Supply constraints, inflation, volatile currencies, frequent competitor price adjustments, bundling and unbundling of products: pricing teams are overwhelmed with new data streams and volumes of data that are necessary to make the right pricing decisions to drive profitable growth. Even the most robust segmentations fall victim to data sparsity and relevancy. Join this session to learn how neural networks revolutionize profit optimization and empower pricing professionals to make the right adjustments in real-time and across all of sales channels to not only adapt to frequent market changes but leverage these dynamics as competitive advantages.
Eldho Kuriakose: Hi. I'm Eldho Kuriakose, Director of Product Management at PROS....
Kaavya Muralidhar: I'm Kaavya Muralidhar and I'm a Product Manager of Price Optimization at PROS.
Eldho Kuriakose: Thank you. In this session, what Kaavya and I want to do is really share with you what it means to bring on AI. And specifically, Gen IV AI as a partner in price optimization.
We'll deep dive into the capabilities of the AI, and how your team can collaborate with it to outperform.
And outperform really is the operative word here. Right? Because, you know, according to our friends at Gartner, IDC, etc., 85% of all companies in B2B are going to employ some kind of AI come 2024. So just by putting AI in, you're going to be amongst the 85%.
So how are you going to outperform? Right? If eighty five percent are doing it, what's the key to outperformance?
And that's what we're going to dig into with you.
Before we talk about price optimization, let's look at a slightly different problem.
We're all familiar with chess and what's been happening in the world of chess over the last 50 years. You know in the 50’s, even programmers started to try to build an AI that can defeat a human being. And as we all know, by 1997, they got to the point that IBM's deep blue defeated Gary Kasparov.
Right? So since then, the AI has just continued to improve, so maybe at that point the takeaway was that, well we're done. The AI's figured this out is just getting better and better we're just going to sit there and watch AI play against AI. There's nothing for humans to do.
Except that would have been a wrong conclusion because Gary Kasparov wondered, what if humans and computers actually played not against each other, but together against another machine.
So in 2005, He had a chance to kind of understand and play that hypothesis out. Right? So, in a virtual chest tournament, he had combinations of human players, and machine players, and as expected, the machines always defeated the humans. No questions.
But then you put a human alongside a machine and play another machine and guess who won?
The team with the human and the machine constantly won against the machine by itself. This is a pretty incredible finding, right? Because you have billions and trillions of transistors and perfect execution on that on that AI model and yet you put a feeble human being along with another one and boom, you're beating that machine.
So there's something we still have that the machines haven't figured out. And in fact, when we collaborate with the machine, there's something that is just undefeatable.
And so here now starts to get into the idea of how do you outperform with AI?
And in fact, when we as product managers at PROS think about how to bring this amazing science in a productized form, this is exactly what we're trying to solve. And today, Kaavya is going to share with you a lot of what we do every day to make AI a wonderful partner for every human being that's out there, especially on your pricers.
But why are humans and AI such great partners? Well, the human modalities of thinking, are a big part of that. How many have read this book by Daniel Kahneman?
Great. If you haven't read it, go to your local library or a Kindle and definitely give it a read. I'd say its life changing because now I can have a great conversation with my wife the way that I couldn't before. Because she, actually things a lot more faster than I do.
And sometimes, she runs circles around me. And I'm sitting there slowly analyzing stuff and I can't figure it out. So, once we kind of recognize we have different modes of thinking, we're able to communicate much better with each other. Right?
Thinking fast is what humans do and especially my wife - she does that better than me. We satisfice rather than optimize. And we try maybe not to fine tune the decimal point but maybe protect the downside. Right?
So we're pretty good at those kinds of things. We had to be. We've been here a million years and we had to do that to get to this point. Right?
We find patterns and navigate even when things are not that obvious.
We use personal experiences. Sometimes you call it prejudice, but a lot of times it is what saves you. Right? So there's some values still there. And most importantly, we excel at communicating and being creative.
Now generative may have some scratches on that claim of ours, but creativity is still very much a seed that we hold, and we build relationships.
And relationships are just things you can't fake. Right?
So those are the strengths that we have as humans in the way we think. Now as technology, of course, it has its own strengths. They perform consistently. They never make mistakes. They don't get tired. It's able to remember and process complex instructions like these complex price methods that maybe many of you have seen.
It's unaffected by anecdotes and emotional factors.
And it doesn't struggle very much to extrapolate just from a few data points. In some of the cases like in the generative scenarios that Dr. Michael Wu showed, we saw that AI was able to fill in the pieces very nicely. We still struggle with that in truly tabular data as he was showing. But we're going to get to a point where now human and machine can collaborate to make that part happen as well.
So really what is the formula for outperformance then? It's to deliver insights at scale. And when I say insights, it means things that you and I can understand.
Deliver that at scale. Optimize at scale, such that we don't have to redo those calculations each time. And bring in a place where the creativity and curiosity of humans to play a part. And when you create that braid, that's how you get outperformance. I think we answered that first question. It's not just to go put in AI, but it's how to create this brain that generates outperformance.
And this really comes down to the fact that, you know, AI will not replace you, but a team using AI definitely will. And this is where I want to hand it off to Kaavya, where she is going to start talking about exactly Oh, sorry. I got one more piece here.
Before we get into that, want to take a step back and look at the journey that we have been on at PROS.
About 15 years ago, and if Royce is in the room, he can correct me if that number is right or wrong. But at 15 years ago, we came up with our first generation of AI.
And I'm giving these rough Harvey ball grades on three dimensions.
1. The sophistication of the science
2. The amount of transparency we had out of the box for that science
3. The amount of self-service that we've enabled in that science
So Gen I was a relatively simple science. It was what we call a cartesian segmentation. So, think about cartesians like you cut a cube x y z, cut it up into little pieces and see how many observations you have in each piece and make some recommendations off of that.
Now at that time, we didn't spend a lot of time on transparency or self-service. We had, you know, big PS implementations.
Gen II advanced the science a good deal, where instead of just cutting up everything into equal little boxes, we were able to detect various patterns and keep as much data as possible to get a good information or good recommendation from the right sets. The only thing was that the technology at the time made that rather brittle and rigid. Because after the implementation, it was struggling to learn as the data patterns changed.
But there also we didn't have the extent of transparency and self-service that would allow anyone to just start implementing.
Gen III was a huge step up. We tweaked the science to the just the right amount of sophistication such that we could build really good transparency on top of it. So that's at the end-to-end system, whether that's the implementation, the science team, and the salesperson, the end-to-end was actually optimized because now the salesperson could understand why that recommendation was what it was. So that transparency was a big jump. In our Gen III.
We also spent time ensuring that it was as self-service as possible. So, you could do it through kind of like you're filling your tax form. You could configure your guidance run.
Gen IV is where we are taking this to the next level. We're now the science, we don't have to sacrifice in the sophistication of the science because we've really invested in the transparency.
Usually, what happens is that when you have really sophisticated science, it's hard to explain and if it's easy to explain, maybe the science is not so sophisticated. We've actually reached a good pinnacle on both with Gen IV and that's what Kaavya is going to show you. But equally as much, we want to make it as self-service as possible.
So we ask you for some attributes and then we've built in the hyperparameter tuning so that it does that tuning for you without you having to do it yourselves.
And this is super important because, you know, we want to put this in your hands and it's not the first implementation that matters, it's actually what happens 6 months later, 9 months later, 12 months later. Right? You need to be in the driver's seat. Right? And that's exactly what we're shooting for. So Kaavya is going to show things that we've done already and the things that we're implementing and investing in the second half of the year to make all these green boxes the reality.
And you know, our efforts on this haven't gone unnoticed.
When we talk to Gartner, IDC, they've really recognized that our AI is top notch. We can actually stand against any of our competition and say that we're doing it correctly, we're doing more advanced and as responsive in a way that is well ahead of what is in the market today. And we're making that all of that with transparency.
So now I'm going to hand it off to Kaavya to really get into the details of those.
Kaavya Muralidhar: Thank you, Eldho.
So as Eldho showed you, PROS played a leadership role in AI for several decades.
Some of the Gartner and IDC quotes that you saw on the earlier slide were actually also referencing our previous generations.
But with our fourth generation of AI, we've really surpassed our standards for what AI looks like, what it does, and how to meet each of those criteria that Eldho shared.
So, to start talking about PROS fourth generation of AI, I want to start by talking about what the fourth generation of AI is not. Which is a segmentation-based approach. And this is important to discuss because PROS has used segmentation for the last couple of decades. Some of you might be on our segmentation-based products, or those of you who are using other tools, even those of you who are using Excel, may be using a rudimentary form of segmentation. So, this has been a very popular form of understanding the industry for a very good reason. The goal of segmentation is to identify like groups of people with the assumption that they have similar behavior to each other. So, let's play out a segmentation happy approach and see when that happens and when that doesn't quite happen.
Let's say that we're talking about two people. We don't even have to talk about pricing. Let's just talk about a different example such as targeted ads. So, you have person one who is a male born in 1948, grew up in England, married a second time, you know, how many children he has, you know, where he likes to spend his holidays and that he likes dogs, that's quite a lot of details.
And now we found person two, who has exactly all of these same characteristics, so he's also a male born in 1948 who grew up in England, has the same number of children as successful in business, spends their winter holidays in the alps.
The assumption that we make with segmentation is if these are the attributes we've chosen, this looks pretty comprehensive, so these two people are likely to be very similar to each other and respond in very similar ways. But we see that that's not always the case. In this case, person one is King Charles, and person two is Ozzy Osborne, a heavy metal singer.
So we can see that although they have an incredibly surprising number of attributes in common, I had to cross check some of these facts just to make sure. They are very, very different people with very different characteristics, and they are probably not going to respond the same way to almost anything regardless of what it is.
So, what went wrong here is that we chose a certain set of attributes. We assume that that set of attributes, such as gender and age, and where you like to spend your holidays, is indicative of behavior in the same way in every situation. And maybe for another group of two people, it might be but for these two group of people, this wasn't the right set of attributes, and we should have looked more closely at their unique case to understand what actually matters and who are the right peers to look at for King Charles, or for Ozzy Osborne in order to truly determine their behavior?
If I bring this example to pricing, now we start seeing an analogy.
So, if you consider an electronics parts outlet selling a product such as LED string lights. You have a variety of things that characterize the product, such as its color changing bulbs or solar powered battery. You also have a variety of things characterizing the customer that they're selling to. In this case, if it's men's home improvement, we have their location in an urban geography in California. And finally, he have a set of things that characterize the sale. Is this happening during the winter holidays, during a time when competitors are offering 8% discounts.
So, when we try and solve this kind of problem through a segmentation-based approach, here's where we get.
You might look at your entire transaction dataset, and given that you've chosen, for example, product as an attribute, you split up your dataset by product. So now you're looking at the 12% subset that is LED string lights. But you've also decided that the fact that their color changing is really important, so you split it up further, so you're only looking now at the LED string lights that were color changing.
You've decided that geography is important. So now you're looking only at the times that you sold those string lights in California.
And you continue segmenting to a point when pretty quickly you get to a segment that is too small. You don't really have very much data in it, and you can't go further. So, you're forced to stop there, or even go one level up in order to get the data that you need.
We think that there is a better way to do this.
With segmentation, you might get to a pretty good price, but in some cases, there are still going to be drawbacks. For example, you're not necessarily going to get to use every one of your product configuration attributes.
In this case, perhaps you couldn't use product length because there just weren't enough times that you sold to that product and that region.
You also might not get to use seasonality. So, in this case, you're selling these LED string lights during the winter holidays. That is a key concept to remember.
But if you've never sold these particular lights, during the winter holidays before, or even if you just haven't sold them enough in the winter holidays, that's probably not going to be a segment you're you can use. And you might miss out on a key insight that LED string lights are much higher valued during the winter holidays.
We also aren't going to be able to look at continuous attributes in quite the same way so let's say competitors are offering 8% discounts or costs are high, in this case. We may have previous examples of how competitors have priced but if your competitor has never priced at 8% before, while you've done this kind of sale, it's going to be really hard to know what the relationship is between what your competitor prices and what you should be selling at.
And finally, the core of segmentation is that we are using transactions within a particular segment in order to determine the price. And this means that there is wealth of additional history represented by the rest of the circle here that isn't necessarily getting used and the insights from those transactions aren't playing a part in determining the price for this transaction.
So, this is why we believe that there is a better way, and that way is now available with Gen IV.
So with Gen IV, instead of treating attribute values as discrete entities that cause you to split up your entire data sets, we explored what you could do if you were actually able to leverage all of the data from your entire dataset and leverage the insights from each of those transactions based on the attributes that you see. So, an example of how that may play out, if you've never sold the ten feet string lights in California before, but you have sold the ten feet string lights and multiple other states, and you've sold multiple other types of string lights in California.
You may not be able to get a segment if you were following a segmentation, but you have a lot of information about ten feet string lights and how they behave, and you have a lot of information about California, and how that tends to change the price of similar products. You have enough information to come up with a pretty good price if you have an algorithm that can draw conclusions in this way.
Similarly, if you've seen the seasonal patterns of other types of lights around the winter holidays, even if you've never sold this particular product in the winter picking up on that seasonal pattern helps you identify that the value of these lights may change a lot during the winter holidays, and that's pretty different from for example, a light bulb, which doesn't necessarily have the same seasonal pattern. So being able to identify that this pattern applies to this product is essential. And finally, this is an entirely customer specific value driven process. So, we aren't just setting a singular price for ten feet string lights everywhere, but we're really looking at that customer, in this case, Ben’s Home Improvement: How have they priced before? Are they an underperformer? Are they an overperformer? Do they really value these lights?
And based on what we see in the data, we are able to provide a price. So, we talk about value-based pricing a lot. This is true value-based pricing because you're able to look at each of the attributes and understand their value in determining a final price.
So with neural networks, a few benefits to close out, you're able to leverage a much higher number of data points.
For example, competitor data, market data, because you're not creating a segmentation.
Second, you're able to understand that every unique sale has its own influences.
So, string lights may be influenced by seasonality in a different way from light bulbs.
And finally, you're able to provide a specific price for that combination of product, customer attributes, and market conditions.
So, there isn't the need to provide a higher level price or to provide a price that is based only on data that you have had in the past.
To provide some specific examples of where Gen IV is able to surpass what we've seen in the industry today with segmentation-based models, I wanted to go over a few points. The first is looking at trends and data so seasonality, volatility, inflation. With a segmentation-based approach, you can cover some of that by segmenting based on the time of year, for example, but the amount of patterns that a neural network can pick up on as well as how quickly it can pick up on them is very different. And this is something we've seen in our preview data as well, which Eldho will share. But we're really able to see that Gen IV is picking up on patterns in a way that is able to account for unprecedented situations like the inflation of last year. You may not have a segment that shows this level of inflation, but if you understand the pattern and you're able to see what is happening when you put a price out, a neural network can respond to it.
I've covered the use of attributes and number of data points, where a segmentation may be restricted to using a limited set of attributes and data within the segment, while a neural network-based algorithm like Gen IV is able to recognize that every sale may have a unique set of attributes that are significant and is able to use those data points more effectively.
And finally, explainability.
So this may be an interesting point to a lot of you that explainability is something that we consider a key benefit of Gen IV. I'm really excited to share more of why that is. We believe that being being able to see what is in this segment when you're doing a segmentation is helpful, and that's something that we've always offered with our segmentation-based tools. But being able to understand what your value drivers are for every unique sale, that is incredible value and information that not only helps you understand the recommendation, but helps you understand what's going on in your business.
So, a quote from one of our customers that we worked on a preview with, they said, well, that's just how the neural network works. You give it data. It'll figure out what works, what doesn't, and it changes over time. Really speaking to what they saw in its adaptability and its ability to pick up on really fast changes in inflation.
So, to give you a quick overview of how it all comes together. I'm going to geek out here for just a moment so for those of you who are interested, this will be interesting.
And for those of you who just want to skip to the demo, give us two minutes, and we'll get there.
So the input data goes into transactions.
We have two steps.
We have a price prediction step, which is where we use the neural network in order to look at all of the historical data, be able to assess what the patterns and historical data look like, and predict what the price would be without any intervention.
But the point of price optimization is to intervene. So, we have a second step called price optimization, where we look at win rate elasticity and how that changes at different price points, and are able to recommend a target price point that maximizes your margin or revenue based on taking your win rate into account. So, you're going to have an optimal point where If you go much higher than that, you're at risk of losing the sale. And when you go much lower than that, you are at risk of losing margin. What is that sweet spot and how do you calculate it mathematically? That's what Gen IV is doing here.
Finally, underlying this entire process is model interpretation where we are able to show what is happening under the hood what Gen IV is able to pick up about your business, and how we can play that back to the user in a way that does not require them to understand anything about AI and neural networks. If you look at our UI, which I'll show you in a moment, it barely mentions neural networks. And this is because we believe that this is a tool that anyone should be able to interact with, if they understand the business, and they understand pricing. They did not have to be an AI expert. That's what we and the product are here for.
So in interest of time, I'm going to skip over a couple of slides, and, yeah, I'm going to actually jump into the demo. So what I really wanna show you is explainability.
When you have a sale that you're pricing, you have a series of questions that you may want to know. And you can kind of split these questions into localized questions versus market wide questions. So, a localized question may be about one transaction or one recommendation understanding what attributes are most influential and how your win probability may change at different price points. But you also may have market wide questions. So, these may be questions are explored at an executive level or maybe explored by a general manager or a product manager, looking at in what areas of my business is cost inflation leading to decreased margins, or which attributes are influential in this entire product series.
We aim with Gen IV to answer both those kinds of questions.
So here, I'm going to skip over to a demo and show you what the UI looks like.
Thank you. On the screen, we what we've done so far is run the Gen IV AI on a set of transaction history and as soon as it finishes running, you get a view into your potential opportunity and uplift with Gen IV. So let me zoom in here so we can see what that looks like.
So, you're going to get $636 million based on this dataset over the next few months, and we're also able to show what that's based on historically in terms of the margin that you had in the last three months.
If you want to dive in deeper, we're able to show the price changes that contribute to that uplift. So in this case, you see a lot of variation and price changes because Gen IV isn't recommending the same price change for every transaction. It's actually looking at every unique case individually and assessing, in which cases things need to change. You're going to have a lot of cases, most of them, where the price changes are pretty minor, and you're going to have some cases where maybe there's a severe underperformer or there's a sale where you're really losing out on margin and Gen IV is able to notice those as well and recommend an increase.
So that answers some of the market wide questions, although I have a couple more than I'm going to come back to. But the second type of question is a localized question, which is how do I understand the particular nuances of one transaction?
And there, we have this second workflow where you can enter in any combination that you want. So, note that you don't have to have sold this product to that customer before or maybe you don't have to have sold that product to that customer in that geography or in the same channel. You can really enter the combination that you need a price for and generate a recommendation.
So, what this is doing is actually querying the neural network model in real time and returning with a result. We call that real time scoring.
And here you see not only a floor, target and expert that give you that give you a view into what your recommendation is, but also, this win rate chart that shows you your floor, target, expert, and what the risk adjusted margin is at each of these points. So, you'll see that target has the highest risk adjusted margin, but there are cases where you may want to price at expert if you are a really confident salesperson and confident of winning this sale.
There might also be times when you want to come down to floor if this is a risky sale and you really want to lower the risk a bit, and you're okay with taking that hit to margin.
So, we provide these options as well as the actual probability that is estimated for each of these to give you the transparency to make that choice, and we've seen customers use this in different ways. So, for a lot of customers, they like to provide the entire envelope to their sales users on our quoting tool, CPQ, while others may want to just provide target or they may want to change things a little bit. So, we have a lot of flexibility with how this actually ends up in the final quoting system.
One of the biggest examples of value of Gen IV is in this value driver chart. So here we are able to show you a detailed breakdown of how every attribute is influencing the sale and specifically, which attributes are significant to this sale, how much and in what direction?
So, in this case, for example, you have a product baseline but the fact that the specific geography that it's being sold in creates an increase in price from that baseline and you have some other characteristics that cause a decrease in price. So, this is a really detailed view that you can look at to understand for every sale, how does the customer attribute value to this particular product and to this sale? And how does that change when I look at different sales?
So, what I'm showing you here is for an individual transaction, but in a moment, I'm going to show what that looks like collectively.
But before I do that, I want to jump over to a third localized chart we have coming up over the next few months, which is our what we call our peer transaction or comps for short. So a question that a lot of people may have is, well, it's really helpful to see the value drivers, and it's really helpful to see the simulation of win at different price points, but I actually just want to look at the raw data. What are some transactions that you've priced at with this product or with similar products?
And what's what have those transactions been priced at? Sometimes when you're interacting with a sales user, you may just need to get to an actual example of a historical transaction that has been successful in order to be able to really make your point on adoption.
So on this UI, we have that. We have a chained algorithm called k nearest neighbors that runs over our neural network and what that gives you is a list of similar transactions to the one you're looking up. So, for example, you can see that you've sold LED string lights to a few other similar customers. You also see another view that is pretty analogous, and you can open up any of these to look into the transactions in more detail.
Additionally, we also have an export if you want to, just look at all these transactions together, and to be able to explore how each of them differ from each other in your own BI tool or through your own analysis. We have that available as well.
So switching back to the slide deck, I want to show you what we're thinking in terms of market wide analytics, especially when it comes to that value driver chart.
There we go. So, what I showed you earlier was a view of the attributes for an individual transaction, but this is an incredibly powerful way to look at what that looks like across your business. So, in this example, I've filtered by a channel, and I've also chose to compare different countries to each other. And I can do my own slice and dices, so I could change this from country to state, or I could change it to product series or whatever I choose. But here, I'm looking at five different countries, and I'm looking at the attributes significance for each country. What I'm able to see is, where am I seeing similar patterns and where am I seeing a divergence? So, in this example, I can see in most of my countries, I'm following a similar pattern where the color temperature of the bulbs are the most significant attributes followed by product ID and power type, but I also may see something an unusual behavior here that gives me insight.
So in this example, I see that Germany has a really different pattern in pricing where product series is determining value much more significantly than in any other country. This is insight that is incredibly difficult to get. And is going to lead to a deeper understanding of what is driving pricing in different areas of your business.
To go one level deeper, you can also choose to zoom in on one particular attribute. So here, I've chosen to zoom in on color temperature and look at how each value of that attribute drives pricing. So, you may see some things that are intuitive and already known to you. So the fact that warm gold or muted colors tend to drive up the value and have a higher sale and that flashing neon tends to drive down value. But potentially, you may also see surprising things over here that lead you to take an action whether it's exploring what you didn't know about what was driving your pricing, whether it's something you didn't know about your customers, or perhaps it's something that it's an assumption that is being held in your own business by your own salespeople that you hadn't even realized. Now you have the tools to explore that, and to understand what exactly is going on.
So with the Gen IV AI, Explainability is key. Explainability both in a localized environment being able to answer questions about one transaction, as well as an aggregate.
And again, we deeply believe that being able to have an intuitive user interface where you can use it effectively and get value without knowing the data science, without knowing the AI is where we're planning to go and where we believe we are with this interface.
So I'll hand it over to Eldho to finish up.
Eldho Kuriakose: Awesome. Thank you so much, Kaavya. I think this is just amazing. I mean, I was a product manager at Dell a long time ago. And it took us months to come anywhere near close to this level of analysis.
And it's just amazing that this literally takes minutes for us to pull together. So even as a product manager, I can imagine just looking and pouring through this information to figure out, am I making the right decisions on my product, etcetera, etcetera? And of course, there's so many other such use cases for this.
So, but these results really do speak for themselves. Right? So, a lot of this is what we showed are some of our mocked up data and UX prototypes. But we actually shared this with some of our customers, and we ran their data through it, worked with them on a preview program last year, and the results actually speak for themselves.
One of the customers were a US food manufacturer and there we actually saw that the price prediction was so much better with Gen IV in that inflationary environment. We saw that when you see the data over the times series, you can see the inflation happening. And of course, it's hard to express that in a segmentation model. But in the neural network, it was obvious right away. And that means that we are able to price ahead of that inflation and get an uplift on the attainable revenue. Similar results we saw with the customer we did a preview over in Europe in a different industry. So, we're super excited about bringing this to you guys and seeing your data run through and seeing the uplift and the predictions that we expect to see from your data.
And you know the excitement around this is quite palpable. I'm glad you guys are all here in this room. And I think the booth itself was very much swarmed by a lot of interest. So, we're super excited for that.
We're always happy to continue this conversation one on one or at the booth. This is one of the comments from one of our previous customers. “From the first conversation, this was super intriguing.”
And it just continues to confirm the fact that this is the right direction going forward.
And this also opens up the avenues for bringing in additional data sources. Because again, in the past, when you're in segmentation based more data sources you bring, you're cutting that data up and you're quickly running into sparsity. Here we're able to just continue to adopt that and bring that into the model.
So, this is the closing slide. I was trying to come up with a way to really impart that spirit that we started off with, that, you know, outperformance is not just about putting something in. And I was trying to figure out a way to convey this standing in my hotel room, and I opened my curtain, and this is what I saw was across the street. So, this is the Clifford Still Museum, and this is what's painted on their on their brick wall. So, he said, the canvas was his ally, the paint and trowel were his weapons.
And in his case, the art world was his enemy. Obviously, we don't we don't have any beef against the art world. But we do have beef against status quo. So, status quo is our enemy.
So, though how does it apply to us? You know, the canvas that we're engaged in is our business. It's what we do day in and day out. Our employee,s our partners, our vendors, that is the canvas.
We are all artists, believe it or not. Right? And we're operating on that canvas.
And the tools that we operate with, maybe Excel, maybe pen and paper, but we hope you see the value of bringing PROS in as your tools. And let's kick the status quo in the butt. Right? That's exactly what we're here for. Let's outperform.
Thank you very much.