Name: Why AI chat for support fails to deliver ROI (and how to fix it)
Uploaded: 2026-01-30T20:11:05.808Z
Duration: 40 min 8 s
Description: Why AI chat for support fails to deliver ROI (and how to fix it)

Transcript for "Why AI chat for support fails to deliver ROI (and how to fix it)": Hello, everyone. Happy Friday. How is it going? We are super excited to be here today. We're gonna wait our customary minute or so to let people join. We are really excited to chat today all about AI chat agents for support and why a lot of common deployments are failing to produce a return on investment. Before we get into it in that minute, if you are here, let us know where you're coming in from. Where are you joining us from? We are an assembled we are based here in San Francisco. We feel very lucky that, we have not been dealing with the snowstorm that the rest of this country has been dealing with. Very spoiled Californians over here for sure. Awesome. Rebecca from Massachusetts. Oh, from Ukraine. Amazing. Salt Lake City, Utah. Terry, that's killer. Love to see it. Yeah. We really cannot complain as Californians. We put on one little one little sweater when it's 65 degrees and we're cold. Oh, yay. Salt Lake City. Awesome. Glad to hear it. Glad to hear that things are better in Salt Lake City. Great. Now it's been a minute. We'll just jump right into it. So, today, as I mentioned, super excited to talk about AI chat agents for support, why common deployments are failing to produce a return, and, an introduction on us here. My name is Cassandra Stumer. I work on product marketing and assembled for AI voice and AI chat products. I'm joined here by my colleague, Mongku, who is a product expert. He will jump into the product in a little bit and show you stuff live. A few housekeeping items here as always before we get started. Please definitely make use of the right hand console. If you have any messages or questions, go ahead, chat in there. We love it to be interactive. We also have a docs tab here. Check that out. We've, added some content, that we'll also discuss a little later. And this session is being recorded. If you have to jump early or anything like that, we'll send the recording after the event, so no worries there. So by way of agenda, we're gonna just first give a little quick introduction, to assembled for anyone that's unfamiliar. We'll also chat a bit about the state of AI ROI, sort of like what we're seeing, what we're hearing, why AI chat deployments at the moment aren't really making an impact on the bottom line, what we believe the solution is here at Assembled. And then Monk is just gonna jump right in the product, like I mentioned, run through a live demo of our AI chat agent and some of the features we've built specifically to close that gap. And then finally, we're gonna chat through how some real support orgs are approaching this in the wild. And we've also got a little bit of additional time for Q and A at the end here. So if you do have any questions, like I mentioned, go ahead, pop them in the Q and A there, on the right of your screen. At any point, we'll answer them at the end. And if we don't, we'll make sure to follow-up with you via email. So quick introduction here on assembled for anyone that's unfamiliar. So assembled, we're the orchestration layer that makes AI and humans work together seamlessly. So that means you can plan, staff, manage your humans and AI agents on one platform. And spoiler alert here, that's a lot of what we're talking about today. Historically, a lot of these teams, they've been separated. They're separate worlds. Right? Workforce management tools for humans and then your isolated AI vendors for automation. And assembled brings both of these together so you can see and manage your full operation in one place. We really think of these as part of the same team and not two separate teams. We're connected to your tech stack with, of course, all our integrations. We've got all your favorite stuff for ticket platforms, telephony providers, knowledge bases, productivity, QA, plus, of course, we have our open API, for really any custom connections, anything that you could possibly dream up. This all adds up to our platform. We do have workforce management, as mentioned, Copilot, AI agents for every channel. And these products, we believe, really work great together, but can also just be used individually to meet you wherever your org is at. And our products do share a data layer, that's why we think that they're best used together. And the importance of that is that your workforce management insights really make your AI agents more effective, more dynamic, and your AI agents and Copilot performance data can actually help inform future capacity planning and workforce management. We'll get into that in a second. But today, we wanna just talk about how that manifests in our AI chat agent more specifically. So a little lay of the land here, as I mentioned. We've all seen the headlines. I mean, you guys have all seen versions of these stats. Right? We're seeing it over and over that organizations are generally struggling to capture ROI from AI. We've seen these a ton, but these are pretty recent. This is this latest PWC survey that came out at 2026. So what is this? Less than we're January 30, a month old at this point. So still an issue. They cited that over half of CEOs, 56% here, are still seeing no positive impact from AI. No positive financial impact. Even worse, though, 12% are actually reporting a negative impact, so their costs actually increased. And you might say, well, hey. Like, maybe we're making this new investment in this new technology, and it's expensive, but maybe it's improving our customer satisfaction scores or our customer experience. And that's still not necessarily the case, unfortunately. I'm sure folks here might be familiar with this Forrester survey that they conduct annually now about, the global customer experience index. These are leading brands, so we know they're deploying AI. It evaluates the ease, effectiveness, the emotional components of a customer's interactions with a brand. And despite the fact that we have all been, like, on an AI craze in the last couple of months, years, In the last year, 21% of brands, and these are global brands, big brands, their rankings dropped, 21% of them, and only 7% improved on this global rankings index. The rest just stayed the same as well. So we're not seeing a financial return on our investment, and customers aren't any happier with us despite all these big promises that AI has made to us. What is happening? Right? And before we get into how we view this, we're really interested to see, like, I'd love to hear from you. We'd always love to hear from, the market, our customers, support leaders. How are you dealing with this? What results are you seeing? So, also, everyone's in a different place. You know, maybe some folks have rolled out AI for chat or for maybe a copilot. Maybe you're seeing some real containment, some resolution. Maybe you're just trying to figure out if this stuff is paying off for you. So go ahead. There's a poll in the, right hand corner here. You can see in that little module there. Go ahead and let us know. We're really curious. Where has AI made the biggest impact on your support org? We've got a couple options here. Is it, impacting containment, deflection? Are you are you containing more? Are you deflecting more? Are you improving your customer satisfaction scores? Maybe you're actually seeing some cost savings, some real ROI. Let us know. I'm really curious. I'll give the, oh, answers here. So okay. Early answers. We got quite a few votes on higher containment and deflection. Awesome. Yeah. That is also what we're seeing. We've got zero votes here for customer satisfaction. Get that. No votes on cost savings, and we've got the winner here is too early to tell. That makes a ton of sense. Too early to tell, and we have one vote for no impact, zero impact. So that sort of lines up with what we're seeing for sure and, what the market's telling us too. Very interesting. Thanks, guys, for delivering some of your, insight there. Very much lines up with what we're seeing too. A lot of folks voted here on higher containment and deflection. Awesome. I think here is our thesis that just easily rolls up to what you're all seeing here. A lot of AI chat deployments are failing to impact ROI. Why might that be? Here's our thesis. We're seeing containment. Why no money? We're all too focused, we think, on the wrong part of the equation. We know that for the last couple of years, this conversation around AI has really been dominated by this promise of cost cutting through containment. You know, you can bolt on AI, bolt on your AI chat, deflect, You'll contain your volume. Maybe you'll reduce your headcount and profit. Right? That's the story we've been told. But everyone here that knows support, they might have been told this story. The people that don't know support say, okay. AI is great. We've worked really hard to reach, let's say, like, 50% containment. How do we make money off that? That means we can have the number of human agents we have. Right? Let's just cut that in half. I think we all know it's not that simple, unfortunately. Everyone is reneging on that approach. The idea that business value is coming just from cutting costs or cutting people, it's, like, already outdated. And we hear this all the time. I mean, for one, we all see in the market, like, OpenAI even reversed its no humans in support stance. So everyone thinks support is simple and can be automated away, clearly not. Klarna, also very famously. I don't know if everyone's seen this sort of PR disaster about Clarna going all in on AI, and then they actually had to rehire folks, on the support team. And we see this all the time in in conversations with support leaders. A leader, in fact, at this large rideshare company, one you've largely, you've more than likely, used before, by the way. They literally told us, like, we spent two years building automations, and we still have the same number of agents. So, like, what the hell is going on? Makes sense why they'd be upset here. Here's what's going on from our perspective. So, we believe that measuring ROI just from a cost cutting, perspective, it's just a small part of the equation. Containment's important, but really just one piece. So if cutting costs or cutting people is not creating value, where is the value created? We believe that when you really think about the business value of AI and support, it's really more like multiplication rather than subtraction. So we see these, multiplication drivers in these three areas over time creating ROI. So first, your savings from efficiency gains. And this is what everyone's aware of. This is, where we focused all of our attention. It's the most familiar piece. This is your containment, your deflection, maybe even higher resolution rates. Great. This is where you can also contain sort of the repetitive, the low value work, the, like, password reset, your FAQs, order tracking. But this could actually represent kind of a small portion of your total ROI because efficiency alone isn't really gonna change, your cost structure if your staffing stays static. And that really brings me to my next point, which I've even highlighted here with my cute little dot, capacity monetization. This is where so much of the untapped value lives. Because at the end of the day, what AI changes fundamentally is just the fact that it creates more capacity. Right? But that capacity is not going to monetize itself. You obviously have to relocate it effectively. So, sure, of course, we need to contain more with AI, but if that's the focus and it's happening in a silo, you could totally miss what's actually driving cost downstream. And if you don't understand the new volume that's hitting your agents now that AIs contain the easy stuff, if you don't understand the volume and you don't understand the complexity of that volume, you could be spending a ton of money on your new AI vendor and just saving nothing because you haven't done anything with your human strategy because it's hard to do that. It's hard to see all those things in one place. And finally, improvements to the customer experience. So if AI chat's happening in a silo, how can you actually design better customer journeys that make your customers happier at the end of the day? I mean, what do your handoffs look like from AI to humans? And how are you ensuring also that, like, humans are doing the work that humans need to do? The stuff that's high stakes, the stuff that's more emotional. And how are you ensuring that AI is focused on, like, the containment stuff? Like, what is it that they that AI should focus on and what that human should focus on. Right? And this is our key thesis at the end of the day. Like, today's AI tools aren't really linking these three streams together. It's really hard to get a full picture of your entire operation across humans and AI and much less monetize the capacity that AI is creating. I mean, even, like, this huge rideshare company that you think would have endless resources to get, a million data analysts on this problem to understand it, to reallocate their humans. It's a really hard problem to solve because efficiency may create capacity. That capacity isn't automatically monetized, and AI in a silo, at the end of the day, we all know isn't necessarily gonna provide a better customer experience. Everyone's automating. Everyone's using AI already. We're automating more and more, but we're not actually seeing a business impact. It's kind of why we're seeing so many AI pro programs, so many AI deployments. They look really successful on paper, you can say, hey. We've achieved 70% containment or resolution, and they might still be disappointing in your budget at the end of the day. So this is where we believe and fundamentally what we've built, assembled around, support orchestration that brings together humans and AI because we believe this is what captures ROI. Support orchestration comes in, as I mentioned before, this is our platform. It's really the layer that connects these three streams across humans and AI so that value actually can show up. It's an operational layer that really unites AI and humans into one dynamic system that you can view and change. And this is what where we believe truly that the the actual massive opportunity lies. We think that early moving organizations that can take this orchestration mindset with humans and AI together, tying together your human staffing and AI capacity, I mean, we think these orgs will really have a much bigger five to ten year head start over companies that aren't taking this type of action right away. So the sooner that you can put this into your workforce planning, the better. And how does that manifest? So the efficiency we talked about. Obviously, that comes from meeting the repetitive work, but also when you can orchestrate humans and AI together, it also comes from just removing the operational friction between humans and AI. I mean, you are, connecting systems, unifying queues, routing. You can identify where knowledge gaps are and, where they're perhaps preventing further automation opportunities. And now capacity monetization, the really important one here. This comes from knowing where AI is creating capacity and just having the ability to actively reallocate it. So, tie your AI metrics to your staffing and planning where you can see literally where hours are being freed up and reallocate that in the moment and for the long term. So we mean, like, distribute agents across skills, across queues and channels based on your actual AI coverage. You can even hand off based on real time agent capacity and queue demand. So if your team is super swamped, for example, you can say, hey. Let's make sure AI focuses more on containment. But if you have a bunch of agents available just sitting around, hey, why not, like, send a customer their way? Make sure you have a great customer experience because they can talk to a human faster. And then finally, CX improvements. If you're orchestrating humans and AI together, you can really ensure your handoffs are seamless. Humans and AI, again, are effectively allocated to the places that they need to be, pinpoint which issues exactly AI can handle and which still need the human touch, because we don't wanna get rid of humans exactly. A 100% automation or AI should not be the goal, and we all know that. There's always going to be, support tickets that will need a human touch and should have humans involved. And, again, that's just like our our guiding thesis here at Assembled. It's really the core of why we're building what we're building. We believe that orchestrating humans and AI together is key, especially, like, support's evolving really fast. We weren't having this conversation, like, a year ago, two years ago, six months ago. It changes. New models are out every day. But, you know, when you can master really, like, a blended what we call blended workforce management, so it's humans and AI together, you can deliver just better experiences, lower cost, but you can also just move from reacting reacting to these issues to really preventing them in the future and making sure you're prepared for what's for what's next. With that, I know I'll get off my my pistol here, and I'll just let Manku just get right into the product, and he's gonna take us through a demo of our AI chat agent and some of the features that we've built, the orchestration features to ensure our customers see real ROI and not just containment or deflection. Perfect. Thank you, Cassandra. Let me share my screen if you wanna stop sharing yours. Awesome. Thank you. And then pull up the right window. Cool. I'm gonna make this full screen just to make it a little bit easier for people to see. Can we see that? Looks like we can. Alright. Yeah. Appreciate appreciate, walking us through all of that, Accenture. But I am really excited to kind of walk everyone through today on a high level demo of assembled chat agent. Where I'm gonna start off is walking us through a scenario where a customer might want to do something like, hey. They wanna upgrade their membership. Maybe they have some questions around, like, their their different the different tiers that they can upgrade to. This is an example that I think everyone can kinda relate to and understand, but understand that the core fundamentals, I'll kinda show you how everything is kinda set up on the back end. But this example is a combination of a, you know, somebody that can ask, like, self serviceable information as well as a situation where typically a human agent might need to go into a back end system here or there, and, you know, actually make some changes. So what I'm gonna do is I'm gonna ask a question. I'm gonna say like, hey. Like, I'd like to upgrade my membership. If you'll notice on the right hand side, one, this is our chat testing area. This is kind of where you can go in and see like, hey, like, are the workflows and things that I'm building out, do they make sense? Are they working? This is your area to be able to QA easily, and you can kind of see the thought process on the on the right hand side of what the AI is doing. The first thing I wanna call out here is, one, we're doing adversarial detection all the time. Somebody comes in, they wanna ask you about doing, you know, like, how do I rob a bank or something. Right? Like, obviously, stuff like that. Not relevant to to your guys' use case, and so we can identify those types of things and shut that down. The second piece here that I wanna call out is we didn't detect any escalation in this message. Every single message that a human has sends and and every single, basically, part of the conversation, we are always checking, is escalation required in this moment? And the reason that that's important is because we don't wanna have your your any situations where your customers are getting stuck in AI doom loops. Right? We have the capacity, and I'll show you this later as well, to kind of say, hey. I want a maximum number of attempts before I I just automatically have the AI handoff. But this is also a second lever you can pull where it's, like, more dynamic. You can define, hey. When do I wanna escalate? Hey. Anything related to lawsuits or threats of lawsuits or anything like that, trust and safety, I wanna be able to escalate those immediately. Right? Somebody shows high levels of frustration. I wanna handle I I wanna be able to escalate that immediately. You can customize that so the AI is intelligent enough to be able to say, hey. It doesn't matter what I'm doing right now. A human being can change their mind. Right? And if they if they change their mind, I'm gonna hand them off to a human being. But going back to this, what we've identified here is we've identified that, hey. Alright. Cool. We're gonna, activate this workflow. Basically, hey. I wanna be able to check what is your full name, what is your member ID. Let me look you up in my system. But I'm gonna interrupt you right now and say, like, can you tell me more about the different tiers that I can upgrade to? I'm on the basic plan right now. So even though we're kind of interrupting the flow of the of the conversation, hey. Here is a, it's basically giving me the basic breakdown of, hey. Here's the basic tier. Here's the premium tier. Here's the ultimate tier. Here's how much it costs. Here are the basic things that I need to make sure that, you know, that that are the critical differentiators between each of these different tiers. So I can say, great. The ultimate tier sounds like what I need. Can I upgrade to that? Cool. And so we're going right back to where we were before. Right? Picking up right where we left off. It's asking for my full name, my member ID. Member ID is 421. My name is Mongu. And so what it's doing right now is you can see the the different events and things that are happening right here, but this is a retail database that we set up. Basically, here's my information right here, my name, my ID, the account type that I that I have right now. And what the AI is doing is through just an API call, it's just looking up all this information, and then it's saying, hey. I I can validate that you are who you say you are. Right? And then I was asking for confirmation. Can I upgrade you from basic to ultimate? I'll say yes. Confirmed. And now it's gonna go back in and go into the retail database and update my plan from basic to ultimate. And so this is a really, really basic example. You know, obviously, just updating a field in a in a retooled database is relatively straightforward. But you can imagine this is like, hey. Look. I wanna go into my billing system, and I wanna update credit card information. I want to go into, a Snowflake warehouse or, an ERP system and I wanna pull information or change information there. Basically, anything that your human agents are typically having to do manually, our AI agent, as long as we can access that information via API, our AI agents may have to automate all that stuff. So here we have the seamless integration between, like, hey, self serveable information. Our AI is intelligent enough. There's a lot of things you can control around, like, style and everything to make sure that the AI sounds exactly like your like your human agents. But we can go into the workflows right now and show you, hey. What does this look like, on the actual back end? So this is the specific workflow that we were working around. And the thing that I wanna call out here is that assembled is built by customer experience experts for customer experiences support teams. Right? So you don't need to have, you know, making adjustments, QA ing, troubleshooting issues, or making tweaks to the AI to be able to to make sure that you're, the AI is delivering the exact experience you want. It's not gonna require, like, entire teams of forward deployed engineers and a bunch of engineering work, and it's not gonna take you weeks to be able to, make these types of changes. All of this stuff, it's all very, very easy to set up. This is a simple trigger step. All of our, all of our rules and everything here, all of the guides, what we call them, are all written in natural language. You can work with your engineering team to set up like some of these reusable APIs, and now you can just reference that API call right here. Makes it really, really easy for people that typically your support teams don't have a lot of engineering resources. It makes it really, really easy for teams to be able to deploy things super, super quickly. The other thing that I wanna call out here is I know we're focused on chat right now, but you'll notice that when I build out this workflow, I have this thing here called manage activation. When you build out a workflow, you build this out for all of your different channels. And with the simple click of a button, you can launch this on your email channel. You can launch this on your voice channel. There's no having to do a ton of work to be able to go in and say, like, hey. Like, alright. I need to build out a workflow for voice. I need to build out a workflow for chat. I need to build out a workflow for this. The the the important thing here is that all of this single source of truth, your answers across all of your different channels are gonna be the same. But now that we know that the AI is going to perform and that we're gonna be able to achieve some of those high resolution rates and deflection rates that your teams are seeing today, the next and probably the most important thing is, what are we doing with these results? And that's where I wanna take you over to the dashboard area right here. And this is an area where it's blending workforce management and AI data. And so we're giving you a visibility into, hey. What does our is what what do our net staffing requirements look like, across all of our different channels here? Hey. You can throw in additional forecasts and things here as well. But you can also track, like, critical metrics like, hey. What's my average channel time? What's my CSAT score looking like right now? But you can also see, like, hey. How many cases are we solving and how many of those cases are being solved by AI? And so now you can see a breakdown of, hey. How many of our tickets are being solved with Copilot, you know, by an AI, completely or by a human using a Copilot? How many are being resolved without AI at all. And the important thing here is this gives you that kind of, like, high level visibility into, like, alright. Where are my where where do I need to be focusing my time, and my efforts to make sure that I'm delivering the right amount of, the right basically, where do I need to be looking in my organization? Sorry. I'm seeing in the chat right now. If you need me to zoom in a little bit more, I am happy to do that. I guess that doesn't really change too much of the the dashboards. But Well, cool. good. I think it looks great, Manko. I'm just thinking some people might wanna zoom a little here and there if they wanna see something in particular, but looks great. Cool. Alright. But this this is, one of the key areas where from an overall metric standpoint, this is how you can take what is your AI doing. Right? Obviously, we're gonna be able to deliver, like, high resolution rates. We're gonna be able to deliver the exact experience that you want, but we're gonna be able to also do take the next step, which is turning that AI impact into actual decision making data for your business that's gonna have real impact. And we're able to support all of that with data, and all of that's gonna be consolidated in one place as well. AI and workforce management really shouldn't live in separate silos, but they do in in today's environment. Right? Assembled is one of the only vendors out there right now that's actually consolidating and bringing all of those those two different things into the same platform. They need to be integrated together so that you can do really cool things like this. So I can go over to the chat agent. I'm gonna go into one of these profiles. I'm gonna go over to the handoff area. Tons of other things that I can get into here. For the sake of time, I'm not gonna get into it, but the handoff area is where you can do one of the really cool things of saying, hey. I want dynamic handoff rules. Before I get into that, we have different basic out of the box sensitivity things here of, like, hey. You know, when we're handing off, whether it's because somebody is demanding an escalation or whether it's because we're, you know, not we just need to make sure that this gets handed off to a human being maybe to take the next action. Right? We have different sensitivity criteria here. So standard is basically saying, hey. We'll we'll try to resolve. Right? If somebody says, hey. I wanna talk to a human. Maybe we'll push back a little bit on that. Cautious is saying, hey. You know, hey. We don't wanna, deliver a negative experience here at all. The moment somebody asks for a human agent, let's just hand them off. The containment focus is saying, hey, we're gonna push back a couple times. Right? We're gonna try and make sure we'll explain that, hey, the AI is gonna be able to help you with x, y, and z. I can help you with this. Let me try. But the cool thing here is I can add these dynamic handoff rules where I can go in and say, hey. If my average wait time goes above ten minutes, then I want my handoff sensitivity to be containment focused because I wanna make sure, right, that I am not flooding my my human agents. I can see that I'm, like, understaffed right now or that I'm just getting slammed with a bunch of phone calls or a bunch of chats right now. And so I wanna try and alleviate some of that pressure from from my team. And so there's there's really, really cool things that you can do here. You know, obviously, there's a ton of different factors here as well. Hey. How many available agents do I have? What does my occupancy look like? What does my real time SLA look like? You can you intelligently make decisions like this because the data, the AI data, and the workforce management data are in the same place talking to each other, and you can have one influence the other. Now I know I've gone over a lot. There are some other things here that I can get into, like the QA review area, some of the there's tons of other areas. I can probably spend another two hours talking about this, but my time here is unfortunately limited. So feel free. Reach out to me directly. My email is monkhoo@assembleofhq.com. We'll drop it into the chat or or somewhere. If you want a personalized demo, I'm happy to set that up. I know we have a little bit of time for q and a later, so I'll be on to that. But, I'll I'll hand this back off to Cassandra for now to kinda get into the real world results and things that we've been able to actually drive some of our customers. Awesome. Great. Yeah. I love seeing our product live, and we've gotten feedback before. It's funny. Like, we always at assembled, and I'll get back into slides here, but we always show, real live product. We we don't like to be in slide world too long. So we've gotten great feedback on that. I hope that's something that resonates with y'all and that we should continue to do. But, again, it's helpful. If it's personalized, let us know if we can, you know, dive deeper into your specific use case. Like Monkhu said, just wanna dive a little deeper here into what we have seen and what our customers have seen with, actually work actually using Orchestration Live. So, it's a unique approach, but I think people are resonating with it. A lot of our customers that use both workforce management and our AI agents have seen really massive gains here. So, Canva, we just came out with this case study that we're super excited about. And PS, check that docs tab right where where the polls are. All of these customers here I'm mentioning do have a customer story case study you can actually look at on our website. I've linked them all in the docs tab. But Canva, we're super psyched about. They use assembled reverse management and AI Copilot, and they're automating almost, like, 29,000 really complex cases every month. I mean, they have refunds, a lot of print quality issues. I didn't realize so many folks are printing stuff with Canva, order tracking, but what's really important to them was this, human in the loop, like, safeguards that they created. And it was really great to work with them to see that, like, we could scale their automation while ensuring their quality. And that also happened because we're, their workforce management platform, and we're saving them a ton of time, with automated scheduling, forecasting. So Canva is a big company, and they're global. They come several different BPOs. They have their in house team. They're all over the world. And we're saving them a massive amount of time on their scheduling and forecasting, of course. Scheduling all those folks across, like, different skill sets and time zones was a huge issue for them. They're saving a ton of time with all this automated scheduling forecasting and also real time management of all their their thousand plus agents. FlexCar is also a really interesting one. So FlexCar is a a month to month Carly service. They're, like, this great new startup. I'm not sure if anyone used them. I think they're great. They're only on the East Coast. I wish they were over here. But they use dynamic handoff to keep their twenty four seven coverage intact. I mean, they were spending a ton hiring more folks for overnight. They're trying to get twenty four seven coverage. You can imagine people are calling them all the time when, like, their car is, somewhere on the side of the road for some reason. Also, they have issues with their lease. They wanna get a new car. They're calling constantly. So they needed twenty four seven coverage, but they also just made sure, like, we have to have our AI agents results in the low complexity stuff when people are just messaging or calling in. We use, or FlexCar uses us for phone, chat, workforce management. So we're too big 85% containment just on the easy stuff. So already their agents don't have to deal with the complexity of the calls they might get or the chats that we're going to their, their life support staff, so that's great. We're only escalating those really urgent ones now to humans. Thrasio, also a very interesting company. So they're an Amazon brand aggregator. They have a ton of brands on Amazon you wouldn't even think of. They've been using us to automate a bunch of their decision making, especially around seasonal peaks. So you can imagine they have a ton of brands on Amazon. Like, Amazon Prime Day is crazy for them. So they've always had issues with sort of, like, seasonal peaks, and they've needed to really, like, weigh the trade offs of how they can deploy AI or how they can hire temporarily to accommodate that sort of, like, peak with Prime Day. With this sort of long term planning, they've saved a ton, 1,800,000 so far, and, you know, we need to update this case study because I think this is from a while ago. I'm sure it's more at this point. So if you'd like to learn more again about any of these customers and their use cases, how they did this, I've linked this, in the docs tab. Check them out. Again, you can always message us. And with that, I'll just say sort of leads us to the end of our content. Thank you so much for joining us online today. Monkou and I will stick around here, answer any questions you might have. But before that, just wanna say final reminders. Again, you can check your inbox for a recording of this webinar. You'll get it, automatically. You can also always visit our website. I've got those links for those specific case studies, but check it out. Lots of information there. You'll also be able to stay in the loop. We, run a webinar every month. So check out some of our past webinars, sign up for an upcoming one. You can see those all on our webinars page on our website. So that will leave some time for q and a. And if you don't have any, we will also let you all go. Let's see. None so far. Great. No news is good news. Great. Thank you. Thanks. Thanks so much, Elazaida. I hope I'm pronouncing your name correct. Glad you enjoyed. And, also, again, we'll plug that, like, if you want to learn more about how this would manifest with your use case, feel free to just message us. Reply back to that email. Message Mongku. Sent him an email. We'll go through that specifically. Let's wait for, like, two more minutes if anyone has q and a, and then we'll hop off and let you go. Thanks, Jim. And, also, happy Friday to everyone. Like, one happy Friday. Where did this week go? Where did this month go? This was, like, a very quick January. Right? Just us. That flew by. So thanks, Roman. Glad you enjoyed it. Happy Friday. I hope everyone has a warmer warmer February and a great weekend. Great. Let's give it one more minute, and then I think we'll just hop off. And if you have any questions, again, reply to that email we send you or message Manku either way. Thanks, Alex. Great. So glad y'all appreciated it. Cool. Amazing. I think we'll let you go, and fantastic. Thank you all for joining us today. Keep in touch. We really appreciate everyone who takes time out of their day to come and listen to us here at assembled chat about support ops. So great. Have a great one, everyone, and we'll sign off. Thank you.