Name: Beyond Copilots: AI That Actually Executes and Explains
Uploaded: 2026-02-24T18:16:37.126Z
Duration: 34 min 8 s
Description: Beyond Copilots: AI That Actually Executes and Explains

Transcript for "Beyond Copilots: AI That Actually Executes and Explains": So thank you so much for joining us today for our demo on Beyond Copilot's AI that actually executes and explains. And let's jump right in. So let's it's kinda just to sort of ground our conversation today. You know, why can't workforce management be solved with just assistive or, you know, black box AI? So what really happens in a lot of cases is more traditional platforms may rely on things like dashboards or alerts or, you know, things that require human intervention. But as we start looking at modern environments, you know, you really are wanting somebody who can help you automate decisions that, you know, continuous continuously also balance all of the different criteria you need to think about, like labor laws or union rules or preferences, budgets, you know, all the things that go into your decision making and be able to do that at instant speed. So really freeing up your managers so that they could focus on the decisions that really make an impact on your business, not the daily tasks. So Legion AI was built to handle exactly that, and that's what we're gonna talk about a little bit today. It's how Legion AI bridges the gap between people, data, and automation. So just to kinda give a little bit of background, Legion was founded ten years ago, and we were founded and built on AI from the very beginning. So everything we're gonna talk to you about today is, based on Legion's experience and expertise in AI. On a daily basis, you know, you can see some of the metrics there for some of the, you know, scale and operational maturity that we have. Each one of these metrics, it represents real workforce decisions that are executed automatically. So not, you know, human in not just insights that are waiting on your managers to take action. And then we've already been recognized. So Legion has been globally recognized for our AI innovation and our excellence in workforce management. We've included just a few of those awards there today. So with that in mind, I'm really excited to introduce you to Joe Cornish, who's gonna walk you through a little bit more in-depth all of our Legion AI capabilities, as well as show you some of these capabilities. So, Joe, if you'd like to jump in and and take away. Yeah. Hi. Thanks so much, Malysa. Thanks for the introduction. For for those who haven't met me, my name is Joe Cornish, and I work in the solution consulting team here at Legion. My role is essentially to work with organizations that rely on frontline workers, helping them use their workforce data and AI to make better schedules, better decisions, and ultimately to make, better employee and colleague and customer experiences. It's really exciting to see such a diverse group of organizations that are joining today from grocery, luxury retail, aviation, health care, and more. So as we go through this demonstration, I'll try and keep the examples operational and principle based so you can map what you're seeing in the, into your own environment. So before we get into the assistants themselves, it's kind of worth clarifying, what we mean when we talk about AI inside of workforce management and what makes us different as Legion, because we don't see AI as one single capability. So there are different categories of problems that this solves, and they all build on each other. And at Legion, we kinda see this as an involvement of our a of our AI capabilities over time. So we've kinda got three pillars here. So predictive AI is all about anticipation. So it's looking at things like historical patterns and signals to forecast what's likely to happen next. So in workforce terms, that might mean how do we predict customer traffic, how do we forecast transactions or workload, How do we identify those peak periods? And how do we anticipate demand spikes around things like events or seasonality? Essentially, what it's doing is it's reducing the uncertainty. And if you know tomorrow is likely to likely to be busier than last week, then you can kind of plan accordingly. But predictive AI still leaves that decision with the human. It doesn't tell you what might it tells you what might happen. It doesn't tell you what you necessarily need to do about it. So then we go to the next layer, which is our deterministic AI, which is rule based automation optimization. This is essentially where we take some of that business logic, and we apply that consistently. So, for example, generating labor guidance based upon that forecast, building those schedules of compliance rules, applying different skills and requirements and availability constraints, respecting that contractual regulatory guardrails. So this is kind of bringing that structure and consistency. So instead of managers building schedules differently, the system applies that logic uniformly. So even here, someone's still got to interpret that output, and they still got to review it, and they still got to adjust it. And then finally, our third pillar, which is AgenciKi, and this is the next layer. So this just doesn't predict. It applies it doesn't just apply those rules. We interact with AgenciKi. It explains. It can recommend. It can execute within defined guardrails. And this is the real shift where, essentially, we're moving from here's your data to here's what's happening, and here's what I recommend. Before we go into the platform and I'll show you how Legion's solving some of these problems, I kind of wanna re quickly frame what we're what's live today. These aren't concepts to Malysa's earlier point. These are assistants that are operating inside of Legion right now, and we're gonna be demonstrating some of those today. But they support different roles across the organization. So today, we'll kinda look at four areas. We'll have a look at our upper field assistant. We'll look at our schedule summary assistant. We'll have a look at shift explanation and forecast and explanation. And each one of these can solve a slightly different operational challenge. So some of these will be field focused, some of these are store level, and some of these are about trans transparency and trust. So rather than talking about a theory, let's start with our first one, which is our upper field assistant. So the upper field assistant is designed essentially for field leaders, so regional managers, district managers, but really anyone who's overseeing multiple locations. And as we all know, the challenge of a, regional district manager is too much data and just not enough time. They want to be able to open they don't wanna have to open 20 different reports, look at store reports and information. They quite simply want to know and understand which districts are over and under budget, where are we understaffed, and what location stores teams need my attention and support. This assistant essentially brings a high high level schedule health check view across regions and districts. But more importantly, it translates some of that complexity into action. So it's flagging outliers. We're highlighting risk, and we're supporting some of that prioritization for the regional teams. So instead of a reactive reporting process, we've got some proactive coaching and insight I can use in order to support my team. So rather than describe that in theory, let me show you how that works inside of the Legion platform. So let's, for a moment, step into the shoes of a regional leader. I'm responsible for multiple locations. These might be stores, teams, clinics, client advisers. The challenge is always the same. I don't have the time, essentially, to open up every schedule, run multiple different reports, manually reconcile by budget versus hours. I'm looking for clarity really quickly. So here's my divisional, view, and I click on my schedule assistant here for upper field. And I'll ask you quite simply, how are we performing this week? What you'll notice, immediately is this isn't this isn't the this isn't providing me with raw data. This has given me a translated operational summary. So I'm reporting on the budget by forty nine hours. My published hours are reconciled clearly against my labor budget, and I've got a store by store breakdown. So for a regional or divisional manager, this is compressing five, maybe more minutes of analysis into five seconds. And that can really matter because field leaders don't suffer from the lack of data. They suffer from a lack of clarity. So the faster I can understand whether we're healthy or at risk, the faster I can understand where my intervention matters and what I can do to support the team. So if we take that to another level now and we look at, how we can move that from a really simple summary to a level of prioritization. So we'll ask, and you can converse with all of the AI assistance that we're gonna see today and ask the specific prompts. But we've just asked what stores are trending over budget for this week. So, again, instead of having to manually scan every location, the assistant can identify that there's actually only one location that's slightly over budget, two hours over in this particular instance, less than a percent variance. Is that oh, sorry. Three point five hours over in this particular instance. This isn't necessarily a moment of panic, but what it does do is it gives me early visibility. And, essentially, early visibility is what's helping me protect my margin. You can imagine three point five hours in one location might not sound too dramatic, but if we multiply those variances across forty, fifty sites, across fifty two weeks, we would see that information before that compounds, and that's where a lot of our leakage could happen. So what you'll also notice from this prompt here is it's not just flagging the store. It's showing the budgeted hours, suggested hours, and published hours. And I can see whether this is an optimization logic, a manager override, or something that might need coaching, and that transparency builds trust with me as a regional leader. At scale, this could also standardize how the labor performance is interpreted across regions. So the AI assistants that sit within the environment understand your language and what's your, schedule is telling you is making it better or worse. And so instead of having each leader analyzing their data slightly differently, you could use this to create consistent decision logic across the entire organization. So, essentially, what we're seeing here is is live interactive insight, but this capability can also scale beyond basic ad hoc questioning. So Legion can automatically generate concise AI driven weekly summaries of labor efficiency performance across regions, roles, and departments. And for senior field leaders, those summaries can be delivered directly through to email, giving you a kind of structured narrative of overall performance, what trends and, variances are emerging, and some of those focus areas. So instead of kind of logging in, digging through dashboards, looking at, the different signals, the the the leaders can receive that proactively. So this can kinda transform complex operational data into something that's clear, digestible, and immediately actionable. So I wanna kind of, move on from upper field assistant into our scheduling assistant. So this kinda takes that what we've seen from the upper field assistant and brings that closer to the store. The scheduling assistant is essentially built for the store manager. This is or the person that is reviewing those schedules and publishing them on a day to day basis or a week to week basis. And as we'll know, auto generation predictive scheduling is super powerful. But when the manager doesn't understand why a schedule looks the way it does, they'll make changes, and they'll edit that. Sometimes that can be unnecessarily. So what we're trying to achieve with the schedule summary assistant here is an assistant that provides conversational clarity into things like labor versus budget alignment, individual staffing risks, hours by role, and any improvement opportunities. We know when the manager understands the schedule, they trust it. And when they trust it, they edit it less. And when they edit it less, you protect that optimization. So, again, let's have a look at the schedule summary assistant in action in the UI. So, again, as a store manager or a site manager, this is essentially my reality. I've got my schedule in front of me here. You can imagine I've got all my shifts across the week. I've got my coverage. I've got a budget alignment I've got to stick to, and I've got all my compliance considerations. And this is essentially where the complexity lives. All of this complexity looks slightly different depending upon your industry, but I'm sure you'll agree that the pressure is universal. As a store site location manager, your job is essentially you're accountable for getting the right people in the right place at the right time without any overspending. Traditionally, when we review a schedule, this means we have to scan day by day. We're looking for gaps. We might be reconciling that guidance and going back through making sure the hours match the particular work roles that I'm looking for. Maybe I need to go back into the forecast and understand the shape of the day to match that labor. And this can be where store managers' friction or site managers' friction can creep in. And that's not necessarily because the system is wrong, but it can create that uncertainty, and that uncertainty can create doubt. So scheduling assistant in Legion, instead of kind of asking for that or scanning through that information manually, I can quite simply ask the scheduling assistant for a weekly summary. I'll be provided with that summary here. And what this assistant is essentially telling me in simple terms is we're slightly under budget overall. We've got some early week coverage which we need to align on and which you'll see in staffing and coverage concerns here. So it's it's it's letting me know what my shape of my week's gonna look like potentially. And then later on the week through Thursday and Saturday, we've got some coverage concerns. And then right down the bottom here, we're summarizing all the takeaways and recommended actions from the entire schedule process. So if I want to get to a quick tangible understanding, maybe I'm looking at this halfway through the week, for example. I'm being given information that's tangible that all the data points that are sat in the system are being combined together to give me some recommended actions. And these things are actionable. I don't need to look at percentages. I don't need to know where the risk is. All of that information is being populated for me. And it's also kind of highlighting one open shift, and several unscheduled employees who could close those late week gaps. So so you can see I've got 55 with one open shift. I've got some minimal, task risk, and then I've got 18 eight employees that are still remaining unscheduled for this week. And this this information could be really important because when managers don't have that clarity, they'll then make changes or they'll compensate. They might add some buffer hours, and they'll override different shifts. They might build safety into the schedule. And that's not necessarily because they're being careless, but that's because the uncertainty can drive that defensive behavior. The scheduling assistant is designed to just reduce that uncertainty for them. So instead of analyzing guidance, instead of looking at demand and staffing all in separate reports and go through different screens, is creating me one operational narrative. So we're not just summarizing. We're highlighting risk, and we're recommending certain actions. So, again, as with the upper field assistant, let's follow-up with a further prompt. So I'm short on a particular Friday. So I will ask, again, Converse with the scheduling assistant, and I will ask you to add an open shift on Friday for me from nine till five. Okay. So as you can see there, this is the real shift where we see Shared Genesis adding the biggest value. So we've moved here from an explanation and a summary of how my shift's gonna operate through to execution. So I put some language in, and I've got an operational change out. Bear in mind, the system is aware. So this change is still governed. It's still compliant. It can still align to different labor rules, but it's done significantly faster. And this really matters because managers aren't just scheduling. You know, they're running operations. They've got multiple different priorities and things that they're working on throughout the course of the day, week, month. Reducing the clicks isn't just about improving that experience, but it's also reclaiming their decision making bandwidth. And and for the manager, this can reduce loads and loads of friction. For the business, this can protect that optimization. All the work that goes into that predictive scheduling in the first place is being protected because the manager is really clear around what their schedule looks like for the day week, and they can make those changes knowing that they're the right changes. And the bigger point here really is automation automation only really delivers value when the managers trust and use it. And when managers stop overriding out of things like uncertainty, optimization actually delivers the measurable results that you're looking to achieve. So what we'll do now is we'll go one level deeper than that, and we'll talk about the shift explanation assistant. So we've looked at visibility from a, divisional regional level. We've looked at interaction. So let's talk a little bit about trust and transparency. So automate auto auto generated schedules are only as strong as the confidence that managers have in them. And one of the friction points that we see in workforce management is the question, why did the system do that? And so this assistant is designed to answer that question. It will show us how a shift is created, how a shift is assigned, whether or not it was auto generated or manually added or copied. It will show things like edit history, but most importantly, it will explain the logic. So if a shift is generated, it quite simply tells you why. If an employee was automatically assigned, it fits the, skills, availability, preferences, and compliance. And this is really critical because when managers don't understand the system that the decisions that the system is making, they will override them. When they override them, as you'll know, you'll lose the optimization, and sometimes you'll lose that element of compliance and protection as well. So this assistant is designed to bring transparency into the decision making layer and not just why it not just what happened, but why it happened. So, again, let's have a look at that in our, UI. So we're back to our, scheduling screen. As we know, auto generated schedules, as I said before, are only as strong as the confidence that the managers have in them. And so why did the system do that is a common question, and we're gonna answer that question now. So let's take a single shift in our schedule. We'll click, and then we'll ask, the scheduling the shift explanation assistant to explain why that's taking place. So this assistant explains that it was auto generated to meet the forecasted demand. It's tied directly to the labor requirements and the scheduling rules. So now as a store site manager, I know this wasn't arbitrary. I know this was demand led and rule based. So let's take it a step further and understand why this particular employee was assigned. So we'll, again, converse with the shift explanation system, and it'll give us some information into why this particular person, Bauda, was assigned to this shift. So she fits the rules, skill set, availability, weekly, and the weekly limits. So this is structured structured logic that is going behind understanding from the store manager's perspective, why does that shift exist on that particular day, that particular time, for that particular work level, and why is that individual picked for that role, and what makes them fit that particular area. This isn't guesswork. This is surfacing also a compliance consideration. So you can see here down the bottom that Valder hasn't currently has some compliance issues with her minimum working hours and weekly shifts required, and that can be really, really important. In many environments, whether you're working in retail, in aviation, health care, compliance isn't something you'll know is optional. These are things that are contractual. They're regulated, and sometimes they're even reputational. And and so surfacing this proactively can reduce that risk silently oh, that reduce that silent risk, sorry, building up in the background. And when, essentially, when managers understand why a shift exists, they override less. And you'll know when they override less, we are holding that optimization. And so trust is what's protecting that automation at scale, essentially. So that's the shift explanation system. I want to move, slightly upstream now and talk about our forecast and explanation assistance. So we've looked at trust and scheduling decisions. Let's go one step upstream and look at the forecast itself because as you all know, everything we've seen so far is only as good as the signal that's driving that information. And so one of the common questions, across retail and other verticals is why does tomorrow look different to today? What's changed on the forecast? What things are influencing that number? Is that event included? Is that included? We don't have the full understanding of why the forecaster decided to make that decision. And so the forecasting explanation assistant is designed to answer some of those questions. So, essentially, what we're doing here is we're explaining daily forecast drivers. We're influencing understanding influencing factors like seasonality, events, local conditions, and, really importantly, any edit history. And this can matter because when managers don't trust the forecast, we know they compensate. They might overstaff. They might override things. We lose the optimization. All the things that we've talked about before are only as good as the forecast that's feeding it. And so they could then be building things like safety hours or or using expenditure of labor where they don't need to. But often, the reverse also is true when they understand forecast logic. They're more likely to align to it. They understand it makes sense. They're clear, and they understand the areas that it's matching against. And so this system doesn't just show the number. It explains the story behind the number, essentially. So, again, as with, the other parts of demonstration, let's talk a little bit about, forecast explanation assistant within the UI. So our forecast screen is available visibly to the store manager at all times anyway. They're able to see their forecasted demand. They're able to see the influence and factors that go into it. They're able to filter the individual volume drivers that are important to them so they can see that visually alongside looking at things like historic information and last year. We can also leverage this explanation, assistant again directly from the UI. So everything we've looked at today is only as good as the signal demand that's driving it. We know this, and this is where the real risk can sit. If the forecast is wrong or it's not trusted, everything downstream will start to drift. And and in every industry, the same behavior will show up. Managers don't trust the forecast. They'll do all the things that we talked about before, for overstaffing, understaffing, building safety into the schedule, not necessarily because they're being inefficient to a certain degree, but they're buying certainty with you by using labor cost because they want to make sure the customer and colleague experience is great. So the question then becomes, can I trust this number? The forecast and explanation gives that context. So it's showing us demand by category based on your individual volume drivers. It'll compare to historic patterns. It will surface some of that trend movement, and it will explain those influencing factors. So we're not just looking at output. We're understanding and looking at reasoning. The store manager and me will always want to understand what how that compares to last week. So let's have a look at that now. And, again, I'll converse with the forecast explanation assistant here. And, again, so I'm getting some information back, and it's essentially telling me there's minimal movement. So there's slight variation. There's overall stability from my forecast from last week to this week. And that can be so important because when managers understand that the movement is marginal and not dramatic, it reduces that emotional decision making, and then replace that instinct with evidence. And so this is where measurable impact can start to begin. So when we when defensive staffing behavior reduces, we see labor efficiency improve. And that's not necessarily because they're cutting hours, but it's because they're aligning the hours where it's most important for the business. And you can see this is what we see increase in sales performance and all those kind of metrics that we'll you'll see happening from day to day because the hours are aligned when this the business actually needs them. And so forecast explainability essentially does two things. One, it increases our trust, and two, it increases our adoption. Because, essentially, when managers understand why a number looks the way that it does, they're more likely to align to it instead of compensating for it or making panic split last minute decisions potentially. And this can be the difference between, or this is the difference, really, between predictive forecasting and explainable operational forecasting. That's not just what the number is, but why does that number exist and what's driving that and what things do I need to do to make sure that I'm in the right place at the right time with my team making the difference for my customers and colleagues essentially. And that's why that exists. So today, we've looked at four essential areas of Legion, but there's so many more examples that we can share in different follow-up meetings and have a further product demonstrations if that's something that would be of interest. But and we can share how we're supporting organizations to improve their workforce operation. But, essentially, what we've looked at today is AI isn't replacing managers. What it's doing is it's removing that uncertainty or reducing the uncertainty and reducing that friction. We're reducing unnecessary overrides. We're improving decision confidence at every single level of the organization. So if we look at the upper filter system, we were at a regional level, reducing that clarity and consistency. We looked at site level, reducing, improving that interaction and adoption. We were looking at shift level. We're looking at that transparency and governance. And then at forecast level, we were looking at the explainability and alignment. At scale, this can create something that's bigger than individual features. It can create consistent data driven decision making across that organization. And that essentially is where you will see your efficiency improve. You'll see, compliance strengthen, and most importantly, where the engagement of the store managers and colleagues in will improve because we're not just managers aren't just feeling support are feeling supported and not just second guessed. And so with that, I'll invite Melissa back on to, open up any q and a. Alright. Great. So we have one question here from Sarah. The labor demand, it's adjacent to the AI assistance, assistance, but the labor demand for the schedule is primarily based on the forecast of specific retail drivers. Could you talk a little bit more about how your tool uses AI and retail drivers, historical data to create a forecast? I think it'd be great to also talk about, you know, how these custom models are, you know, developed per location and demand driver. So I'll pass that one. to you. Yeah. Great question. Thank you. So yeah. So the the forecast is driven through a multitude of different factors. So we will take historic information, and that can be any level of historic information you have from sale, transactions, footfall, whatever volume drivers that are associated with calculating labor will be the first layer of where we understand your historic data. Our Legion AI system is able to utilize that, data essentially to build and understand trends and patterns that are happening throughout the organization. We'll then take things like promotions and all that information that influences the different levels of forecast that calculate demand. Then we also partner with a company called Predict HQ, which is a global, forecasting analysis business, which essentially can look at things like weather, planned events, and trends, and patterns that take place that sit outside of your organization level potentially, but will influence the things that happen within your business. And utilizing that data runs through our Legion AI forecasting optimization engine. That's a word for. But then it will also allow us to understand a forecast at a location based level to Malysa earlier points. So we're every single forecast that you saw in the forecasting side is down to a location level, which means the store manager interacts directly with the things that are impacting them most. Excellent. Thank you very much. And then we have another question. I think you've you sort of answered this, a little bit, but how have you seen the explainability piece help with adoption, or has it? And and I know from my experience, you know, the explainability piece, I've seen it really help with, like you were saying, giving more trust in the schedule or the forecast, so that managers really understood the the why behind, you know, the the schedule or the forecast. But can you talk a little bit more about that? Yeah. And you're you're absolutely right, Malysa. Yeah. We were we've talked about a little bit about how that can improve trust and guidance. I think a lot of people can be worried about what AI can do and whether it will replace I'm I'm a firm believer that AI will never replace your best stock manager, in any capacity whatsoever. And so what it's able to do is it's able to support and help, and that increases adoption because people will utilize that tool to understand and make better decisions within the moment. Legion ultimately has created that schedule from predictive, point of view anyway. And so we can leverage what's already existing in the system as the management teams, whether that be the upper field or or the the store manager to utilize and understand why it's made those decisions, number one, to help me, but then also help me improve. So you'll saw in one of those screens, we had some takeaways and recommended action. These are things that are able to speed up my time. I don't have to go through various different reports to understand how I can improve my business. The the tools that are available already within AI Legion AI's assistant system give me all the information I need to make good decisions for my team, or to make decisions really, really quickly. Some of these, tools like Shared Joy Assistant sit within the mobile device platform as well. And so I'm able to access that on the shop floor. All of a sudden, I can make changes really, really quickly, which mean that I can spend more time doing the things that are important to me because, ultimately, we don't want store managers in offices going through schedules and creating rotors. We want them in front of customers and colleagues and supporting. So that's where we've seen the biggest game. Excellent. And then the last question I would say we have is, are these assistants live in production, or are they still beta or limited availability? Yeah. Cool. Great question. So, yeah, all of these assistants that you've seen today and other assistants, there's a few we haven't gone into, like how we can support with communications, how we can support some of the configuration formulas in in the back end of the of the system in terms of setting up that labor calculation demand and things like that. All of those are live in production today, you can utilize all of these tools, yeah, from today, essentially. Well, not even from today before today. And then we have a follow-up of the first question. Think, you know, just to better explain that the third party component is really just for those external events. But is labor part is Legion partnering with a third party to create the forecast of the retail drivers? Follow-up question from your answer to my previous question. So question there, maybe just clarify how that plugs in and, you know, how we're really creating it from their data as Legion, but just plugging in some external. Sorry. Go ahead. Yeah. Absolutely. Yeah. So there's two separate elements. So Legion AI is generating the forecast based upon the historic data, in terms of your historic data and information, your promotions and events, the things that you're clear and understand and know about. And then we you partner with a company called Predict HQ, which do everything that sits externally to a certain degree, things like weather and events, and that's that's that's connected directly, into Legion. So that all that information is sat within Legion. It's forecasted in that way. Yeah. And we see people that can take in things like school calendars and, you know, they wanna pull that in, because that has an impact on their And so that's that's another angle that we can pull in some of those external drivers. And, you know, we can we can both forecast internal drivers such as sales, promotions, as well as those external drivers. Exactly right. Yeah. Yep. Did we answer your question, Sarah? I hope. If not, we're happy to follow-up with you after this call and share some additional information and content as well as, you know, provide a a a demo as needed. Okay, Joe's. I'm gonna take one more quick pass. I don't see any other questions. So thank you so much. Great job. Great information. We encourage everybody to reach out. If you want additional information. Please do visit the docs tab if you would like some content as a takeaway that further explains some of these AI assistance that Joe's gone through today. And, we look forward to talking to you again. Thanks, everyone, and thanks again, Joe. Nice to meet you. Thanks.