Name: Unlocking your support team’s potential in the age of AI agents
Uploaded: 2025-07-25T19:06:09.643Z
Duration: 49 min 40 s
Description: Unlocking your support team’s potential in the age of AI agents

Transcript for "Unlocking your support team’s potential in the age of AI agents": Hello. Hello, everyone. Welcome. Welcome. Gonna do a quick mic check here. Can folks hear me? Great. Fantastic. Hello from foggy San Francisco. Joining you here from what they're saying is the coldest winter on record or coldest summer on record since 1982, I believe. So got our sweaters on here in San Francisco. Let me know in the chat where you guys joining in from. Welcome. Welcome. Fantastic. Utah. Love it. Great. Minneapolis. New York. Great. So it's all sorts of or afternoon for some folks here today. 11AM for us, here in Pacific time. Very nice. Great. We've got everyone joining LA. We've got all coasts represented here. Fantastic. Great. Well, I'm just gonna jump right into it today. The session is recorded for anyone who may join late, and just say right now, welcome to today's webinar from Assembled, unlocking your support team's potential in the age of AI agents. Thank you all so much for being here. My name is Cassandra Stumer. I work on product marketing here at Assembled. I'm also joined today by Andrew Steinberg, resident Assembled expert, who will take you through a live demo of Assembled's AI agent Copilot. We'll also be joined by Michael Lim from our AI team. He's gonna join us to pop in for the, to support for the q and a portion of this session at the end. By way of agenda today, we're gonna provide just a brief intro to assemble as a refresher, for anyone unfamiliar as well. We'll also chat about, what we're seeing in the market, what we're hearing from customers, and how they're approaching the changing role of live agents in the age of AI. Then Andrew is just gonna jump right into the product. He's gonna run through how Assembled AI Copilot guides agents through tough cases. And after the demo, we'll talk a bit about what we're hearing, about the impact and the results Assembled customers are seeing after implementing Copilot. And finally, as I mentioned, we are excited to close out the session with a bit of live q and a. So if you do have any questions at any point throughout the session, feel free to go ahead, pop them in the chat or the q and a on the right side of your screen at any time as they come to you, and we'll save them to the end and make sure we get to them. So a bit of background on us. Assembled has been in the support game since 2018. Seems so long ago now, pre COVID. We have about, a 120 employees and growing, with offices in San Francisco, where I'm joining you from, in New York, and we recently just also opened an office in London. We partner with hundreds of the world's most most interesting and innovative companies, to help them deliver world class customer support. And how do we do that? A bit of info on our AI support platform. With our platform, we're already connected to the contact platforms and the tools that drive your support team. And then with our three pronged approach, we can provide a suite of workforce management tools, that will properly account for your team's capacity. Customizable AI agents, of course, across phone, chat, email to automate away, close tickets, and bill pilot solutions to support your live agents, which is, of course, what we're chatting about today. Underlying all of this is our data and insights layer. It's like a command center. It gives you complete visibility into your entire support operation across humans and AI all in one location. These three products work really great together, but they could also be pulled out individually to suit where your org and your needs are at today. We see, a consistent outcomes from our customers like increased customer satisfaction, boost in agent productivity, which we'll talk about today, and, overall, like, the ability to just automate more end to end. Well, that's to say, really, TLDR is assembled. We really, really believe that great customer support requires AI and humans in perfect balance. That really brings us into today's topic. So how can we better support live agents now that AI has and will continue to reshape their role? I'm sure you're all seeing stats like this everywhere. Right? I mean, across several different surveys, we are hearing that agents are reporting how their role has actually become more challenging since the AI boom. Agents are saying that their workloads have grown just in this last year. They're using more tools, more and more tools just to find answers. And on top of all of that, as always, customer expectations are rising. They're never falling, are they? AI was supposed to make everything easier for everyone. Right? So what is happening here? Well, Ben Levick from Ramp actually, I think, just said it best. It's pretty obvious. I mean, now that all the easy tickets are automated, your support team is left with all the complex cases. And this is a challenge we're really hearing from across our customer base. Ben, shared this recently at a panel we hosted with Etsy in New York on AI support secrets, and this was a big topic of discussion there where where Ben had shared this insight. Another customer here, FlexCar, a really awesome, startup that does a flexible subscription car leasing. They're seeing the same thing. We have an upcoming case study in the works. Stay tuned with, Leslie and the team from FlexCar, and they really mentioned, how their approach to to hiring really into evaluating their live agents overall has changed completely after implementing AI agents. They were spending just a ton of time on training before just because their nature the nature of their evolving product, it's a start up. Things are moving fast. And making sure agents knew each feature or policy, it's like it was really just about training them on the right response to customers, making sure they get it right, but not on how to deliver those answers over the phone particularly, which is a big channel for them, or even over chat, which became, you know, all the more important once those really easy to answer questions are automated. And, of course, we all know these aren't just the harder or more complex cases. As everyone here is really well aware, any case that gets escalated, that's often also the higher value case. Right? Those are your VIPs, fraud, billing, complaints. So getting those fewer but more challenging cases right is all the more crucial. It's really why we believe today that one of the fundamental unlocks and the maybe one of the easier wins for maximizing productivity for your support team, for their potential as our title would describe, title of this session, as more and more is automated in this new paradigm, right, one of the easiest ways to maximize your support team's potential is an agent copilot. We're seeing it's a really, really essential part of the AI journey. In fact, Ben's advice at that same session, at that same event in New York was to deploy AI agents first, AI chat agents specifically, and it's a great forcing function to get your knowledge base in order and then deploy your AI Copilot immediately after to solve for this. Make sure that your agents aren't left holding the bag with the complex cases once you start automating more and more. And, obviously, it's a fundamental unlock, but how? Obviously, we cut into the repetitive work, spending a ton of time searching for knowledge and drafting replies. Right? Leslie from Flex Car mentioned that this was a huge issue. Sometimes their agents are, you know, searching around all their tools for knowledge and then drafting a reply that turns out to not even be correct. Also, we know it gets new on on new hires onboarded really fast. Right? They already have access immediately to your best knowledge. It can help maintain accuracy, consistency across all your brands, make sure your responses are all on brand, all in your policy. And, also, one of my favorite ones that I don't think would be possible without AI, obviously, is, turning all your agents into polyglots with instant translations. I don't think, any amount of training would make everyone know 50 plus languages immediately. Right? So that's a pretty cool one. But getting a little ahead of ourselves here, these are, of course, all of the features we're gonna dive into in our demo of our Copilot today. So without further ado here, I'm just gonna hand it off to Andrew who's gonna share more about REI Copilot. Awesome. Thank you, Cassandra, and hello, everyone. I'm so excited to be here with you all to give you a live walk through of Assembled Copilot tool. Without further ado, let me go ahead and share my screen here, and we can get right into it. K. Alright. So I will be showcasing assembled Copilot functionality through a few different ticket examples. But just to ground ourselves a little bit. What we're looking at here is a live Zendesk instance. This is where CoPilot will live directly embedded in the same workspace where your agents are already doing their work and answering their tickets. So really designed to meet folks exactly where they are without the need to contact switch and swivel chair between different tools. For the purposes of this demo, we'll be using Zendesk because that's what we use internally at Assembl. That said, Copilot is really designed to work across a wide range of platforms. We are going to kick things off here with a ticket from Mindy. In this instance, as you'll see, Mindy is writing into Assembled as a new team member. She's looking to get up to speed. She's asking two different questions here. One about our Slack integration and another about our AI product. So this is clearly a bit more of an involved ticket, you know, a multipart question. There's a bit of nuance. So I think a really great example of a ticket that most likely would have not been automated away and a good candidate to show how agents can really lean on Copilot to move faster and respond with confidence for some more of those complex questions. Now if I'm an agent picking up this ticket, I think my first instinct would be to make sure I'm hitting my first, response targets. That usually starts with a quick initial reply, something to let Mindy know that we're looking into her question. And rather than manually typing that out here into the text box, with Copilot, I can actually just go ahead and hit this initial greeting button and drop a quick ready made reply right into the editor to let Mindy know that that we're on the case. And, you know, I think shaving ten to fifteen seconds off of a ticket here might not seem like too much initially, but can definitely tell you across hundreds or thousands of tickets, the effect is real. Especially when it comes to improving folks' first response times and reducing handle time at, at scale. Not to mention, I personally shadow dozens of agents and and can tell you firsthand how painful it is to watch folks manually typing this stuff in, every every single ticket. Gonna go ahead and clear this conversation. The language detection and summary functionality, we'll we'll get into on another ticket. What I wanna highlight next, I think, is one of our most used and impactful features. It's called our our draft reply. And this is essentially where our Copilot will read the incoming ticket and attempt to answer the customer's question, generating a suggested response by pulling in from all of your available knowledge sources. So your SOPs, your internal docs, your help center articles, really whatever you've got. And the response we'll generate is not just tailored to the actual content of the ticket, but also to your brand voice and tone, which I'll get into more deeply in in just a second here. So I'll go ahead and hit draft reply. We'll let Cal, our AI agent, happily jump around while we're generating the answer. Once the draft is generated, the agent will have a few different options here. So they can, of course, send this as is. They can move it directly into the editor if they wanna make any minor tweaks, or they can also use what's called our rephrase, reply option here. What I've seen from a lot of our customers let me actually just zoom in here a little bit so folks can see. What I've seen from a lot of customers, especially those that maybe some more like outsource teams, is that after an agent confirms the content of the answer is correct, they actually spend a pretty surprising amount of time just, like, rewording and massaging things. Whether that's, you know, adding a little bit of empathy, tightening up some phrasing. Maybe in this specific example, you know, the AI failed to mention that Mindy is welcome is is new to the team here at Semple. So maybe I might wanna rephrase this reply and say something like, you know, make more concise, nothing more than seven sentences, and don't forget to welcome Mindy to the team. Something like that. I can now very easily just rephrase that reply with the additional context, that I added that I added back in here. And we can then go ahead and drop that back into the editor and send this back to Mindy. You'll notice here as well we have some native QA that's built in so that your managers can begin to audit and report on what's actually helpful and what's not helpful for your agents at at large. Now I think as I'm showing you some of the draft reply functionality, I think the real catch here with with draft reply and Copilot use more generally is that none of this actually matters if your agents don't trust the responses that are generated. And when we first rolled out Copilot to a bunch of our customers, one of the top questions we kept hearing was from agents was, you know, like, where did this answer come from? Because so many other AI tools on the market are are sort of just this black box. Right? So what we'll also do whenever we draft your reply, is we'll ensure that we're, one, servicing the rationale so that the agent can actually understand how the AI is thinking through the response. We'll also always run a sentiment and confidence analysis on the ticket as well. And then lastly, and I I think most importantly, we'll always ensure that we're citing our knowledge sources, both internal and external, right under the draft that the that the agent can, can see and reference if they need it. And I think, generally speaking, like, this sort of, you know, ultra transparency is is really what's gonna drive adoption with your agents because at the end of the day, the only way you'll unlock those, you know, incredible efficiency gains that, you know, we've been boasting is by getting agents to actually use the tool more. So some of these quality of life things like, you know, rephrasing responses or like this transparency we're giving through rationale and citing of knowledge is all going to help build trust with your managers and team leads and and ultimately your agents, of course. And I think that trust is really what will what will drive adoption for for teams. I'm gonna go ahead and clear this. Lastly here, on the topic of knowledge, we also have this nifty search knowledge bar, where agents will be able to actually interact and and ask questions and get answers from your knowledge base with very, very, very easily. So, you know, whether it's a policy or a, I don't know, product detail and an edge case, you can almost think of this as like a custom chat GPT search that's specifically trained on your company knowledge. So I think where this is is super helpful for folks, for for teams who might have, like, newer or, you know, nesting agents that might not have that same, like, foundational knowledge that more seasoned agents have. So, you know, if we're gonna use this case again as an example, maybe I'm a newer agent here at Assembled. And before responding back to Mindy, I might just be curious. Like, what the heck does our Slack integration even do? Right? So maybe I might wanna respond or say something like, you know, what even is Assembled Slack integration? What value does it bring to customers, and what do I need to know about it as a support agent. I can go ahead and, you know, run a quick search here so that now next time a ticket like this comes around, I'll get some really, really good knowledge around what our stock integration is, what value it brings to our customers, and what I might might need to know as as a support agent, next time around. Cool. Let me go ahead and clear this. The last piece of functionality that I wanted to show off here is is what's called draft with note. You have a few different options here. So firstly, for teams who are heavy macro shops or, you know, maybe you have seasoned agents who are fairly habituated to using macros, we really wanna meet you where you are and and how you work. So we also do have the option here to draft a reply back to the customer using an existing macro as a template, and then maybe add in a couple bullet points to massage things a bit. So, you know, maybe we had a Slack integration macro. I don't know if I have a a great example here, but maybe we'll just use the Copilot macro and say, you know, ensure this is concise or something like that. Right? The other option, that we have here as well, which is actually really, really popular for teams, is drafting a note through, like, a few different, you know, bullet points here. So we can actually come in here and put in some free form, notes to kind of give the draft reply a framework to generate the response. So maybe I actually know the company that Mindy is reaching out from, and and I know that they actually have their Slack integration already enabled. So maybe I'll I'll draft a response with a few bullet points here and say something like, you know, answer Mindy's question, ask her to make sure she checks whether the integration is already enabled. I think it is. And then lastly, I see she mentioned our AI tool. You know, mentioned that Assembled as an amazing Copilot tool. I can now draft this note back to Mindy using those different bullet points, as as a guide or a framework for the AI to to draft the response back. And then, of course, can move it directly into the editor to send off. Awesome. Gonna go ahead and clear this. Switching gears for a quick second here. I imagine that, you know, a question some of you might have let me just zoom in here tiny, tiny bit. I imagine a question some of you might have had when I was going through the the draft reply was, how do we ensure that the responses are actually consistent with our brand tone and style? And I think being experts in support, we really understand better than anyone that sometimes there is an equal importance of not just getting the content of the answer correct, but also ensuring that the answer is communicated in a way that matches how we like to speak and show up to our customers. So what we're looking at right now is called it's called our style guide, and this is where we give you real unprecedented control over how the AI is generating responses back to your customers. So for example, right, like, how brief versus detailed we want our responses to be? What is our ratio of, you know, casual to professional? What should we maintain there? Are, you know, are we a silly support org that likes to use a lot of emojis and be humorous? Or are we a bit more serious and and buttoned up and and professional? So for a few, you know, funny anecdotes here, I I, you know, I work with one customer who is obsessed with ensuring every time they respond back to their customer, it has to end in a purple emoji. Right? But on the other hand, I have another customer that sells medical devices, obviously, a bit more serious of of a topic. And I think we know we all know how, like, off putting it would be if you get an enthusiastic response saying, you know, like, yay, your syringes are being delivered with a a fire emoji. Right? So all that's to say, just really, really important to ensure that not just the answer is correct, but that it's communicated in the right style and tone. Something else we hear a lot from folks, they oftentimes will will come to us with, like, a list of preferred or voided phrases. So, you know, we might never wanna say we apologize for any inconvenience or we, you know, regret to inform you. And the settled will give you those those guardrails and control there as well. And then lastly, we, can also set up custom style guides with different settings for chat and email. So maybe our chat replies, of course, need to be a bit more short and to the point, where emails can be a bit more, you know, detailed and and long form. And this, obviously, will ensure that agents are writing in the right tone for the right channel without having to come in and, like, you know, manually adjust every time if it's a chat or an email ticket. And what I actually recommend to to my customers here, I'm sure there's plenty of folks on this call who have those, you know, you know, superstar agents that you wish you could just clone their responses from. We can actually kind of do that here. Right? So what I always encourage folks to do is to actually take a few different examples of their agents. You see here. And then we can use that as a guide and a framework for the AI to reference anytime they draft a response back to their customers. Just do a quick, refresh here so I can show you what I'm talking about. Here we go. Cool. On another admin related note, the other question that folks might be asking us is how and where can Assembled actually ingest our knowledge? And Assembled integrates with just about every tool and knowledge base out there. So you'll see some examples on your screen right now, whether that's Notion or Help Juice or Confluence or maybe everything's just in a Google Drive. We'll ingest all of your knowledge through our integrations, and all of that knowledge is stored right here in our in our, tab it in what's called our our knowledge management section, where we can see all the different knowledge articles that we're pulling in and denote whether it's an internal or external knowledge source. And to take it a step further, you can even set up what we call a knowledge mapping rules. And these essentially let you tag articles so that Copilot is only pulling from the relevant content based on the nature of your question. So for example, if you are a, I don't know, a, like, marketplace with both buyers and sellers reaching out. Right? Perhaps you might want Copilot to pull from different articles depending on who's contacting us. Or, of course, you might wanna ensure that internal docs will never surface in a in a in a customer reply. So having this in place really gives you the right guardrails so that the AI is only pulling from what's useful and that nothing should be shared if it, you know, isn't relevant to that specific contact reason. Awesome. Gonna keep things moving here a bit, and we're gonna move into our, second ticket example. So this next ticket example we're looking at, as you can see, is a fairly long back and forth. In this specific instance, we can see that, you know, the ticket here is clearly resolved. But another awesome use case for Copilot is when tickets are escalated from, you know, say a tier one to a tier two agent, or maybe there's a really long back and forth with no internal notes. And it could take a few minutes just to kind of sift through and get up to speed on on the relevant context. And this is where our summarization feature can come in handy. You'll see just by clicking that summarization button in a few seconds, I can now get up to speed on the historical context of the ticket. And I can even, in the back end, customize how to structure these interactions to be most helpful. I can also go ahead and move this into an internal note if needed. Cool. I'm gonna go ahead and delete this here for a sec. Hop back into email. On the topic of note and ticket hygiene, I also speak to a lot of teams that have, like, specific wrap up templates or standardized ways in which agents are instructed to close out their interactions. And our wrap up templates down below will help agents really quickly just summarize those resolved conversations in a in a really clear professional format. So I can come in here and and hit wrap up. These templates are also totally, customizable here on the back end as well with our wrap up templates. And you will see that when we wrapped this up, we have a quick summary here. We have some next steps that the agent needs to take, and then we also have some, attachments in case those were were responding to the interactions. Helpful to ensure consistency with resolution summaries. We have follow through our next steps. And and I also hear from a leadership standpoint, it's helpful for those feedback loops to see, like, how issues are being handled. Cool. Now the last ticket example I want to show here is to talk a little bit about our translation features, where we can actually translate the content of a of a ticket. What I found personally to be really interesting about translations is that a lot of our customers, before realizing we could do this through Copilot, would actually handle this type of work in entirely separate tools, like, just to literally do translations, which, of course, from our mind, should ideally be handled in the same Copilot. Now we're able to unlock all these other efficiencies. So you'll see I just opened up a new ticket where the language detected is in Turkish. I don't personally speak Turkish, but the good news is I don't need to if I was an agent using Assembled. I can come in here and just hit that translation button. Assembled will translate the subject and content of the ticket. But what I can also do as well is hit draft reply, which I can then translate the content of that drafted response back to the customers all in Turkish. So really, really cool. And another cool anecdote, you know, one of my favorite stories, like, I I had a customer who was expanding into, LatAm, and they figured they would need to hire locally or, like, open up an office just to support those Spanish speaking customers. But with Copilot's translation, they they actually didn't need to do any of that. Right? They were able to handle all the Spanish speaking conversations using Copilot without actually needing to add headcount or or set up a new team. So it really completely changed how they thought about scaling internationally and and obviously saved them, a lot of time and and money in the process. Cool. So that just about covers it from the functionality side of Copilot. What I wanna spend a few minutes on next, which is really, really important, is the actual reporting side of the tool. And the subwoof has a lot of great reporting to show how we can actually, like, track and and and monitor and measure the effectiveness that Copilot is having, both on, like, the macro level across the KPIs or support org cares about, but also on the micro level to see how it's actually boosting each individual agent's productivity. The first thing I wanna show you here is our QA review page, and this is where we actually have have an audit trail of every single interaction that our agents have with Copilot. So you'll see the the the last one here was, you know, the the the the different buttons I was just pressing in this specific demo. And I think this is really helpful for managers to understand, like, how exactly folks are interacting with Copilots. One of my favorite parts of this QA review page is I always like to see, like, are there any discrepancies amongst how the answer that Cal, who's our AI agent, is generating, and then the ultimate message that the human agent send back to the customer. And we measure that with something we call semantic similarity. It's basically just a measure of how similar the two responses are. And the reason why I think this is so cool is that it's a great opportunity for you to see what types or categories of tickets have the highest potential for automation. So in other words, if if if agents are hitting draft reply and sending a response back to the customer without changing any of the content, I think that sort of begs the question of of whether or not it actually needed to hit the agent at all. Right? So really good opportunities to identify, good candidates for automation. The other thing I'll also wanna track as well is is the actual ROI. Like, hey. Is this stuff actually working? And how are some of my NorthStar metrics that I, you know, purchased a Copilot tool for, whether that's CSAT or or average channel time, how is it getting affected by my team's Copilot usage? And in our team performance report, we can get a really awesome insight into how our metrics are performing both with and without Copilot. So as you can see here, I I can look for the last month. This is obviously a demo account, so the the data admittedly isn't isn't great. But I can see how is my CSAT with Copilot and how is my CSAT without Copilot trending? Or how is the number of cases we're solving per hour influenced by our Copilot usage? And I can also get those same comparisons at the individual agent level as well. I can also get further insight into my specific team's Copilot, usage as well. So I might wanna see, like, what are the most frequently used tools on Copilot that my agents are using? So if I notice there's, like, one superstar agent whose AHT is plummeting and their cases are solving power is skyrocketing, I might wanna double click and, like, see what how they're using Copilot and if there's any correlation to how they're using Copilot to those individual statistics. Lastly here, I can also get higher level trends on the number of agents who are using Copilot on specific percentages of tickets. So, you know, in this example, like, how many of my agents are using Copilot on more than 75% of cases? Or what team leads and managers are actually driving the highest adoption among their agents if we have some sort of, you know, usage mandate that we're enacting on our team. And then what's great is I can start to, again, cross reference that usage to any correlation it might have on something like my average handle time. So for agents using Copilot on more than 75% of cases, how does that group's average handle time compare to those who use it on, you know, say, who have, like, lower or moderate usage, for example? So as you can see, we have a lot of, very comprehensive reporting that will really allow you to, like, monitor and prove the ROI around how your team is using Copilot and the direct effect that that usage has on, on your team's goals. So that is about it from my side on the live demo front. I will stop sharing, and I will pass it back to you, Cassandra. Thanks, Andrew. Love it. As always, fantastic to to actually see things live. We are, we're very bullish on not recorded demos here at Assembled. We wanna give you the give you the full experience. And then with that, you know, I love hearing just how you know, we know Copilot cuts down our petted work. It helps your agents deliver better support. All the features you've entered mentioned help with those things. But I just wanna share, some concrete examples that we're seeing, from, what what impact our customers are seeing right, from, their implementation of Copilot. So FlexCar, who I mentioned before at the top, that monthly Carly subscription, they, you know, have that constantly changing knowledge base, the fast moving start up. Their agents were spending a ton of time drafting replies. That could be incorrect. Right? With Copilot, they were actually able to cut that time hunting for knowledge, drafting replies, and they doubled their resolution rate. As we mentioned before too, they also just completely changed their strategy for hiring and for training. Now Copilot is a key piece of agent onboarding at FlexCar, and their agents can really focus on building relationships and not just responding as fast as possible. Another really great one we have here is Honeylove. Honeylove, if you're unfamiliar, is ecom company. They have bras and shapewear. So as you might imagine, they have a lot of complicated questions about sizing and returns. Shapewear turns out to be a complicated space. And, what was happening a lot at Honeylove was there were a ton of escalations happening to their team leads, to their subject matter experts. It was a huge problem. It was just increasing handle time, cost per case, agent frustration. And so with all pilot, what they were able to do was actually just decrease that escalation rate at this massive level by 20% because more regular agents were able to completely resolve those more complicated cases. We actually did, when we did their pilot, we saw that even beyond the reduction in escalations, CoPilot power users increased sales per hour by 54% compared to the average agent. And then also those teammates who weren't power users, but, just assembled Copilot users just on a handful of questions, they also saw, an increase in sales per hour by 32%. So some major improvements they're seeing from Honeylove. Thrasio, finally, is a really interesting one. Thrasio, is it a company that acquires ecommerce brands, mostly Amazon products? They acquire really fast moving ones and scale them, and they have a 190 unique brands that do this. So they launched Copilot across a 190 of those unique brands, and they were able to not only slash their response times to a minute across all of those customers, but they were also, a lot more on brand, more accurate across all of those different brands. It was actually something that increased all those factors increased their customer satisfaction rate by 10% across the board. But what we also love to see here is that, we also saw a major increase for Thrasio in agent satisfaction. Employee satisfaction rose about 8% when, agents had a Copilot to support them. So not only at the end of the day are your customers better served, but your agents do feel much better supported when they're dealing with these more complex cases. We really love to see it. We can you can also jump into a few of these case studies on our website, Flex Car, soon to come as I mentioned, but you can read more about Honeylove or Thrasio in our case study and customer section on our website. And with that, we're just gonna jump into a bit of q and a. I will hand it over to oops. Let's move this over here. Q and a. Great. I will welcome back Andrew and Michael, and we will get to some of your questions here. And feel free to pop them in the chat, but we've got a few already we'd like to address. So first and it our first question is from Ivan. Ivan, fantastic. So Ivan says data protection concerning yeah. Often a major hurdle when it comes to AI usage. Yes. Tell us more about where the data is processed and stored. I think, Michael, this would be a great question for you. Sure. Definitely. So knowledge based data is stored with, Pinecone and Algolia, and all other data is stored with our cloud providers. We use AWS here, and it's in the AWS US east one region in The US. Your data is also not used for training by the base model providers nor is it stored by them either. So you have full control over data both data storage as well as deletion. We also don't train our models on multiple users' data, so you don't have to be worried about having your data appear for other customers. Privacy and security is paramount paramount here at Assembled, and we've implemented strict controls, to keep the data separate from other customers' information. Each customer is provided with their own isolated environment, which is completely separate from other customer data. Great. And that's sort of, I think that was really answering Larry and Ivan's questions here about data security and privacy. Again, if you'd like to know more, we have more information on our website about how we handle data security privacy, but we also have all the standard certifications. So another question here from Diana. Does Copilot currently support phone based channels as well, or is it primarily focused on chat and email workflows? And, Andrew, you wanna hop into this one? Yes. I can. So, Assembled does have a voice AI product where we can, you know, build workflows that do that do support, voice or phone as well. We can also do, like, voice handoffs that will that will appear in the Copilot tool as well. The primary use case is for for chat and email there as well, but we do also support, voice for, like, more more automated ticket resolution through through workflows as well. So through automations and handoffs, I would say are the two ways that we we also support voice. Yeah. Just to expand on the answer here as well, we have several companies who are using Copilot for the phone channel, but they tend to use it more for looking up knowledge as well as for capturing wrap up notes since it could be a little bit tricky and expensive to get live transcriptions from some telephony providers. Great. And I have a few questions here. I know we've got I think we've answered all of them in the chat. So I have a few, you know, common questions we get often about Copilot. I'd love to have us discuss here. So one we get often is, do we need a perfect knowledge base before implementing Copilot? What if our knowledge base is poorly organized? This is a really common issue we see. Right? Michael, could you talk a little bit about that? Most definitely. This question comes up all the time. It's a great question. So the AI models have become sophisticated enough to provide reasonable answers even with gaps in your documentation. Our Copilot works effectively with limited knowledge bases by using past tickets, internal knowledge, and existing content, and it doesn't need polished customer facing documentation to work well. That said, the, you know, the better your documentation is, the higher the out the higher quality output you will get as well. We've also seen successful teams without robust documentation see immediate productivity improvements, And waiting for perfect knowledge prep can really delay your productivity gains when agents are, you know, needing better tools today. One other thing that I would like to note is that AI really excels at navigating and searching through disorganized information, which is particularly challenging for human agents. So the Copilot is really acting, like a much better needle in the haystack finder, which also makes poorly structured knowledge bases more accessible and usable to the end agents. Great. And another question we often get is how quickly can we get this up and running? Implementation is a key question for a lot of customers. Yeah. I'm I'm happy to take that one. So I say I would say that the TLDR is that we move very, very quickly at Assembled. So, typically, we're looking about at about, like, one week for implementation and setup, and and that will include, like, connecting your internal systems, configuring your workflows. Configuring your workflows. From there, we would typically, like, kick off, you know, really focused one or two week agent testing period where we wanna have, like, a nice feedback loop, see if there's any tweaks or adjustments that need to be made. All, you know, of course, wanting to build that, internal confidence. And then after that, we'll begin sort of like a scaled rollout across your team, which usually happens over the course of, you know, two to three weeks. So I would say all in usually live across the board in about, you know, five to six weeks. Though, of course, if your team is aligned and resourced, I've seen customers move move much faster than that. We're, you know, we're can move as as quickly as as you can on your side. Great. And I see too we have a hand raise here, probably accidental button push. But if you do have a question, go ahead and pop in the q and a. We can't, access the hand raise feature here, unfortunately. But, I have a final question here for the team. I think what we hear a lot too is there are a lot of AI companies out there today. It's really hard to evaluate them all. There's almost feature parity too. Like, we we all are talking about the same features. We're all using the same models. I will Assembled different. Yeah. I'm also happy to take that. Yeah. I I also hear that all the time. You know, things are, starting to get commoditized in in some way. So I I think there are a few different areas in which we Assembled differentiate. I think first of which, as Cassandra was mentioning earlier in the presentation, we are the only support operations platform where our AI agents in Copilot share the same knowledge and workflows and data as your workforce management operation. So every part of your support operation, it's all working in sync and not in silos. And and, of course, that means we're all driving towards the same KPIs and and sharing the same knowledge. I think, secondly, we've also just been in the support game longer than a lot of these newer sort of AI first players. So I I really do think, you know, we know support better than anyone, and that shows through our features, through the adoption of our products. We're also, you know, incredibly customer obsessed and are partnered with some amazing, amazing support organizations that have really, like, played a pivotal role in in shaping our product and road map. And then lastly, I would say it's it's really just the the the quality of answers that we're able to generate through our quality measures, through our multi model setup. And also just like that style and tonality that I was giving you a walk through of our style guide. I think we're just really able to maintain a very, very high quality of standard through a lot of those different measures that that we put in place there. Love it. We love to hear it. Great. So I think that wraps up our our q and a portion here of the session. I would love to prompt you all if you do have more questions to, of course, get in touch with us. And I wanna leave you with, a few housekeeping notes here. So first of all, thank you all so much for being here with us today. We know you've taken time out of your busy day to join us. So, truly, we appreciate it, and we hope it's time well spent. We'll send you a recording, in a follow-up email, if you'd like to revisit any piece of this session. If you'd like to get a custom demo or just learn more about Copilot or anything on the assembled platform, you can simply just reply to that email we send you or head to our website, of course. Also, we run webinars like this regularly, and I'm really excited to, share a few coming up in our August lineup. On the thirteenth, we're gonna be joined by our partners at BlueWave Technology Group to share some common pitfalls in AI initiatives and how to avoid them and really how to support your team for real change. Obviously, you can bring on a product, but making sure change management happens in the right way is an essential piece of your AI journey. So really excited about that session. Then on the twenty seventh, we're gonna host a product focused webinar that really dives into exactly why you need modern workforce management in the age of AI. It's sort of what we're really bullish about in assembled combining AI and humans in one platform to see all of your workforce, your hybrid workforce. So we're really excited to share those with you. We do hope you'll join us. So keep an eye out. We'll stay in touch with the details. And thank you all again for joining us. Have fantastic Friday and a great rest of your weekend. Thanks, everyone. Thanks.