Name: Claude for WFM and Support Leaders
Uploaded: 2026-06-04T18:16:15.146Z
Duration: 1 h 56 s
Description: Claude for WFM and Support Leaders

Transcript for "Claude for WFM and Support Leaders": I've got my three drinks here. Alright. Okay. Hello, everyone. Hello. Hello. Happy Thursday. We'll wait about a minute for everyone to join. Just turned 10AM here in San Francisco, but we're gonna get started here shortly because we have a lot of content to go through. Hello, SF Sunny skies. We've got our chat window right here if you'd like to tell us where you are coming in from. Always fun to see where folks are are calling in from, so to speak. Yay, Austin. This is where Thomas is today. We will wait one more minute. Let us know, guys, and then we'll get into intros. Maybe thirty more seconds here. Detroit. Oh, Liverpool. Alright. Home of The Beatles. Bend, Oregon. Oh, Chapel Hill. Amazing. All over Cork. We have folks from all over. Amazing. I love to see it. Oakland, you're close by, Ryan. Oh, no. That was oh, chat went too fast. Las Vegas. Great. Oakland, California, Texas. Temple, Texas. Amazing. Alright. Well, we are about a minute in, so we will get started here. Just wanna welcome everyone. Thank you so much for joining us today. I'm Cassandra. I work on product marketing at Assembled. Assembled, if you're unfamiliar, we're the first AI native workforce management platform built for modern support teams. And I'm so thrilled today to be joined by Thomas O'Rear, the senior director of operations at Future Health, who's gonna show us how he's using Claude at Future Health today. And, Thomas, could you just give everyone a little intro, a little background on yourself and Future Health? Yeah. Absolutely. Sandra, thank you so much for giving me this opportunity to speak today. So as you mentioned, I'm Thomas Rear. I work here at Future Health. We are a direct consumer telehealth, solution that's in the medically assisted weight loss space. I run both the CX sales and clinical partnerships teams here. Been with the company for two years, and I've been bouncing around in CX and operations for the better part of, just short of twenty years, which is a nice way of saying how old I am. That's okay. And, future, how crazy, I mean, with all the the GLP one craze. I know you guys have grown, I mean, in just the last couple of years to, like, millions of patients across almost every state at this point. So really crazy growth you guys are seeing, which definitely puts pressure on, I'm sure, for you and your team to be using AI, which is a really great, segue into our our combo for today, which is all about cloud for workforce management and support leaders. So just a couple of housekeeping items first. I also wanna mention, backstage, we have a couple of assemblers. We've got Alex Hajir. He's one of our solutions engineers on with us. Alex will be answering your questions in the chat throughout. So if anything comes up as we're having this, conversation and we're demoing things, feel free to pop that in the q and a or the chat. Alex will answer it async. We'll also we've slated a little bit of time here at the end of our session for some q and a, whether that's for Thomas or for the assembled team. And a final housekeeping note I wanna mention, I wanna alert you to our little Zoom button since we are sharing our screen. Lots of demos today. There are a couple little magnifying glasses there where you can see that we're sharing our screen. You can zoom in, adjust the screen accordingly, so go ahead and do that if you need to. And with that, I will just jump into some agenda items here to orient us. So, as I mentioned, we do really wanna spend most of our time today in Claude and in conversation with Thomas to talk about some of his real workflows. But for that, a couple brief slides, I wanna give you a brief primer on, a few core concepts in CLAWB, calling it some CLAWB one zero one. Then I'm gonna show you just a a demo of what I call the possible, just a few quick examples of what you can do when you connect CLAWB to your workforce management data. And then Thomas is gonna take us out of demo mode. We're gonna have a conversation about how he's using Claude every day at Future Health, and he's even gonna show us a couple of the really cool workflows that he's made. So let's get into it. Oh, and I mentioned some q and a too. Let's get into it. What I would like us to start with a little bit here is our framework for our three stages of AI adoption. Just wanted to set some context in the sense that there's a lot of folks here today. Everyone's at different stages with using AI, whether that's Claude or ChatGPT, Gemini. I mean, there's a million tools, a million models that changes so much. But we have this really simple framework, that that we like to use, and this is also what we've, what we've seen the journey most most folks go on when they start to use Claude or another AI tool. We call it ask, act, Automate. So pretty simple, but in the beginning, if you are just starting to use an AI tool like Claude, you're kind of just using it like a Google, a better version of Google. You're opening Claude, typing a question, getting an answer. Most people are at this stage, past this stage even. The next stage we like to talk about is act. Ask, act, and act obviously is when, AI is just taking the next step for you. So imagine rather than just getting that answer, you're also going to let AI post an update to Slack or update a schedule, file a ticket. You ask, and your AI assistant is acting right from within that chat window. And then finally, automate. I mean, this is the stage where work is running without you starting it. Some people call this, like, being a or whatever you wanna say. This is really where work is running without you interfering. You're just setting the rules. Something's running in the background. Perhaps you're using a Claude cowork, but you are the operator, and AI is just working in the background for you like your own employee. Most folks move along this line somewhere, you know, between phase two, phase three. Just wanted to say that everyone's on different stages today. We only have an hour, but what we hope to do is really accelerate you maybe from one stage to the next. And whether, you've never even opened Claude before or if you're super advanced and you're just looking to hear what other leaders in the space are doing or hoping to hear more about the assembled MCP in particular, we hope that this session can move you forward on that journey a little bit, and we'll certainly have follow ons, in the future. So let us know how else you'd like to learn about this topic. So with that in mind, let's just start with a little bit of one zero one, just a couple core concepts that we'll need for today. Two things in Claude, that unlock all those stages, ask, act, and automate. So connectors and skills. These are two things that, we need to know about for today. So connectors are very simply how Cloud reaches your data and tools. You may have heard the term MCP. This is the technology that underlies, and makes tech, connectors possible. This is also, what we launched and assembled recently. We have an MCP, so now we have a Claude connector. And a Claude skill, also an important core concept here, something we use all the time in assembled, It's a reusable set of instructions that you give Claude to perform a certain task. These will both, illuminate shortly. We're gonna do, we're gonna see these in action in a minute, but a few analogies that help me to to understand these concepts in cloud a little bit more plainly. So simple, a connector. It's just like an integration. You can see the example here of the many connectors that I have that we have at Assembled. It lets Cloud work securely with your systems. So whether that's assembled, whether it's Slack, Salesforce, HubSpot, your BI tools, whatever it may be, there's a growing list of hundreds of connectors these days. You can work with those real systems without all the custom integration rigmarole. And the great thing about connectors is they go both ways. Right? So it's not just access to your data, but you can also take actions in those tools, otherwise known as read and write access, some folks would call it. But either way, you can access your data, but you can also make changes to those tools right from Claude because you are connected to them via connectors. Now skills, very plainly, I like to say a skill is just an SOP for an AI agent. Right? It's essentially like, when you document a new process for a new hire, but for Claude. So you just define it once. You're gonna run it with one command. And what's great about a a skill is, effectively, it's it's just a zip file. So, a skill, it's something that's reusable. It's basically a prompt and some instructions, and you can distribute them however you like. What's great about today, as we mentioned, is we've created a pack of skills. If you go and look in your little, chat window, there's also an item called docs. You'll see there a a link to the free Claude skills pack. We've created a set of Claude skills for workforce management and support leaders, and you can go and download them, add them to Claude, modify them however you like, distribute them however you like. And what's great about skills too is if you have a shared Claude account, everyone on the account can access the same skills. But if you don't have a shared cloud account with someone, you can still send them in a file, the zip file, the same way that we're doing today. Skills are very, very useful for these repeatable processes. Again, something we use constantly, and I'm excited for you to try the ones that we've created for you out, and you'll see more of them today. Now finally, I think the most powerful thing we just want to express today is when you can, connect and reuse workflows with connectors, with skills in Cloud, Cloud just becomes almost like your primary interface. It's a super powerful interface that you can use, in your everyday like, imagine you are, creating automated analysis and reports that combine your workforce management data, your BI tools, all of these things in one place. It's almost like, why do you ever even need to log in to any of these tools at all? I personally use cloud as as almost my primary interface. If I never have to, log in to Salesforce again, I'm I'm very pleased. So, enough slideware, though. I'm very excited to get into Cloud right now, and we can test a couple of these things, show you how it works. Let's just get into it. So first, I would like to show you very quickly how you can connect your workforce management data to Claude, and I'll show you connectors very simply. I'm using the Claude, browser app, by the way. Desktop is somewhat similar, but I'm gonna go to customize here, and this is where you can see my skills and my connectors as discussed. Now when you add a connector, very simple, you can see all of mine here that I use every day. I'm already connected to assembled, but if I wasn't, all I have to do, hit the plus sign. You can browse connectors. They're as I mentioned, Anthropic has a major list of many, many connectors. Everyone's gonna have a a clogged connector, I believe, in the future. But, we are still waiting on, our inclusion on this marketplace, coming soon. But in the meantime, what we have on the assembled side is a custom connector, which is also pretty easy to set up, but I'm just gonna show you how very quickly. So to add connector instead of browse, we're gonna go ahead and hit add custom connector. All you do here is you're gonna name this assembled, whatever you'd like to name it, if it's a different custom connector. Then this remote MCP server URL, it's just a link you can copy paste. It's the same for everyone. The assembled one is the same across the board. I've also linked this if you are an assembled user. It is in that same docs tab, just a text file if you'd like to copy and paste it. It's also on our help center, but just so it's handy for you. Very simply, all you're gonna do, add that right here, click add, and then you'll be all connected. It really takes five minutes. Truly, very simple. But then a note on configurations for connectors. As I mentioned, you have read and write access. Right? So you're gonna wanna configure whatever connector you have. What we always recommend is any read only tool, so this is just accessing your data, we always recommend that you configure this to always allow so Cloud can always read your data. The one thing we recommend requiring needing approval is any right access. So anything that's going to change assembled, if you're going to publish an overtime slot, delete one, create a forecast, adjust a forecast, we recommend that you, set this to needs approval so that Claude will ensure, it's gonna double check with you in the chat. Are you sure you want me to update assembled before I do it? Always good practice there. Alright. So we have set up. Why don't we get into an example here? Okay. So, following our framework of ask, act, automate, we're just gonna ask a question here in plain language. We're gonna ask here, why did we miss SLAs last week? Alrighty. And this is a question this sort of, like, root cause analysis is one of the most popular use cases we're seeing at the moment for the assembled MCP. You can see here we're pulling up forecast versus actual connected to assemble. That didn't need to say go look and assembled. I Cloud will understand that context. But what's great about this is, you know, you could say, well, I have my my team that can do this manually. Well, Cloud can sort of do the same thing that a workforce manager would do manually and sort through your volume, absenteeism, shrinkage, productivity, adherence, schedule alignment, all the ways that you normally would manually. And what's great is it's gonna produce a little visualization here, building it by hour. Great. Pulling this all up. But the other great thing is this loads I I wanna mention is since you're connected to all your tools, I mean, this analysis can be super powerful in a way that no single tool is. Right? So imagine I mean, you're investigating an anomaly across your entire tech stack. So, like, my actuals were way above my forecast. Like, look across my CRM, Slack, or even online. Find anything that could explain this. That kind of investigation is way easier in one interface rather than switching between a ton of tools and, cross referencing all of that. And, you know, speaking of analysis, what's great instead of, you know, creating a ton of custom dashboards and Power BI or what have you, you know, you can you can do this all in one place, cross cut all that data in one place. We have customers, for example, who are correlating staffing performance directly to their customer experience metrics, Tons of different examples like that. I mean, the the examples are endless. But another one I really like, now here we're loaded, but, is have you ever had I think we've all had this experience no matter what field we're in. Any of your leadership or stakeholders are just popping and ask a random question that you don't have a dashboard for. What are your volumes looking like right now? Or are there any trends you can tell me about? And, it really cuts down the time you're gonna spend on that type of ask significantly. So great. Okay. We've got the full picture here, and what I love is you've got some great visualization. Alright? We're missing SLA a little bit here. But, also, you have some analysis on why this happened that Claude can do. Oh, why did this occur? Okay. Well, the next step that I wanna take, again, following our framework of ask, act, automate is I wanna act on this. So why don't we say send this report to my team leaders on Slack and advise them of next steps? And my Slack channel, I've got, it's called support team leaders. Let's make sure we know that. Claude will obviously ask if it's unsure, But let's make this happen in a different tool so I don't have to export any of this. Yeah. We're not just going to to ask questions, get reports. I wanna do it all here. K. Cool. We're pulling it up in support team leaders. Here's the thing that's great is, I mean, you can imagine all the other actions you can take in any of your tools. I mean, depending on the connection, you can and as I mentioned, assembled, you can create over time, PTO slots. You can adjust your forecast with more actions to come. But Salesforce, you can create new records. There's so many things you can do, really. The, the options are endless. K. Sent. Now I'm gonna pull over to Slack here and share this tab. You can see this is my Slack thread. You can see 10:18AM. This just sent. It's gonna say, hey, guys. Hey, team leads. Here's a summary along with what we're seeing and what needs to happen next. Here's my root causes. Let's do this by Friday. Great. Okay. Now I don't need to do that myself. Alright. Now we go back to Claude here. Now I wanna talk a little bit about automating things. This is where skills come into play. So imagine as most of us do, we have to report out on things on some sort of, you know, annual cadence or maybe not in some timely manner, whether that's weekly, quarterly, daily, monthly. I want to have a a repeatable process for that type of weekly readout. I already did this investigating here, but let's do my, consistent reporting. And I have a skill already created for that that I've made, based on the needs of our operation. And to access my skills that are preloaded into Cloud, I'm just gonna slash, and you will see all of my skills here. I've got quite a few, and I'm gonna hit my performance report skill. And I don't even need to tell it to do anything in particular. I'm just gonna say performance report. It's gonna run this based on my information. Also, the prompt that I've created that makes this skill great. We're reading the performance report skill. We're gonna pull data from assembled. It could pull data from anywhere to create this report, of course. And just for time's sake, I have preloaded this in this tab right here, my weekly performance report. Check this out. This is a doc, but this could be any type of, any type of asset. Claude calls this an artifact because it's something it created that you could use elsewhere, but this could be a slide deck. This could just be a readout, something you send on Slack. And a management here that, you know, this is something in a skill. This is this is sort of just the beginning of of skills here. You can totally, create an end to end workflow. I've got a doc, but you could also have a skill that's more powerful and actually creates this doc, then it sends it to your key stakeholders. Also, if you're working in Cloud Cowork, you could set this on a schedule that triggers automatically. It could update a spreadsheet at the same time. It could, you know, update assembled. The options are really endless when it comes to skills. But what's great is it's all in one prompt that can be used by anyone and shared by anyone. So, really, now that we have gone through a lot of the, examples here, ask, act, automate, I'd love to actually turn it over to Thomas and stop sharing here. We've got the full adoption journey here, but it's really just a again, just scratching the surface. But I wanna talk, less about examples and more about what, Thomas, you're doing at Future Health. I won't, get too ahead of ourselves. But I know, for example, I've got this report going, which is great. Everyone, by the way, has access to the skill and those docs that you can use and make your own. You have a report that is, like, a 100 pages long, like a a business review you've made with Claude that's, like, a hundred hundred slide deck, which is amazing. So that's how far this can go. Yeah. But, to set the stage for everyone up, first, I just wanna say, or ask you, Thomas, what does, you know, the support ops org look like at Future Health and just sort of what what are you on the hook for day to day just to set the scene for for the folks here? Yeah. Absolutely. So, as I mentioned up top, I'm effectively responsible for three main verticals. So we have core customer service, which has the similar dynamics as we all see, and and Symbol is a big part of that. I also have our sales and retention team. So I have revenue generating, groups that roll in here. We honor clinical partnerships. So our telehealth and pharmacy folks, we manage them, like, in in a lot of ways. Like, we do BPOs in in the CX space, as well. So a lot of what I'm trying to understand on a day to day basis is what's the health of the business. What's going on in in customer service? Are we meeting our goals? What's going on on the on the partnership side? Are we hitting targets on telehealth, side of the house? And really then also trying to give myself and my team and the rest of the business an understanding of what are the areas where we need to focus today. And so a lot of what we've built is very much around this this that kind of kinda same framework. Give us a form to be able to ask the probing questions, see what the the opportunities are, you know, take action. So we're doing a lot of yeah. We are a remote early stage start up, so we all live in Slack like everybody does. There's a lot of how can we push out, give these insights. Also, how can we use that to adjust and be able to tell that kind of that narrative story? And then as you alluded to, we've done a ton of automations. So to that point, because we have a lot of functions, what I've done to help with our team, so, like, myself or my leaders, there's a lot of reporting out of the business that we do. And so, great opportunity there as we built through, through cloud. We we worked some effectively Python scripts. So at that point, we're we're basically vibe coding of how can I take all of these metrics that we're responsible for, build views that are executive ready so that we can take those and and put those out and then do that across all those functions? And so I do have my monster business preview deck, which I can also give you a little bit of flavor of as we go. Totally. And, you you said something to me too that that really, really struck me recently. I mean, you are definitely an an early adopter and and pretty advanced in this journey, but even, with the release of the assembled MCP and and the ability to sort of talk to your workforce management data more conversationally, you said, like, the job is just completely different than it was six months ago, and you you kind of were were mentioning some of the ways that it's changed, but, what are you doing today that you just you couldn't six months ago? Yeah. So one is you kind of showed a bit in in your presentation. There is a bit of us really taking a lot of the the kind of legwork out of some of the insights we would have. So, like, a great example, I'm an old school workforce guy. If you're doing but, on the hook for those various understanding of tell us why we didn't meet service level. What are the what are the goals? Those are the things right now. Today, I have a lean team, but I can effectively go interrogate, Claude, use the MCP server. We'll be able to get those insights within, like, ten minutes on that. So there's a huge gain there, but, also, it's allowed us to do much more complex pieces. A great example, because we're fairly young, the forecasting that we do on the CX side, it's a lot more complex. If your business has been in in place for twenty years and you have really strong seasonality, awesome. That's that's super easy, and Assemble does a very good job of that. If you're in a world like us where we're very focused on what is the marketing plan, how are we pacing towards our revenue targets, what does that imply in terms of our customer acquisition, that gets a lot more complex. And a lot of what we've been able to do is leverage the MTP server for assemble, bring those metrics out, do much more complicated forecasting using profit, which is, you know, great library in the, in the Python world that was developed at Meta back in the day and be able to do much more of that multivariate, forecasting. And then the great thing about how Assemble has built the MCP server, I can go take that, build that plan, iterate on it, and then push it back in. And so then, effectively, now all of that work to go and plot that out on a weekly, daily, hourly basis, that's all taken off the plane. Amazing. So a lot of time savings and then just bringing in a lot of those different we've talked a lot about the the different data connectors we're bringing in. It's like all these cross cutting across your different tools. But to sort of follow our same framework of, you know, our ask, act, automate, what's the first thing maybe that you will ask Claude each week Just as an example. Yeah. So there is a little bit of tell me how the how the day is gone, how the week has gone. So. I'll go in, say, ask those questions, summarize for main service level over the last week. What were the main areas of variance? What are time frames I should be aware of? Also, to kind of expand that out, how did our marketing acquisition what was our what was our new customer count last week? Does that line how does that look compared to what we see on the customer service side? So I'm able to ask these probing questions to really get a a flavor of what is the state of the business, And then that gives me a clear sense of, okay, what are the areas I need to start to dig into? 100%. And what I love about it too is, yeah, you're not flipping between tabs, and you can also sort of ask, like, some of those what if scenario questions. Like, what will happen if that marketing campaign goes as well as we hope it does? And we. get 30% extra extra volume next month, next week. It's great for that sort of, like, long term long term planning and short term planning. And so in the, the sets of the the automate, or we're gonna go through the the automate phase. Tell me about some of the automations you're creating and whether that's, like, skills, agents and co work, what you've been doing. I'd love to hear more about that one. Yeah. So, this this would actually be a good transition to show you some of the the fun stuff that we've been working on. I'll One yeah. So so one one good example that we've been working with here is how can we get a good hold on our financials in the the CX space? So, we're an organization. We're heavy BPO users. The vast majority of my staff is outsourced. And so what we've even done before Quad is by having our teams in assemble, it has had a really rich view of us being able to to do this sort of BPO billing reconciliation. What are the productive hours? How does that how does that work? That said, there's still an exercise historically of I get the invoice. How do we reconcile? Where are their variances? What are the questions that we need to ask of of our partners so we can go and then, you know, complete and make the payments out? So one of the use cases I've been building was, first, build a corpus of all of of all the financials that we have across our business. So, I have a one stop shop now where I can go and see, here's everything I've received from my BPO. Here's everything I've received from my tech partners. So here are my my envelope costs. Here's my Zendesk costs, and then even from that telehealth side. But the piece in terms of the day to day month to month exercise is really wanted to focus on how can we solve that that reconciliation. So to give you a bit of a flavor, and I'm gonna share my screen here. So this here is this is my quad instance. So what you can see here is if you look at this prompt, I'm basically telling the telling the tool that act as a finance finance analyst. You have expert knowledge of invoicing and the operational management. And the task here is to reconcile the invoice that that we received for the month of April with the hours out of Symbol. You'll see here so there's a lot of you today in the MCP, server on your operational productivity and all that. There is still some deduping that's happening on the the BPO billing page. It's not quite there, and we talked with the team about. But even in this case, you can see what I did is I effectively did that, export of the metrics from that page. So now that I've given it that context, I'm saying, what is the task? We have the financial tracking skill. You already have this invoice. You can see I'm not even having to present that to to the tool. Look for the productive hours to avoid duplication. Here's the site you should use. Here's the main queues to look at. We obviously have some, you know, line items and teams that are not necessarily in our workforce tool, providing the context in terms of what it's doing, and telling it, here's the format that we're looking for. Here's what I want the output to be. And you can see here, it goes through this exercise. It it finds the invoice. It it pulls it out of the Google Drive. It's it does the analysis. But if I flip over, I'll show you what the actual output was. So here, an example in the the lovely view. On this, this is the artifact that it created. So it's telling us here's the total invoice, invoice hours, here's the the bill by hours that we see from assembled, what are variances, And the season cost is telling me what the the gap is. And then here, it goes through this exercise of comparing all the hours, winding up the line items to specific roles, and we see these hour comparisons. What are the dollar variances? Even if I slide down here a little bit, I'll just kind of snapshot. It even gave me the individual descriptions, the different line items. What are the hours? What are the rates? What's the cost? Is it included in this reconciliation or not? So, again, this helps us kind of close a gap of before, this would be a long term exercise that we would have my my vendor manager would be going in, pulling these reports from assemble, and lining up lining all of these up, interrogating it out. And now we have something very clean and effectively where we do have variation. Here's something we can share directly with the vendor. Here's your you here's what we're seeing. Help us understand where these gaps were these opportunities. That is fantastic. Yeah. I love it. I love the visualizations there too. And what other tools could you describe a couple more of the, like, connectors you're using for that? It's obviously assembled, but you've got, like, your vendor too. What what else are you using to create that report? Yeah. So in in this case, specifically, this is using the the assembled connector, and then we actually built a skill internally to effectively house all of our invoicing portion. So in in other use cases, Mhmm. maybe you have you'll have a connector to your ERP and being able to pull that in from, say, the accounting side. But to your larger point, the sorts of connectors that we're able to use and and drive with this, so we have assembled. We also have, connections with Zendesk. Some of those are API based. They're working on some stuff, and that's in that case, it's certainly from your your ticketing system that you have that as well. Another really powerful piece, we do have we we use Google BigQuery, so we have the integration there. So I can also effectively use Quad to go and structure SQL queries, pull that out of our core product, data as well. From, like, a product analytics, perspective, we also use posthog. So another. great piece that I can do here is, like, tell me what the lead traffic has been over the last thirty six hours. Compare that to customer service volume. That sort of, like, deeper analysis is so much easier now because all those tools are effectively talking. And as opposed to putting in the work to build all the business intelligence, data engineering to to bring all those together, you know, a nontechnical, you know, person like myself can go and effectively structure that out over an afternoon. Right. And I'd say, it's interesting. You know, assembled has always been an open platform. We have a great API. It's why, often people choose us for that level of customization that that we allow for folks. But I think what's great about, Claude and and other AI tools like it is there's really a, what I call this, like, an extra persona, another person that is not so technical that they are gonna get into an API and custom code something, but is technical enough that if you can speak to Claude and use natural language, you can build out a dashboard. You can create some automations. It just makes it so much easier for for that type of person. I'm I'm the same way. Like, I'm not getting into an API. I don't I don't like doing that. That's the job for an engineer, but there's, definitely some some workarounds now that we can we can do without, without needing engineering resources now with Claude. Yeah. That's that's a great oh, go, ahead. I'm sorry. ahead. Yeah. Actually, to to riff off that, that is a great point. So, certainly, there's a lot that you can get in terms of you leveraging Claude to drive that intelligence and be that, you know, kind of support there on there. What you'll see in is that if I get into this, a lot of that kind of business, review reporting that I'm talk that I was talking about. Effectively, there, most of that, we are just using APIs. So what I what I was able to do with that is basically build a Python script, gave it a set of, API keys or various tools, and it it is running that on its on its own. So from that perspective, I've invested the the time and the tokens to help structure that, but then that's that's that's a script I'm running on my local machine on a Monday morning, and it's taking fifteen minutes. And then it it's put puts all that out. Could I have done that a year ago? Absolutely. That technology is is mature. Would I, as a customer service director, have been able to structure that on my own? No. And so realistically, the, you know, the fact is dev time in any organization is very valuable, and it's it can be a challenge to get that invested in terms of internal operations as opposed to what our customers touch touching every day. Now it's not necessarily that sort of dichotomy anymore. Yes. A 100%. And, I'd love to if you don't mind sharing that massive report, I'd love to see it. I think the the folks here would also love to see. I was very impressed by this one. I think there is a question even in the chat about, you know, differentiating or or personalizing visuals and and outputs. This is certainly something you can work on in cloud and and make the report specific not only to your organization in terms of the data it's pulling and the the outputs you receive, but also, yeah, the styling. This is something that we also do at Assembled. I mean, just on the marketing team, we have a weekly pipeline review, you know, all the sort of things that we measure for for marketing that is updated in cloud on a weekly basis from Salesforce, from Notion, our our project trackers, things like that. And, yeah, it's all in our, brand guidelines and style. These are definitely things that you can add as inputs and, even on a a 100 a 100 page, or. slide deck. Yeah. So I'll take this opportunity here. So, again, as you mentioned, what you're seeing here this so this is that unified, piece. So the goal here is not necessarily to say folks need to read a 100 pages every, every week, but really have a unified source. So each of my leaders over the areas that they own can come in. They have a shared understanding of what their key metrics are, and then these artifacts are easy to to pull out. And so to your point, as I mentioned, I'll just do some scrolling here. This is this is 98 slides, on this. Now and Juan asked that that question in here to this point. This is set up in our brand guidance, so it has our color scheme, our palette, and all that. The way that I structured that, to to be clear, is as of as I was building it, I shared with Claude our company's brand deck, our, you know, presentation template, and and asked it to build off of that. So it's it's it's even a situation where if you have those artifacts, it's smart enough to go work off of those and then you can iterate as necessary. But the other part in here so, like, as we go along, it'll have the different sections. As you can see, there's there's core metrics. We have a combination of these, like, scorecards. So you'll see some of these numbers here. If I come up, I'll I'll be able to put an example here. Like, this is a very traditional, piece, and I'll just plop this up. So, like, here's what phone looks like, what between service and retention. You've got some of your very traditional metrics here on it. You know, actuals and forecast. The other part here and if I come, there's these various trend lines. Like, a great example here, I'll take this. So we use assembled, voice AI agent. And so there's a lot of rich reporting sits in in the panel in terms of your, overall performance and what that looks like in terms of your containment and your resolution. But, again, if I wanna go show that to my COO, she's not gonna go log in to that. So here in this example is a much more executive friendly way where we tell that. Here's the trend. You can see what is the volume coming in, what's being resolved, what's what's abandoning, and what is that the total resolution percentage on that as well. Another one that I will also highlight here, I come down. So, again, we also use Copilot. So big thing that we were able to drive on is what is our automated customer sentiment and intent, on this. As folks know, you know, CSAT is very important. There's a great view of that. I get one, two, 3% is a great time frame for responses, and so that doesn't really tell you a whole lot. And, yes, QA would love to go and, you know, have deep evaluation of every ticket, but that's just doesn't scale. So we leverage the tools. And so as an example here, what you're seeing is this is the overall sentiment. So I can I can see if I look across my time period, or, you know, we're kinda hovering in this, like, 60 to 70%, yeah, positive intention, and then that negative piece is is obviously down? The next portion here is intent. So, historically, we did a lot of reporting on ticket tags that are applied by human agents. That has its own challenges. But what we were able to do is, one, we leveraged a tool to be able to develop these these intents, and now I'm able to leverage Quad and our various pipelines to take that out and start to tell that story. So here is specifically in this case, most of our contacts are about the medication servicing plans, which kinda makes sense. And then you have account management being a portion, and that's where you're gonna get a lot of cancellations and all that fun stuff. So this allows us to basically take all these these insights out. These are metrics that we always have access today, but really reducing the friction to be able to see that, ingest that, and then decide what action we wanna take. And, Thomas, I'm curious. How often is this deck, updating? Is it and it's automatic? Yeah. So we we are running it on a weekly basis. So as you can see there, it's it's always set to look at the previous previous week that's that's settled. What does that compare to the week prior? What does that look like month today, quarter today? And we're able to apply that across all those metrics. So I have it I have it set up. There's even there's even some manual reporting that goes into there. Like, we are like a lot of shops. You scratch, a little bit deep in some workflow to end up with spreadsheets. There's a little bit of even direct adjustments in ingesting that's going on there. So we run it every every Monday. Takes about fifteen minutes to build, between pulling out all that data in. But we even set it up to be dynamic. So if I wanna go and say, let's change the time frame and look six weeks ago and how have we progressed? I do have the ability even to do that and basically kind of readjust what time frame it's looking at. Now I'm curious. I mean, sort of to, a segue here, but, something that I I think we've heard a lot, in the community is, security. Like, how has your team thought about, you know, using cloud or an AI tool like ChowDBT for data like this? Some folks say we're just blocked on on using something like this, or we can't use it for security purposes. Like, what has been your approach to that, at at Future Health? And just maybe talk a little bit about how that works for for you and your team. So that's a great question. Because of the nature of our industry, we're in, you know, health care and telehealth services. So while there are certainly firewalls in terms of what's the source of, you know, personally identifiable health information that goes to our partners, One always has to be careful because customers will be very, direct in terms of when they're talking to human agents, and so there there is we treat a lot of our internal CX data like, you know, conversations in Zendesk or recording recordings of conversations that customers have with Cal, the assembled, VoiceBot. We treat them as if they are PHI, and so from that perspective, we are very segmented. So the reason why we as an organization went with Quad is we were able to take advantage of the robust, you know, HIPAA structures that Anthropic has. We have a BAA with them. The other piece on that, our internal data science and engineering teams have an also building an MCT layer on top of our internal data as well. That allows us to be very clear about what are the what are the structures, what's the, you know, data we wanna make available to certain models where the the data we we want to intentionally block. And it also allows us to be agnostic. So this space is is changing. You know? The the models are effectively going through a whole revolution every three months or so or something of that sort on that. So today, quite frankly, anthropoc has the best models. Six months from now, is that gonna be the case? I don't know. It might not be. And so we've been very intentional about building this to be agnostic so we have the opportunity to to take advantage of the best technology. Yeah. I think that's a great approach. Right now, I am also all in on Claude Claude Pild, as they said, is the best tool. for enterprise work. I mean, if you're connecting different, different platforms you use for work every day, I think it's the easiest one. But then now, I Cloud Code was the winner for until just a couple months ago. Now codex is apparently better for coding, and it's such a it's it's moving so fast. It's I'm in AI tools every single day, reading the news every day, and it's so hard to keep up. So, you know, just building for the future, I agree. Being model agnostic is is the right way, but fundamentally, I think a lot of these principles all apply sort of, you know, the sort of system thinking that you need to to work with an AI tool can apply across them since they're sort of commoditizing. That's right. But, That's right. you're right. I I'd also love to ask, you know, in the next couple of minutes here too, like, has this spread on your team? Like, what's the what's the sort of, like, difference between you're using it primarily today? Are other folks on your team using it? Are they using your skills? And do you plan for that to be different in the future? How how is this sort of spreading across the team? Yeah. So we've we've very intentionally gotten access to effectively all my internal team, on that. So, sure, I'm there's a lot that I'm doing and and building some standardized skills so we can make that available to other folks in the organization. But, all my leaders are also setting up their own distinct, structure. So if I brought my service delivery manager on, he would be able to show you, like, all the wonderful things he's done to understand kind of function by function within his organization. Here's the the heartbeat of the, you know, what we're doing on service. What are we doing on, our white glove support function, which is actually monetized support that we do on that. So he's been able to build those structures. So his leads are over each one of those functions, are able to independently work with these tools, and all of that. And then I've kind of set some standards, but but, eventually, the the goal there is we wanna get this in the hands of everybody because I feel very strongly sure. There's always questions about, this is gonna replace my job or take take me out of that. It hasn't, but what it does is it just, you know, raises the bar in terms of what what are the capabilities that I individually have as a member of the team that I can deliver without necessarily needing to bring in, like, the complex tools or the dev time or all those other resources. Alright. I think, you know, for everyone here in the the contact center world, workforce management support, we've uniquely seen, this, like, rise of, okay. We're gonna all, get our frontline. We're gonna you know, frontline agents are all gonna be AI now, and we've seen how that did not play out. Right? That's right. You know, That's right. didn't replace everyone's jobs. The the work just sort of changed. So for everyone too in the back office managing, all those frontline agents, it's not that, you know, an AI is gonna do every single thing for you. Maybe the work is different, but, yeah, it's just about managing those agents and managing the changing scope of your work every day. So I I'm glad that we can all see that almost uniquely in this industry play out more than more than others. It makes me personally pretty pretty optimistic about the future with AI. I'm very, not a doomer over here. That's for sure. I think it will all just make us better. So I think we have about ten minutes here, by the way, for q and a. I'll switch back over to our little q and a chat. We have some questions in the chat window we could answer here. I know there have been a couple we haven't gotten to yet, about ten minutes for this. So if we don't get to every question, folks, by the way, we, will answer a sync. We have your contact information. We will make sure to, email you after the event. And there was a question here about, this event being recorded. It is and it will be sent to your email, about ten minutes after the event, so you can share it. Let me just pull this up right here. Q and a. You have, quite obvious, but you do have the the q and a in the in the right here. So a couple of questions. One we have is at Future Health or any other company, which team drives the skills or development of those AI building? Is it usually driven by the individual team members? That's a really great question. Thomas, would you like to answer for for future health how you guys are approaching this? Yeah. So there's a little bit of tops down and bottoms up that is going on. So on one hand, our data engineering team are building a set of very standard skills, that we're deploying across the organization. A lot of that's where we focus on how do we answer our, you know, questions and leverage the specific internal data that we have. So, like, what's our customer acquisition? What's our patient progression? There's a lot of that. Individually, in each of the functions, we we have taken on developing, you know, kind of specific solutions. So within my CX and partnerships org, that's why you've seen us build some of these defined skills for us to be able to go and say, okay. Here's our unique use cases. How can we solve for that? That said, we have encouraged individual members of our team to to go and develop their own solutions and and let that not only inform their workflow, but then for us to think about, okay. What can we take here and scale this across the organization? Great. And I think, from what we've heard too and and how everyone's approaching this in in support or elsewhere, this is such a growing, a growing part of any organization that I think we're all sort of just testing the waters who who owns what. I've seen roles come around for just, you know, AI ops, AI internal tools builder, tool builders, things like that. So I think, this is all sort of being determined in real time. We sort of take a similar approach and assemble different teams, create different AI agents and skills. But, oh, wait. We do have a question here too I I wanna get to actually about, maxing out on tokens. I think that's an interesting point. This has been in the news a lot recently too, the tokens. Yeah. That's, right. I'd love to hear how you're thinking about this. Yeah. So it's a great question because right now, one, it's it's an uncapped cost. And, also, it's not it's not lost on me. It's not lost on on any of us in the in the industry. The the charges that are the cost associated with tokens today don't necessarily represent all of Anthropic's costs or OpenAI's costs, and so those are obviously gonna change over time. So the way I've been thinking about it is how can we develop so take advantage of Cloud to develop solutions that are not necessarily always have to live in there. So great example, that monster business review deck where we're structuring all the metrics. There was a a decent investment in tokens upfront to build this the scripts, you know, kind of QA the data, all of that. But then at the end of that, what am I doing? I'm running a Python script locally on on my Mac, and I'm just hitting a API key. So that's not necessarily anything different. So from that perspective, that's where we set a foundation, and now we're we're intentionally bringing the workflow out of clause so that we don't have to manage that. The other way I think about it is how can we take these consistent insights and push them out on there. So we do a lot of running things in, in Slack where we go and have notifications. Or, like, I'll give a great example. I've been working with my product team where we do this, like, a unified update across the organization where we're even, like, pulling in, like, as you mentioned, some of my customer service pieces, pulling some stuff from Jira. We're doing fun things where we're tagging, Slack messages with a specific emoji to go to pull in. But, again, what's the what's the point there? Take advantage of what the tool can do. Put it in an output that everybody can utilize so we're not necessarily recreating the wheel every time. Right. That makes sense. And I think, you know, what we've seen as, we run up against our, token limits to it at assembled and what we've been hearing from folks in the valley. I mean, I've heard I'm sure you guys have heard all these hilarious stories of, you know, teams all of a sudden getting a token budget and then maxing it out their their month long token budget in two days. Oh, You. know, I think we'll we'll all adjust the same way. Like, you have a, if you have an unlimited resource, you're definitely gonna gonna use it for for everything and not be so, careful with it. Once it's limited, I think folks will will understand how to, how to better structure their prompts, how to best use their tools. And there was also a question about, different models in Plog. There's, the I I think now they're calling it, like, low effort, high effort, medium effort. It's the different models. There's, you know, SONNET is sort of your standard. Then we have Opus. Opus 4.8, I believe, just came out a couple days ago. It's how fast everything's moving. That's how Claude classifies it very simply. Like, there's a model for lower effort task, better effort task. I would definitely suggest going on the anthropic, help center docs, and they'll describe, more of the granular differences between each model. I think they'd say, like, the highest, the hardest working model, which is the latest Opus, is better for deep research, complex tasks. Through your everyday tasks, you can use more of a sonnet that will, will eat up less of your tokens. So that sort of optimization, I think, is is starting to happen too for folks who are realizing this is not all free. Alright. Let's see if there's any other questions we could answer in the last two minutes. I think you know what? We I'd love to to give a couple, plugs to what we have coming next. So if we didn't get to your question, we will follow-up with you. Again, there will also be a recording here. Also, when we do send you the recording, via your email, you can always just reply to that email. The assembled team will get back to you. We'll follow-up on your questions specifically. So check your inbox for that recording, and we'll reach out proactively. Also, I wanna plug for everyone here. Please make sure to grab your free pack of Claude skills. There's the q and a in the chat window there, but there's also docs. All you're gonna do is click on that pack of free Claude skills. This is what it looks like. You've got instructions for how to add these to Claude. You can even add these to your personal account. Just try them out. We've got the performance report, which we demoed earlier. We also have a skill called shift check. We have a skill that reviews your forecast. There's instructions on how to use these and what they do all in this doc that is linked there, in the docs tab in your window. So check your, so make sure to go grab that. I do highly recommend go try them, test them out, play with them a little bit, and then you can modify them. I think stills are most powerful when they are completely personalized to how you work in your organization. I also want to mention for anyone who is gonna be attending CCW customer contact me Las Vegas later this month, we will be there. We have a booth. Of course, we're also hosting a really great session I'm excited about. It's called From Reactive to How WFM Leaders Are Building the Agentic Operation. So sort of following on from from this conversation. We're gonna be joined by two really great senior workforce management leaders, Matthew McKay from Block, Christopher Trujillo from Memorial Hermann Health System, along with Ryan Wang, the CEO of Assembled. So please, if you are attending CCW, come say hi. Come join our session. We're super stoked to see you there. As always, you can visit assemble.com. Go check us out, learn more. You can get in touch with us there as well. And, again, just wanna say thank you so much, Thomas, for joining us, for sharing how you're working in Claude, sharing some of your great artifacts, your workflows with us, with everyone here. We're just so excited to hear from you, and, it's one of the the biggest users of RMCP. Your insight's always super valuable. So thanks for joining us today, and thank you all for the hour. Appreciate it. I know, time is valuable for everyone here, so thank you so much for joining us, and I hope you have a fantastic rest of your week. Thanks, everybody. Great. Bye, guys. Have a good one. Thank you.