Sales forecast automation for Supervision

How to create outlook signature

a very good morning everyone uh and a warm welcome to this webinar uh automating settings forecast and stock replenishment with AI my name is Jasper hoyland and I'm head of sales of vagina in the Netherlands Denmark Sweden and Norway um we all see it uh it's been a part of our daily news for a long time we experience it in our daily lives as well as in our business we work in we are in times when unpredictable demands in our markets increase the pressure on our supply chain and it has a tendency to blur our ability to see across multiple data points and sources their limitation of human capabilities disabled us from considering patterns and data points and engaging with external data sources and that could be time consuming and challenging but there is a helping tool that can provide insight and give us a helping hand in foreseeing patterns in our supply chain and our demand for this webinar we have invited Mila listers to introduce us to modern tools that help you to foresee your supply chain demands Milan it's a pleasure to have you here good morning good morning yes sir good morning everyone thanks a lot for this nice intro and for uh handing over to me to to deliver this uh this content so um first of all thanks everyone for joining my name is Milan as Jesper mentioned um and just in one sentence I spent uh six to seven years as a head of our data and AI Department um in adriatics region and now my role is to lead a pre-sales again for the data Nai um so pre-sales activities on our on our group level nevertheless um just a few quick intro uh that I'd like you to have in mind so please feel free to drop down quest questions you have in a q a section if we don't manage to address them during the live for any reason uh we'll have we'll absolutely do that as a follow-up so don't worry about it and of course there's a recording being in place which is going to be sent to you uh more or less automatically so please feel free to you know re-share it with your call police or whatever is needed to you know boost the discussion within your organization so moving on to the agenda today so after a quick round of of an inch for our data and AI team we're going to discuss which problems are we solving with this solution you know um how did we come up to to to approach this solution um second thing that we're going to discuss is how do we solve it obviously um and we're going to discuss a bit about solution overview from module perspective and from end user perspective because it's really important um and basically two last bullets are more or less a project oriented so to say and that is we're going to discuss how to calculate business benefits so we have a lock that may be comparing to some other uh um business software implementations here we can really focus on calculating Financial impacts and pure business benefits that this solution can get your organization um and the last question we're going to try to answer here is uh you know how can you start this journey and what would be the first preferable steps so just a quick round of intro for for the team that I'm coming from so um I think that a lot of you or most of you known uh bitterna or either as a whole or as a group or um the nordics team however inside of that team uh there are between 70 and 80 people who are focused on delivering data and AI projects um and we do basically everything from you know initial assessments planning together with you up to the execution and support you know lifetime support uh activities and inside of that group there's a pretty pretty dedicated group of seven people who are engaged in delivering I like to call them applied AI projects you know why applied because we do not engage into some kind of um General Consulting uh or um you know introducing um AI or machine learning as a technology so what we did in the past few years we found couple of really really focused uh use cases in various Industries trees and for now we stick to them you know so the one that we're going to discuss mostly today is demand forecasting and supply chain planning and then if we are not talking about few retail organization but uh extending it to a manufacturing organization um planning and Manufacturing process optimization comes hand in hand with it so we can take a look at this as as let's say one uh solution but just depending on the industry that we're discussing uh let's say slightly more um uh next to it is a topic of iot predictive maintenance so uh you know eating and understanding sensor data from various manufacturing organizations and and many production plots and trying to figure out uh how how you know how to keep the maintenance in a predictive manner um so let's say we can we can put all of these things in in one uh hat called supply chain planning somehow and on the other hand side the the second focused issue we have here is um customer Behavior modeling and that's the topic which is basically in the core of any personalization project that you might have right so if we're talking about retail then it's in in a focus of a loyalty project right it if we're talking about um I don't know Banks or insurances it's a personalized offering if we're talking about Hospitality chains um it's some kind of a personalized offering in terms of you know a recommended stay next best offer and things like that um so basically understanding the behavior of your customer not just the demographic capabilities so this is what this team is pretty pretty focused on and today you're going to see you know the the progress and the response from a market we have from the projects uh on the topic of demand forecasting and supply chain planning so to set the base for this discussion uh you know we're going to answer the question which problems are we trying to solve so what we have here on this map is really um let's say generic approach um from initial financial planning to you know getting Goods on stock to replenishing goods from sent Warehouse all to a markdown execution um so you know various challenges um From perspective of managing inventory and what we did here you know we mapped this process I use the example of let's say that we're producing or selling a pink chair so I just use this simple example uh to see you know which challenges do we have um from managing inventory perspective down the road so in the beginning obviously financial planning we're trying to answer a question you know how much revenue do we want to get from our uh let's say selling this uh this pink chair right um and and then after we have the stop line plans we obviously have to answer the question how is our assortment going to look like so which portion of this Revenue should come up you know to Furniture to chairs and then the pink chair so to say then inside of that there's a skiu planning so inside of the furniture part right we had to answer the question um how much portion of the sales should go to the chairs itself to the pictures to the blue chairs Etc um and then again when when those goods either come on our stock or we produce them as a manufacturing organization right we can go in two steps here we obviously have to make a decision how we are going to distribute this initial allocation across our stores right which stores should have better sales of ping chair and which which not and then this this was so called let's call it a Goods pre-arrival so-called planning phase and then in a Goods post arrival or so-called execution phase we have also various other challenges you know where we could get some support from um from from machine learning solution uh first one is being replenished and from central warehouse uh so you know do we have to replenish our pink chair is the sales going good uh how much should we replenish um in future uh second thing for example really popular in fashion industry is store transfers so basically when we're done when we're let's say out of stock or where this season is over we still have to answer the question you know are there uh for example pink uh shoes in this case um left in one store to transfer them to the other store where they could be you know sold more easily and then in the end and there's a question of markdown execution right so if the sales is not going well can we you know lower the discount uh lower the price introduce the discount in order to sell it more easily Etc um and as you can hopefully see by this lighter pink color I mapped also two phases uh which are let's say uh strictly focused to manufacturing organization so first of all you know um ordering uh Goods in a manufacturing organization actually means ordering raw materials or planning how to produce a semi products in order to to meet you know Financial plans of selling the final product and there's also another component you know which we got an amazing feedback from a market so let's say that this whole plan is already on a paper and it's executed and what we have is actually a list of you know work orders so how are we going to produce uh these pink shares right um but there's still a component that we can optimize here and that's the schedule and the sequence of those work orders right you know order to maximize the output in a given time frame and this is a topic where we really have an amazing results up until now from a market now when we have run through these um you know main managing inventory challenges we're going to try to answer or we are going to answer the question how are we solving it you know at beturna so obviously it all starts with the first module of data input so core of our solution obviously is data input now the thing that differs you know Erp processes or statistical forecasts Etc to machine learning one of the things of course is integrating also third-party data so on top of your historical sales historical stock out historical production historical purchase data what we integrate is also historical and future weather forecasts holiday seasons you know Global shipment trackings Telco data if available in-store heat Maps data if available etc etc and then the first module so the chord basis of this whole process is you know utilizing demand forecasting algorithms now please bear in mind uh this has more or less nothing to do with your financial plans that you want to achieve right so these are the algorithms that are forecasting the demand from a market So based on a sellout data so to say then when we have these uh forecasts which inherently have seasonality analytics in it so if your sales data or consumption data has uh you know anything remotely to uh seasonality patterns um these algorithms will understand it and forecast it ingly and then when we have this information the next thing we do is we recommend orders you know either orders to our vendors or we recommend you know orders quantities that we need to produce in a manufacturing organization uh module of multi-location planning obviously means that we don't do this for one central warehouse or One retail store but for multiple right and bill of materials roll down means that we also integrate bill of materials information from an Erp and whilst when we know the demand from a market for a final product you know we can easily roll down this information to understand the needs and the capacity for the raw materials and the semi products right and then put them in optimizing production module before coming to that a few more really important modules so cargo optimization you know placing an order um either we're ordering a track uh truck or a container across the seas right uh that volume can be optimized now um in order to not keep the container half full or half empty uh we put this data through the cargo optimization modules you know and then try to optimize The Filling of this container in order to reduce the transportation cost then one this order is placed right what we are expecting is let's say expected uh lead time of this delivery but those expected lead times are usually just there on a paper okay so what we do is we integrate with the global shipment tracking apis so if you have an ID or you know of your shipment we can basically understand when this shipment is going to arrive to your warehouse and then plan ingly either or your goods from a different vendor you know or um you know replan ingly or something like that as already mentioned in the Fashion retail store transfers is something which is uh super important so we also have algorithms that recognize you know if the sales of an item is going better in one store uh or worse in the other store and then suggest suggest the transfers between them um and then in the end we come up to a promotion planning so this is really really important component where we feed the system with historical promotional data so the system can understand what was the success of this historical promotions and then we also feed a system with upcoming so future uh um promotional data you know and and this recommendation system is going to take these assumptions and recommend as a quantity to order ingly and the last module um actually second to last module is module of kpis and Reporting now um introducing solution like this is obviously also a change management issue so what it happens in real life is that you uh um tell to a person who's been doing this for I don't know five or ten years uh using Excel or whatever way now you're telling that person hey there's a system which you know can do it somehow better which is just partially true you know because we cannot do it better without the help of uh of a person who has the domain knowledge and in order to bridge this Gap what we do is we deliver them kpis and Reporting in the form of you know self-service dashboard in power bi or qlik sense or Tableau or whatever you're using and the purpose of this is to give more confidence to um a person doing the purchase so that means that you know when this person takes a look at the dashboard he or she can understand the historical sales the forecast simulation of a stock level uh upcoming six or seven or recommended orders right so it can give this person much more confidence to accept the recommendation from a system and the last one that I already mentioned is you know optimizing this sequence of a production in a manufacturing organizations you know so once when we come up to a level of work order we can also optimize these work orders taking various constraints into into consideration you know um capacity labor capacity um you know nominal needs uh nominal work time needed actual work time needed time needed for an item to switch between the lines time needed to switch a format on the same line etc etc so we put all of these information into a one box and provide you with a Optimum sequence of the work orders foreign now when we put this overview into something that we call data pipeline so moving from left to right um I'm sure you more or less understand it it all starts with data input then you know we deliver forecast propose recommended quantities and then integrate so what I want to do here is actually uh just just focus on on two bullets that I didn't mention previously so first one is on top of the whole system which is live and up and running constantly we have something that we call stock monitor so that means that you don't have to manually track if you know there's a risk of going out of stock or if there's a risk of you know getting Overstock there's a constant stock monitor which detects and quantifies you know alerts you detects over stocks stock outs calculating missing Revenue potential if you know we don't manage to produce something by the time detects outliers if you know some sales was an outlier or standard sales Etc and then gives you a proposal to order goods earlier or do something else right so this is the first thing that I want to point out second thing that I wanted to point out is the output and now we're slowly coming to the I like to call these you know market differentiators and I think it's important for us to discuss those so um first one starting from a right hand side is and and it's really correlated to this output so what we're discussing here today that's not another system to maintain right you we we don't give you a Management console you don't have to input the expected service level you don't have to input the promotional data in there if you already have it you know so what we do you can imagine it uh at this phase as a large you know black box and which deliver you we call it intelligence as a service so that means that there are just two integration points of the system first one is direct let's say overnight integration with your Erp okay using apis uh you're gonna see it in the next slides and second one is the integration to a self-service bi application so that means you know understanding sales forecast open orders term contracts stock simulations Etc okay uh second benefit that I wanted to discuss is we really cover process end to end so that means that we don't leave this last mile of optimizing Cargo in a truck alone you know we also take this logistical data pallets tracks volumes pieces Etc and we provide you with the most Optimum feeling of a truck or a shipment we use bi for explanatory that means that you know this front end is not some rigid table or a simple line it's really um global leaders in terms of business intelligence and analytics which really allows this user you know to understand this recommendation and accept it more easily and the last thing which I also think is really important we leave the data where it is right so we we just integrate the data if it's uh sales data it should be in your transaction system if it's logistical then the same thing if it's promotion same thing if it's global weather we take it from the API you know we don't ask you to to provide it so we really try to leverage all the existing data without actually asking you to input something uh something new in it um and now this same pipeline how I call it uh when we paint the same picture from perspective of manufacturing organization it's pretty pretty similar there's just another additional component called Product production optimization and this is the one that I mentioned and that is you know just scheduling the work orders per each line in order to maximize the output in a given time frame and then again back integration of these uh work orders into an Erp in order to actually trigger production from a process perspective right what Erp is is used for okay so what we did up until now we we went through the uh let's say module overview of the solution and now what we're going to do in the next few steps is we're going to see how the solution looks from an end user perspective you know I mentioned here that it's not another system to maintain we don't have a Management console so fair question would be okay how does it look like how does uh even how does your user even use this solution right so here is how it looks like uh what we have on a uh here is the use case of you know purchaser starting the order process and for this example uh we're using a 365 business Central um just for an example right so um let's say a person comes to a work in the morning and opens the requisition worksheet enters the uh vendor that he or she is responsible for um she in this example so select vendor and sees the recommended quantities that came by overnight integration to a system okay so she can revise recommended quantities and let's say that she's concerned over this suggestion of zero pieces for a mango order so the next thing she can do is open a self-service power bi app and have an insight into this forecast and historical sales so basically item by item she can understand what was the historical sales how does the forecasted sales look like she can also understand what was the historical stock level and how does the forecasted stock level look like so this means if a forecast happens as we assumed and if uh you order goods as we suggested this is the stock simulation um in your in in a future an item by item she can also understand some key information so the name recommended quantity forecast Etc um and um you know just feel more confident into ordering zero or changing the quantity one interesting point that I would like to uh show here is you see this color code here so what this actually means if item is read sorry green coded that means that actually a person can have confidence and that our forecast is precise accurate so to say you know if there's a yellow or red mark here that means that person should uh double check the recommendation because you know either we don't have enough information or it's a first filling or it's a slow moving item or something like that uh just to mention this ratio here with our customers who are already in production goes up up until 95 97 on some categories of the items that they just you know accept the recommendation and place the order and now let's say um she decides to um change the zero to uh 174 okay and um confirm the quantity to order and just place this order the same way as the order would have been placed um using just a BC you know standard functionality so it's just that information given in there was much more precise and as a result of a pretty complex machine learning algorithm running in the height um and this is the process how it looks like end to end the last components that are left here is you know then actually just tracking the actual sales comparing to what you ordered uh Goods arrive at some point and then somewhere down the road there is a similar process that's going to be triggered you know either uh another order from a vendor either replenishment from a central warehouse either store transfer either production order right if it's a fashion organization then maybe initial allocation Etc so all of these processes depending on a use case that we're solving from this first you remember it flow of goods right challenges so depending on which point are we solving there we can trigger different processes integrating at different points in Erp but more or less the process is going to look pretty similar to what we saw now so really simple um and now this is now just zoomed up uh screen from a BC requisition worksheet that you see that it's completely standard functionality we just integrated information to be much more precise you know and then the links to something similar like this where you can really see forecast simulation uh all the you know Master data information drill filter relatively easily uh etc etc um so basically this is this is what I uh what I wanted to share with you from the from the product perspective um just a final note on my side what what we showed you here now is as a process and I'm repeating it you know it's not another system to maintain so why did I even mention it you know three to four times so from our experience in the market uh probably the key reason why do AI projects or machine learning projects fail that's because you know they tend to change the process how a person is doing it now completely right we're not trying to do that we're trying to keep the process as similar as it is and just provide the end user with a more quality information and if you think about it in your in in real life the applications of AI that have I would say the best success in your life are the ones where you actually actually have no idea that it's some kind of AI running in behind you know sorting emails in your Gmail recommending songs in a Deezer or Spotify recommending uh what to buy at the Amazon uh receiving I don't know a booking proposal from your favorite Hotel chain etc etc you know so those are all the things which are not changing your experience and that's why we decided also not to change the experience of a person who's actually doing the purchase activity right and now we're slowly moving uh to the last uh two bullets um first one we're going to address is to how to calculate the business benefits and for this purpose I prepared uh two uh points of view on the same uh problem first one is looking from a pre-sales example so what can we calculate already in the pre-sales activities and the second one is from you know project perspective in terms of um actually sharing a case study with you right so from pre-sales example um please take a look uh to begin with on the right hand side so what we have here is the kpis that we are basically able to simulate already in pre-sales phase and typical kpis that we you know try to optimize is for example lower stock out situation right and then we try to estimate how much of a loss sales loss Revenue would this uh stock out situations have an impact to uh second one is less that cash so that means just you know if you were piling up the stock in some categories we're probably going to be able to keep it much much lower and then keep this free cash to your CFO or your organization to invest it in some some other place and then there is of course decreased cost of stock handling um here we didn't even mention some soft benefits as you know lowering uh time that people are spending to this process better negotiating power with your vendors etc etc so why am I sharing this with you is just to give you an information that basically all of our uh you know projects up until now were based on pretty pretty uh clear cost benefit right and um I don't want to over promise here but for now our experience was that return on on investment was ranging somewhere in the area of one year you know 10 months year and a half but somewhere in this in this area of course depending on how much problems through the stock do you currently have and the second thing that I wanted to share with you is this project example this information is coming from a Pharma Distribution Company case study um uh this team is with us in production for uh now I think more than three years they kick off even before this first wave of uh covet lockdown so couple of interesting things here so as I already mentioned so up to 97 of the estimate future demand is correct in their case so that means they accept the recommendations for 97 of the items in some categories uh this is what they stated that they are now spending 50 percent less time and a manual effort on data Gathering preparation Etc so they just you know focus on actually uh placing the right order and this is pretty amazing so you know although we managed to lower their stock on some categories 25 to 65 percent okay so that means lower stock in the same time we actually managed to reduce number of out-of-stock similar situations you know so number of other stock dates and we did it for more than 90 so that that exactly means that a Pharma retailer in those cases you know was not ordering their goods but the goods from the other vendors right and now this situation is sold um and twice as faster inventory turnover but um I don't want to drill too deep into this one uh so let's focus on these kpis here that we uh that we mentioned them uh for now so this is from the business benefits perspective and the last thing that that I wanted to share with you um to to to leave enough room for uh hopefully discussion is you know how to start now obviously um there is a there is really a simple question uh which should validate like do you have do you even have stock issues and that's not if you already have an Erp if you already have bi solution if you already have a planning Solution that's really not important what's important is you know are you experiencing stock issues in real life so are you experiencing stock out situations are you experiencing Overstock situations are there I don't know 5 or 10 people involved in this purchase process is it too much manual effort behind it and if the answer is yes to any of these three to four questions you know then there's probably room for the Improvement and the question is how to start so they're basically two you know action points that we usually engage with our customer customers first one is we like to call it one day POC and that's a free activity so what's behind it is you know that we ask you that you deliver um you know historical sales data and purchase data for let's say 20 items and we just deliver a forecast for those items now this activity is free for you and the reason why it's why it's free is because you know it's relatively hard to validate those results we can you know take a look at the forecast uh wait for the future to happen and then validate it right so that's why we usually do uh something else um and that is another proof of concept but what we like to call proof of value or in the other words feasibility study or benefit analysis or whatever and this use case looks in a way that we ask you for the data on 300 400 or 500 items right for the last two years and then what we do we train our models on a first year and a half and then do the simulation on let's say last six months and then what we can do is actually do the comparison you know in those six months how did the actual cash outflow look like how would it look like in our simulation how did the stock out look like in real life how would it look like in our simulation you know and in general what was the stock level and how would this stock level look like in our simulation so basically what we do after this exercise is you have you know really really precise potential Financial impact which is as close as possible to you know tomorrow's data and to what's actually waiting for you um the market right this is obviously paid activity because there's a lot of added value behind it and basically you know if financials after this exercise say you have returned on investment in 9 10 12 or 16 months you know then customers more or less feel confident into engaging into a process into a process like this so basically um I think I would uh I I think I I put all of this in roughly 30 minutes which was my plan and I would stop here um so um now we we move on to q a sections and so I'm still in a full screen mode uh yes per I would kindly ask you uh to you know let me know if we have um any any questions uh in a q a section uh yes I should address yeah thank you very much uh for for this great uh introduction Milan um we have some questions here uh the first one I suppose this considers lead time and or Price agreements for vendor selection when suggesting plan purchase orders absolutely so um yeah I I'm I I must have forgot something you know in these 30 minutes so basically um what I didn't mention is you know these forecasting algorithms and these recommendation um recommendations take into account uh many constraints um you know that are available starting from you know actually time expected lead time uh Warehouse manipulation time um annual bonuses for example you know if you need to boost sales promotional information as I mentioned uh holiday days Etc so basically all of the constraints which can be either obvious or something which is you know domain specific which differentiate differentiates you in the market we integrate all of those constraints so you can think of it as a box where we just pull in all of the constraints and and get the best recommendation output out there so absolutely yes thank you Milan um we have another question uh we have an older version of novision um and I I guess this question applies for all uh older Erp versions uh so the question is can we use this solution for for an older Erp vision so um absolutely I mean the the core of this solution is you know uh something that we deploy either um on your server on-prem or ideally in Azure cloud or in Amazon cloud or whatever you know wherever on-prem on your service and then as I mentioned we just integrate two components we integrate this information to power bi and we integrate your um this information to your Erp you know and all until we can integrate somehow either via apis or putting this data into SQL database which is behind Erp you know we are we are good to go and we had either projects or pocs with um nav BC ethanol you know ax sap um some local uh Erp systems etc etc so um it's really not dependent on the Erp you have we use your Erp to get the data from there and to get the data back in there or you obviously those are two technical challenges but you know that's not something that that we uh that we don't manage to solve okay thank you um Can this also be used to forecast the need of Workforce as well so I have resources in a in a production plan yeah yeah so um that's a good one also uh absolutely yes so basically you can think of it is this way any any sequence of information you know that we have on your timeline uh this can be shifts by the day you know uh work Workforce uh by the day it can be forecasted so um when we are delivering this uh work order schedule um it's integrating again many constraints and one of those constraints again being uh you know Workforce plan or the nominal Workforce which is available um second example is we did a project within um retail section of gas stations here in Adriatic region and they had a challenge with a large waste of fresh goods so for example orange juices croissants sandwiches Etc and what we did we integrated sales data we integrated traffic data and we gave them forecast two things first one is you know recommended quantities to order or produce number of sandwiches juices Etc and the second one is suggestion of a Workforce by shifts you know so they could also optimize that number so the answer is yes all right thank you Milan um another question here uh we have multiple stores in our brand uh can this be used to propose items per store and assortment so basically what what type of projects we will have per store yeah so um uh similar uh slide as I've showed you in the beginning we have this same slide similar slide sorry um for the for the for the let's say fashion industry right for example and if we engage in the discussion with you we are probably going to come up with a similar slide which is a bit tailor-made for your industry uh and for example in fashion industry um I mean it doesn't have to be fashion industry but I'm just taking it as an example so we do also that right we we plan the um the assortment that needs a split by categories which which you know um accumulate to a full Financial amount and then we plan a full amount number of skus we need to order and then for example in the fashion industry this is one of key challenges maybe in different retail stores also and that's the initial allocation you know so from the first annual planning up until the execution of you know initial allocation of the goods by store it can be two to nine months you know and a lot of things can change on a market so deciding on a split of those goods this is something that we also do you know and you can think of it in the same way that I showed you so the end result you would get is the recommendation for the allocation right we utilize also similar items detection in there image recognition so they're um quite a few techniques uh that we do in order to you know forecast the sales um of the item which is new and which you know does not have any historical sales already there um thank you Milan um uh and another question uh anonymously here uh so when it comes to external databases that could influence uh the AI prediction of uh for instance stock level or other things is there a limitation to uh external data sources that could influence these calculations so there's absolutely no limitations in terms of you know how much and which data can we take but it's the question if the data is going to be significant or not right luckily we do not have to decide that we can pull this data in and then the algorithms itself are going to let's say tell us if some external data is relevant enough how to use it for the future forecasting just an example we did a POC with a large bank and they wanted to predict you know which customers are going to repay their credits earlier uh for the bank that obviously means a lost interest right so what we did in this POC we put in there roughly 60 so 6-0 um various macroeconomical kpis you know and then we run the simulations and forecasted in the same way that I mentioned you know train the model assess the model and we understood that there are basically five or six that have some kind of a significant correlation with it you know and the rest of them uh don't and then we said okay we're going to focus on these five and utilize them down the road you know on our on our forecast okay thank you very much Milan um I think that's the questions that we have received up till now um well if that's the case then I would uh thank you all a lot from my side um and um give a closing closing words and closing remarks uh remarks esper to you yeah thank you so much for for this great introduction Milan um I just added here our contact details uh you are here on on the left hand side Milan uh but your email and your uh that's of there's a QR code for your for your LinkedIn uh so feel free to to connect uh we have Roland levers that's responsible for the Netherlands we have Patrick yanchgold responsible for Sweden uh with this we also have Frederick stay loan ahead of a customer success in in in between analytics and uh and also me here um head of sales of of of of Denmark and Norway um so please feel free to to reach out and engage with it with our uh with us in in discussion uh what you could benefit from from Solutions one final note thank you so much for joining we already have the next webinar planned it will take place on November 9th and it's uh it's a it's a similar topic about how we could use it to uh to to create a better a better day at work uh so this topic is around robotic process automation uh how how robots can help us automate uh automate our workflows um so by that closing remark uh thank you all for joining and have a have a great day thank you so much