the stuff we're gonna talk about today is very heavily in the machine learning side. And then we sprinkle in little little bits of artificial intelligence to help enable the human being do workloads faster, more efficient, more effective. So, you know, what we really wanna stress to folks is, you know, this isn't Skynet, Lochnet, you know, where the bots are gonna come and take your children and and and we're we're not there, folks. I know the media likes to hype that up that that's where we are. We're we're not. What we're focusing on is building blocks, for very benign, very onerous workflows. So how do I use this technology to do something that takes me ten hours, but it could take a machine ten minutes? Right? We are not trying to replace the human. Number one, there's no humans to replace, right? The staffing shortage, does not allow us to replace humans. And, you know, we're not gonna hire ourselves out of this problem. We wanna augment the wonderful people that we have, who are our first line of defense, and they are overwhelmed. How do we help make their life easier? One of the easiest ways to think about the progression is think about your photos. So for instance, I have three daughters, and they're all very close in age, eighteen, twenty, and twenty one. So when I go into Amazon Photos and, you know, I tag, okay, this is, you know, one, this is one, this is one, it get it gets confused. Right? I mean, everyone's done this where it mix and matches your family members or people that look so you have to go in and you have to say this is the right person. K? That's supervised machine learning. So you're telling the machine what is right and what is wrong. The machine is not just magically doing this and no. You have to train it. You have to teach it. In order to do that, you need large datasets. Okay? And one of the challenges that we're seeing in public safety is a lot of the datasets that are used for AI and ML are on the commercial side. So we need to teach the data, right? We need to scope the data and teach the machine how this data is important. And a really good example is the word good. So if I'm in retail, I may ask, was that a good experience on a call? K. There is no good experience in a nine one one call. Right? The word good is not is not applicable. So we have to teach the learning mechanisms what certain terminology is, what the aims are, how they how expected outcomes. So we need the large data sets. We need the supervised learning processes, which will eventually turn to unsupervised learning, which eventually turns into artificial intelligence. Then we need the basic understanding of what this is and what this isn't. It's not magic folks. It's not gonna solve all your problems with one button. Okay? It takes work. It takes time. And we're here to support you. TTEC and AWS are here to support you in getting to that. Lots more we could talk about, Kirk, but I don't wanna take up too much more time on this one, Tom. So integrated artificial intelligence into nine one one operations transforms how you handle non emergency calls. It directly empowers our workforce by automating the triage of calls. This not only optimizes response times for true emergencies, but also alleviates the stress on our employees contributing positively to their mental health. AI's capability to support multiple languages enhances accessibility, ensuring that no caller is left behind due to language barriers. Automate automated triage and predictive analytics for staffing enables us to efficiently prioritize calls and allocate resources. In essence, AI's role in non emergency services elevates our emergency response system's effectiveness and efficiency, marking a step forward in meeting our community's evolving needs. So we can dive a little bit more deeply into how automation and what kind of call deflection strategies are currently revolutionizing revolutionizing contact center operations. Of course, chatbots and virtual assistants are our first line of defense. They can handle common questions instantly so teams can focus on life threatening emergencies. Self-service portals allow people to help themselves get the info they need, cutting down on simpler calls and letting dispatchers tackle the more complex calls. Automated email responses. Everyone gets a quick we got your message email. Keeping them keeping them in the loop from the get go. Analytics and reporting, that's our crystal ball, helping us to see trends and plan better so we're ready for what's coming. Automated status updates. Keeping folks updated via text means fewer calls asking what's happened and where are we. And then proactive communication. When we reach out first with updates or alerts, it cuts down on incoming calls and keeps everybody in the know. Implementing these tactics isn't just about making our lives easier. It's about providing a smoother, more responsive service to everyone who reaches out to us. Thank you, Kurt. I do have a question for the group. In terms of the automation tactics and call deflection strategies, I guess, which ones are the emergency contact centers gravitating towards or taking the biggest interest in? I think initially, you know, anything with a with a chatbot and, a q and a bot, question and answer bot, is what we're seeing the most adoption. But it really comes down to what the workflows are. Right? Because some of them are gonna require data collection. Okay? And data collection, you know, via text or via chatbot is a great way to collect data. Right? Because I'm typing. I'm not using my voice. It's very hard for bots to understand people's inflections as far as spelling the right street name, the right last name, etcetera, etcetera. So a chatbot and a text exchange is very, very accurate for that. If I simply wanna ask what's the phone number for a police department, let's just go with voice, Right? Because that's a very simple thing that, the technology today can adopt. So I would say it depends, which I know is always a lousy answer. But the beauty of what TTEC is doing and and leveraging the the capabilities of AWS, Amazon Connect is, omnichannel. So it works in voice, it works in text, it works in chat. You don't have to recreate each one of those. You just build it in one, and it works in all three. So then the nine one one center has the ability to adopt that technology to the workflow that best suits their needs. Okay. Great. Thanks, John. Yeah. And I I agree with that as well, John. I think that, they're they're they're starting to become a life cycle for the implementation of these technologies in in nine one one centers, and it definitely starts with a chatbot or a virtual assistant. I mean, that allows you to do an immediate call deflection, give yourself immediately some breathing room, you know, by taking calls off of the dispatchers. And then once you have that breathing space, then you can begin to deliver more of these, type of strategies into the contact center. And, Kurt, if I could add one thing to that. You know, for for the group, this isn't, you know, Ma Bell's phone tree. Press one, four, press two, four, press three, four. This is an interactive engagement. Alright. It's it's much like you do in your car, right, when you're trying to get directions. It's much like you do in your house if you're using Amazon Alexa, the exact same technology. Okay. Lex, for a lot of it. So it is interactive. And that's a very, very key. You know, you're not press going through a front tree, pressing sixteen different buttons to get to possibly the right answer. It's it's immediate and it's interactive. Okay. Great. You know, as a consumer, I I often leverage these type of this type of technology when I'm reaching out to the my partners and people that I work with in terms of consuming. And oftentimes, I can't get to the right place, or it doesn't understand my prompts, or, you know, how does this technology differ from our traditional experience in leveraging IVRs? Yeah. Definitely from a technical side. We we've talked about, using AI and machine learning. You know, one example is for the bot, a lot of times what what you'll have to configure in the old days were keywords. So the bot would have to hear a certain keyword, before it would take you down a certain path or respond in a certain way. But with machine learning through the Amazon bots, you can give it utterances. You can give it a handful of utterances that kind of reflect a single purpose for that caller. And then the machine learning comes in where the the the system will look at those utterance you've provided and extrapolate. So if the if if the caller says something similar to that or something that that that the system can map back to those utterances, then it it understands what you're saying and knows how to respond. So it's a much more customized, I think, solution and experience for the caller rather than, as John says, press one if you want this, press two if you want that. You can literally say, this is what I want. And then with AI machine learning, it'll take what you said and sort of map it back to what it thinks you want to do, confirm that with you, and then take the proper action. It's a game changer. That is for sure. Alright. Great. Thanks, guys. On this slide, we're focusing on how streamlined post call workflows can significantly boost process efficiency. And what we've seen is, a lot of the post call workflows are manual. It takes a human to follow-up in a number of circumstances. So we think post call workflows is a real opportunity to improve process efficiency. As I mentioned before, automated follow-up. You know, if we can integrate this with a CAD system, we know we have an open case. We're getting updates to that case. Can we send out an automated follow-up to to to keep the constituent, you know, understanding where they are in the process? Satisfaction surveys, this is a growing, need that that we've been getting pinged on in terms of, like, if it's a sheriff's office and one of their their their their people get deployed, they want to know, did their people do a good job? So can we do an automated satisfaction survey and get that back to to the internal constituents so they can improve their processes and their procedures as well? Summarization is just is a critical capability of, of AI and generative AI. So can we quickly summarize exactly what that call was about? And can we do it on on a wide variety of calls, so we can drill down into the calls, understand why people are calling, and understand our response to them? And, of course, data analytics. That's again our crystal ball. Without data analytics, we really don't have a road map to to see how we're doing and then where we need to go. And a lot of sentiment analysis that gets complex, as John mentioned, because rarely in a nine one one call is somebody gonna be the sentiment's gonna be absolutely ecstatic. It's probably gonna start out pretty bad. We'd like to see it get better, but it's only gonna get better to a certain point. But there is the opportunity to measure, calibrate what sentiment we expect, and then measure that sentiment. So, we feel like a lot of these post workflows will make our processes more efficient, but also enhance the quality of of the customer service. And it ensures we're always learning, always improving, and staying connected with the customer's needs. Great. Thanks, Kurt. Kelly, I have a question. For the sentiment analysis, can you give us a little can you illustrate what that looks like for a customer and how the contact center can leverage that information for future Absolutely. Absolutely. Thanks, Lisa. So that's really taking a full reading of everything that's going on with the interaction. It's being able to read and understand through inflection and tone of voice of the different emotions that may be taking place here. Obviously, if a caller is upset or frantic or something like that, that's gonna become clear in the audible that the person is giving. So the sentiment analysis is able to read through the analytics and understand that and wrap that up so that it's giving a more clear, reading into what's going on in these interactions from a case by case basis. It helps us understand the emotions and attitudes of these callers before they're calling in based on the historical analysis of these interactions. Very nice. Definitely some key information for those call centers to understand and measure in the future. John, did you have anything you wanted to add in terms of the post call data and analytics? Is that something that our our customers are really being able to leverage? Absolutely. You gotta look backwards before you look forward. Right? And and, I think this is something that a lot of centers don't do, simply because they don't have the bodies and the technology to do it. So on the on the non emergency side, which is what we're focusing on here, very few centers have any analytics, the type of calls that are coming on on the non emergency side. So there's a lot of repetition, right? There's a lot of wasted time and wasted effort. So what we do when when folks deploy is, the first question we get asked is, how do I train the bot to answer the questions? And we're like, well, what questions do you have? And they're like, well, we don't know. Well, great. Let's let the bot run for a month. Let's just let it collect data. Kurt talked about utterances. Utterances are basically the subject of the call. So the bot will capture all those utterances and will tell you, hey, this is what your calls you're gonna have twelve percent of your calls are about toes, eight percent are about dogs, you know, four percent are about trash cans. And then you can start building workflows and questions and answers to address those. You're not just stabbing in the dark. You have hard data. You're making data driven decisions by somebody saying, well, most of our calls are about this. I respect the, you know, the experience of these call takers, but you gotta show me the data. Right? And that's what a lot of people are doing. So this post call analytics is absolutely phenomenal. And then we get to have another webinar about what you do with it. Right? Now that I know and, hey, can I compare every time I get a call about a light bulb going out in the park, does crime go up or go down? Right? So I have a non emergency number that affects an emergency situation. How do we start linking those together eventually? Right now, let's focus on what the calls are about and how we could build workflows and interactions to mitigate those calls. So our nine one one operators can focus on emergencies. Yeah. I that can absolutely make an impact, especially on holidays like fourth of July when you're getting a lot of callers telling in about fireworks, and those could you don't have to ramp up anymore on the holiday that it you know, it's a difficult day to staff in general, but now you can have a bot help assist in those phone calls. It's it's We actually Lisa, we actually had two centers that set it up solely for the fourth of July. And they took every complaint about or anything about fireworks, and it went to the automated system. And everyone went to Eamon. So and they stood it up simply for that reason. So it's a great example. And New Year's Eve. Those two days which are the two busiest days for nine one one was fourth of July and years. So it's a great, great option. Awesome. And to piggyback on that real quick that John pointed out, because that is a huge, huge area of need within these centers. I mean, we've talked to nine one one centers. They will go out and outsource a second call center just to handle the overflow of those firework complaints. That runs up thousands and thousands of dollars just for essentially a one to three day window of service. So by deploying these technologies we're describing here, you're able to recoup a lot of that cost back by just standing it up and automating it. And and recently with the, phone carriers going down and people only having access to the nine one one SOS, emergency, I imagine that increased call volume and that's not something that we can plan for. You know what I'm saying? So having something like this in place to help cover some of those calls, I think, would be, make a huge impact. Thank you all. So as we consider integrating advanced technologies and AI into the operations, several critical considerations and potential roadblocks emerge that we must thoughtfully navigate. First and foremost is the hesitation to trust AI. It is a natural concern given the shift from traditional human operating systems to automated processes. Building trust in AI involves a commitment to transparency and a demonstration of the technology's reliability and effectiveness. Engaging the public effectively is another pivotal access, aspect, as the success of new technology often hinges on public acceptance and support. This not only showcasing the benefits, but also addressing concerns and feedback from the community. Additionally, navigating the political landscape is an integral part of the process, requiring collaboration with political leaders and adherence to regulatory and compliance standards. John, I know you got a lot about this one. Yeah. Couple couple ones to hit on, is the hesitation, trusting AI. Right? You you know, take it slow. Right? Baby steps. Just just ease yourself into the concept. We get a lot of pushback on trying to take our jobs. Yeah. Nobody's trying to take anybody's job. Okay. Number one, there's not enough people working in the first place because of the staffing shortage. We're trying to augment, okay? We are focusing on non emergency. If you call nine eleven, you get a human being, right? You're not gonna get a bomb when you call nine eleven, you get a human, right? We wanna make sure that human can answer that call faster and organize a response faster. We do that by deflecting some of these noise calls, right? These non emergency calls. Public engagement is absolutely key. You have to educate the public before you adopt these technologies because they're gonna have the same questions that you do. And expectation management. It's it's hard, right? I could go online right now and I could say, hey, you know, type me a paper about the French Revolution. And boom, I'm gonna get a paper about the French Revolution, right? Hey. It's that easy. No. It's not that easy. Those models have been trained. People took a lot of time, effort, and money to train the models to be able to do that. We're at the nascent stages within the nine one one community of taking nine one one data and doing it with that. Nine one one audio is bad audio. It's not clean, crisp, clear audio. Right? We have a lot of work to do with that in order to make it as good as possible in order to leverage some of these tools. So and it's a cycle, right? We we take the tools, we clean the audio, the audio gets better, we're able to use more, new tools, cleaner audio, and we just keep making that loop. So lots of options, lots of opportunity. Take it slow, take your time, dip your toe in, don't go crazy. Bots aren't gonna eat you. Nobody's gonna replace your job. Wonderful. And what I one of the things, I think, in leading up to this discussion and preparation for our talk today was how we talked about how people can leverage AI in a lot of different capacities and they're gonna expect it everywhere. And I think that that kind of helps soften the blow when it comes to thinking about the hesitation around AI. So if they're capable of doing it on their own, why aren't why isn't the rest of the world adopting? It's not if, but when. And AI is definitely here, and it is it is making an impact. And so it's definitely a place where where we can make a change. And Lisa, real quick on that. I think it's really important what TTEC is doing here because all the stuff we're figuring out, we're doing on the nonemergency side. K? So if we do make a mistake, okay, hey, that dog complaint takes an extra hour. Okay. Fair enough. So we wanna learn on the non emergency side. So eventually, we can start applying technologies to the nine one one ecosystem. And I think that's very important for people to understand is we're we're starting in a very safe space to test and try these tools. So then, eventually, we could apply them. And and, Kelly, I don't know if you wanna add to that, but I think that's a that's a big point. No, John. You're you're spot on with that, sentiment. So, you know, we're not going in to try to automate any interactions for someone calling in for, someone breaking into their household. That's that's not the use case here. What we're focusing on are those who are calling in to nine one one but for their, neighbor's dog barking or something of that effect so that we can clear that those noisy calls out of the queue so that the nine one one operators can fully focus on taking those interactions so it doesn't take an hour and a half for that caller who's reporting a break into their house to get to an operator. Once the technology is fine tuned and everything like that, then we'll begin to look at incorporating it into more, sensitive areas. I think the, main purposes for this slide is to let people know that they're not alone. I'm gonna let, Kelly and John talk about their engagement with with the public sector in terms of nine one one operations. But, we are seeing lots of outreach and lots of centers asking good good questions and trying to get information on how they can implement technology to support their centers. Kelly, I I know you can talk about engagement. Why don't you go first? Sure. You know, engagement has been running very high from across the nation. I mean, this is definitely, an issue that is felt by nine one one centers from east to west, north to south. You know, you hear stories, you read the newspapers, you see them all the time about, wait times being just way too high for someone to get into a nine one one queue, and people need help. You know, staffing shortages have been mentioned numerous times throughout this presentation, and there just aren't enough people. That's a very high high emotional workplace. It is there's this it's a special person that's made for the nine one one contact center, and not everyone's cut out for it. So we need to be able to free them up to be able to focus on the, duties that they need to be handling. And we've had numerous use cases that have spun up through this, you know, wildlife divisions, standing these up for the firework displays that have been mentioned. You know, when weather issues are coming up, tornadoes, things of that sort, flooding. All of these types of areas are areas that we've been reached out to to help incorporate bot technology so that we can just stand up and automate a message saying, you know, if you're calling in for this, we already know what's going on. We're happy to text you over a form that you could fill out so that we could follow-up with you. And there's a lot of success going with those workflow automations. John, what else are you seeing across the space? Yeah. You know, everything you said, Kelly, I'll just hit on a couple success stories, you know, in the interest of time. We've seen, incredible success with adoption. So, one nine one one center, when combining Amazon Connect and and the use case with a home built portal, they are at fifty percent, five zero percent of non emergency call collection. That is, insane, right? That that is awesome. So that is calls that were redirected. A human being didn't have to answer, but they were still resolved. Right? So it doesn't mean like we just hung up. Right? It it means that they were deflected and either the bot or the portal was able to address. And what they're seeing is they're seeing the public learning. So they're actually starting to see a decrease in calls to non emergency and an increase in hits on their portal. Because people are learning, I don't need to call anymore. I can just go to this portal and get the answers that I need. That's exactly what we want. Right? We wanna teach the public where the information is and put a less burden on our nine eleven, personnel. Another one, a twenty one percent reduction in non emergency calls in a month. So one month after deployment, they're already sitting in a twenty one percent, reduction. The other one after two months, seventeen percent reduction. So we're averaging between seventeen on average, seventeen and twenty eight with, you know, a year up to fifty. So we're seeing a lot of success. We're very excited about it. And we look forward to more adoption. We look forward, of course, to, you know, our partner TTEC taking on this partnership of, building these solutions for, for the public safety community. It's interesting because as a as someone that would need emergency services potentially, if it's not nine one one, I don't know what else it is, and it would take me some time to figure out that other phone number to call. And I imagine that not everyone is as apprehensive as me in in not dialing nine one one. So I I think that this is definitely gonna change and make a huge impact on how we help the public, which is what nine one one contact centers are there for. Alright. Moving, narrowing a little bit of focus down to, TTEC Digital. TTEC Digital does have a non emergency chatbot solution, and when discussing the implementation of a non emergency chatbot solution, it's essential to focus on three pivotal areas, security, regulation, and the combined experience and expertise of TTEC and AWS. Our collaboration brings forward a solution that places a premium security, ensuring that all interactions and data handled by the chatbot are protected by the latest cybersecurity measures. This is crucial, not only for maintaining the integrity of sensitive information, but also for building trust with the public who will be engaging with the chatbot for nonemergency services. The collaboration between TTEC Digital and Amazon brings a wealth of experience and a proven track record of delivering cutting edge technology solutions. Together, these factors, security, regulation, and our combined experience, form the foundation of a non immersive chatbot solution that is secure, compliant, and built on a solid foundation of industry leading knowledge and technology. Kelly, did you have any comments about our awesome chatbot solution? Nothing specific. Just that we have seen great success in, whenever we've worked with certain local nine one one and, nonemergency contact centers within the different cities that we've been working with. We've seen a lot of success whenever we, collectively figure out a good let's start with one division. Let's start with let's pull some analytics out and some of your call stats and see what is up there in your top three of what's coming in that are being classified as non emergency that are coming into your nine one one queue, and let's focus on those. And let's start with the public engagement. Let's go educate them that we're gonna be doing this. This is how we're gonna be solving for this, and this is what you can expect of it. And I think something that's key to, obviously, what we're doing with the automation of the actual interaction is the follow-up education. So, you know, we're we're telling the citizen to reach out to this number for wildlife division, if you will, if it's an animal complaint. But we wanna make sure that that person gets some type of record or follow-up of what to do in the future in this case so that the condition behavior of picking up a phone and calling nine one one does not continue. And that's where we've built in other items such as SMS follow-up with the information that was provided during the call. You know, you could have a form or a link to the number, whatever the case is. So I think that post interaction education is absolutely key. Alright. So in terms of next steps, we understand that nine one one call centers are at different places in their technology journey. So I provided some resources to assist you in that journey. A recently published blog called Supporting Emergency Dispatchers with AI Reinforcements. This is a great place to start if you're early on in the process. I've also provided an ebook, The Do It Yourself Guide to Improving Your Contact Center with AI. This is another good place if you're still in research mode. And then lastly, TTEC Digital offers, a new service powered by AWS called nine one one auto direct service. This is an Amazon Lex chatbot designed to augment human dispatchers in handling non emergency nine one one calls. This is a great place to go if you'd like to start the engagement and find out how this service could potentially work in your nine one one contact center. So now we'll open it up for questions. We have about seven minutes to talk, a little bit more about some of the things that you're interested in. So we will wait for those questions to come in. So first question, how much does the a this AI solution cost typically? John? Hey. I'm gonna give you my my standard answer. It depends. And the reason why it depends is because there's a couple factors in. There's a telephony aspect of this, right, because this is a phone call. So there's telephony charges. And then there's the services charges. And then there's the work charges for, you know, TTEC. And obviously, I'll turn it over to Kelly or Kurt, on the other side. So I would say this, there's there's some public pricing, but I would say this. When we look at pricing, what we wanna look at is, what am I getting for the price? And we base everything on headcount. So if we look at it, I'll give you one quick example. A hundred thousand calls at three minutes each is three hundred thousand minutes. It takes about, two point six human beings working forty hours a week with normal vacation days, etcetera, to answer three hundred thousand minutes worth of calls. So we take what would be the human capital. And then we show you what the cost is, you know, on the telephony charges, etcetera, etcetera, and have PSAPs make an informed decision as to whether that offset is worth it. It is. It's it's it's a order of magnitude less than what it would be, to pay somebody to answer non emergency calls. But it depends on call volume. It depends on the sophistication of the system. So Kurt or Kelly, I'd rather you guys talk about this, you know, distress what you're doing at TTEC. Sure. And thanks, John. And I'll turn it over to Kurt for the more, getting into the details of the different elements that are included, but we do have a package service offering that we've based, and we've based that completely on the many conversations and discussions we're in with different nine one one contact centers across the country. And what we found is kind of a first place to start whenever we kinda break out that first use case depending on what, interactions are coming in and clogging up your nine one one queue, and we'll focus on there. But I do wanna kinda asterisk the details with of course, there are different elements that can certainly increase, decrease the pricing as well. I mean, what kind of integrations are you all gonna be looking for? What does it look like to integrate with your CAT system? Is it open? Is it not to integrate with? So a lot of those different, details will come into play. But Kurt can go through the details of our package service offering. Yeah. Definitely. What we what we've what we've tried to do is take to heart what John preaches, which is, you know, start small and build from there. You know, we've talked about reputational risk. We've talked political risk, constituent risk. You want something that you can implement, then you can measure the impact of that. You know, get your twenty percent. You know, if you can get your twenty percent call deflection in a month, that's awesome. You know, then then you know that it's working from a call deflection standpoint. Nobody's really complaining about it. You're not in the front page of the newspapers. You've had a successful launch. So, you know, that base system that we like to implement, definitely will respond with voice. It will it will take a voice call. If it has one question, one answer, it will respond with an answer, or any other information that it has in in in its database of, responses. It will transfer a call. So if it needs to be transferred to another constituent agency or another department, we can handle that transfer automatically. We don't have to have a human involved. And as Kelly mentioned, we like to send SMSs, either one way with the information that we provided so you have it on your phone for future reference. John mentioned two way SMS is something that we're working on. So instead of sending you to a form, we can we can do a two way SMS. But our core fundamental product is going to voice an answer, transfer a call, or send an SMS with a link or information. And we feel like that is gonna give you the initial call deflection that you need and give you a base to build on for the future. So I hope that answers that question, Lisa. Thank you, Kurt. And thanks team for the for the feedback. Where does this information come from? Where what is the source for the chatbot? Going back to that collection period, you we could do it three ways. You can interview all the call takers and ask them what the calls are about. Wonderful people. This is not the methodology I would recommend, because it's not that driven. It's human based. Number two, you could take, x amount of months worth of calls and you could run it through AI tools and say, hey, somewhere around these calls, tell me what the topic is. Number three, which is my recommendation is let the bot run for a month and just collect that. Although you've reached any city non emergency number, how can I help you? The person says what they say, the bot sends them back to a human like you would do today. But what it's done there is it's collected the intent of the call. So now a month later, you can go back and you could say, I know exactly what people are calling about. I know the exact topics. I know the way they're asking the question. That's very, very important. Do they use the word crack? Do they use the word accident? Do they use the word crash? Right? That all matters. Let's get real data, and then we build out the datasets with the answers based upon actual information that we've collected over a three to four week period. That's my recommendation. Kurt, Kelly? Yeah. I think Can I follow-up with one more question to John? So does that mean that it's continuously learning? Yeah. Oh, absolutely. Yeah. And you teach it. Right? So it's learning and say say if it misses, you know, you wanna know if people say animal control or animal services. Well, once you see that people say both, you just train the bot that animal control and animal services is the same, so it should have the same answer. But you're never gonna know that until you collect the data. So the more you collect, the more it learns, the the smarter it gets, but you, again, are supervising it. You're saying this equals that, and it's not just magically doing it. Eventually, it will get to the point where it start it can start learning on its own. But that that takes a while. Right? That's a long way away. I mean, it's doable now, but it's gonna take a while in this process. At least for that to happen because you need massive amounts of information. And, Lisa, that's the point I was gonna make. I mean, to John's point, you gotta have a good starting off point. I mean, you gotta have the data you need to set this thing up for success initially. But to your point, you know, we we would provide managed services to any solution to help you keep that bot tuned and improve that bot. And I think John will agree, what we've seen over time is when you implement the bot, you know, the default for a non emergency bot is to send it back to a human. No doubt about it. If we don't know what's going on, it needs to go back to a human. I mean, we're not gonna hang up on them. You know, you know so we look at what falls out of the non emergency bot, and then we learn. That's how we learn. We made these assumptions. We gathered this data. We configured it this way. It needs to be changed, or we need to add some utterances. Or so initially, you're gonna see a lot of of a lot of changes, but it'll settle down as you begin to tune the bot to to support your constituents. That's great. Well, it looks like we're at time, and I don't see any more questions coming through. But I wanna express my appreciation to Kurt and Kelly for presenting our attendees with some insightful, impact to how AI is working within nine one one contact centers. A heartfelt thank you for our guest panelist, John Persano from AWS, for sharing your insightful ideas. And, additionally, I extend a special thank you to our guests for participating in our webinar today.