AI is everywhere, but understanding what vendors are actually offering, and whether it fits your business needs, can be overwhelming. In this webinar, we summarise the complex landscape of pricing relevant AI technologies and autonomous agents, while exploring how they are changing price management. You’ll learn how AI models work and how they interact with each other. We will explore real world applications and value that can be delivered today.
Speaker
Dominic O’Regan
Senior Strategic Consultant, PROS
Full Transcript
Welcome to our, our webinar, the future of pricing, and we’re gonna be focused on the future of pricing, no surprise, but also detangling AI models from agents and autonomous agents and where are we going and how does that fit within the world of the future of pricing. So it’s not a general future of pricing. We’re thinking about how the future of pricing is affected by the evolution of AI models and agents and and so on.
So let’s start with the the basics and the housekeeping stuff for the webinar today. Firstly, the webinar will be recorded, so it will be available on our website shortly after. I’m not sure what the duration is, but soon after we we finish today. So if you miss bits, if your Internet goes down, if the apocalypse happens, you can come back once we’ve sorted that all out and view things on our website.
In GoToMeeting, you should find in the top right hand corner, I’m glancing now, yes, I’ve got that, there’s a little question mark, that’s the best place to put any questions that that you want answered in. And, of course, if they’re appropriate and we have time, we’ll answer whatever we can. There is a chat function. I’m not sure if that’s available, but if it is, it’ll be monitored by our sports staff.
So if there’s a pro a technical problem, use use that if it’s available to you. Okay. So who am I? I’m Dominic O’Regan.
I’m a senior strategic consultant, at PROS. So I kind of focus on how on earth does our stuff work and who does it work for and how do we understand and and explain that. But I’ve been working in pricing and pricing software solutions for gosh. I haven’t added it up.
Let’s say fifteen somewhere between fifteen and twenty years and across multiple vendors. So I have a slightly broader perspective than just PROS, but I have been at PROS for quite a long time now as well.
For my sins, and I think this is a confession to be honest, I I I really like pricing. I’m quite passionate about pricing and about what we can achieve with pricing compared to kind of other functions in in industry. I think it’s a shockingly powerful lever that we have and that we’re in a position of privilege, and I say we in the sense of not just a software vendor like us, but people like yourselves, because you have the opportunity to make a big impact for the companies that you work for. So I’m passionate about pricing, and that is a bit of a confession because it sounds like a bit of a nerdy niche.
But I’m also really passionate about AI both in pricing and in general. So when Pro started investigating some machine learning, networking stuff, and things like that a few years ago when it will kind of in its infancy, this latest generation of of AI pricing. I was certainly one of those people that got wildly excited and ran off, watching YouTube videos and learning how learning networks work and stuff like that, but that’s just the way I’m wired. I kind of like stuff like quantum physics as well. I know it’s weird, but there we are. So I get excited about this stuff and I dig into it and I try and understand it and then my world is one of trying to help people to understand those complex things, and are they applicable, and how can they be applied?
So what are we gonna cover today?
The the agenda, the focus today is on a little bit of what is AI and light, what is AI, but more specifically teasing that into what is the difference between an AI model and an AI agent and how do we understand AI agents and how do we think about them on a maturity curve, stuff like that. So kind of foundational AI models agents stuff.
We’ll also kind of step into and illustrate that with some of the PROS capabilities. So we’ll use some of the PROS capabilities in the world of agents to kind of understand and illustrate what we’ve been looking at, and then we’ll get to the headline of where do we think it’s actually going, what’s the near future look like. I’m not worrying about ten years from now because that’s so far away. It’s ridiculous.
I’m worrying about what’s happening in the next two or three years. Where might we be from a pricing standpoint as an industry, and of course, as PROS as a as a supplier, so that we can think about what does that look like and and try not to try not let it get too frightening. That’s the aim. As I practice this, we’re about half an hour, so that’s what I’m expecting in terms of the duration, give or take depending on kind of questions and and whether I get a bit overly wordy, which I do sometimes.
Okay.
So let’s, let’s move let’s move the slides. Right. So what is AI?
Well, there’s all sorts of definitions. Right? And this one’s a nice one, but it’s just one of many. And this one essentially is saying it’s some stuff, set of technologies, some stuff that is going to enable you to do things without having to do them yourselves. It’s kind of a IT techniques that can go about delivering something that you would have expected a human to be able to do and only a human to be able to do.
And that’s just one definition. There’s actually quite a few different definitions of what is AI, but I think there are technical definitions today and then there’s what do we kind of recognize and feel is AI? What are the things out there that we kind of go, yeah, that’s an AI versus this is the sort of thing that I’m used to in the past?
So we think about things like formulas and spreadsheets as being something that we know and understand, we wouldn’t classify them as AI, but as soon as something is maybe going, oh, I can suggest what the formula should be for you, then maybe we think this is something that is perhaps an an AI based concept.
Actually, to a degree, people’s natural definition of AI at the moment is almost something that is using IT to do something I didn’t expect was possible yesterday, which is not really a very good definition, but it’s kind of how we are grading things and how we recognize, oh, wow. That’s AI. That’s something I didn’t expect to be be possible. I actually think the definition is a bit pointless to a lot to to a large degree. The real question is what can something do and how do I understand it and how do I understand what it can do today, what I want to achieve, is it aligned and where can it go next? How do I kind of get that level of understanding is really what is more more interesting to me.
I I pulled this up because I was looking at different kind of definitions of what is AI, and and this image appeared. And part of it is a bit irrelevant for us, but part of it is. So so this is looking at a different industry entirely. Right? And, obviously, looking at process automation.
And you can see how when you think about process automation overlapping with robotics, then you get robotic process automation. Maybe in the middle of this picture, get intelligent robotic process automation as three worlds collide.
But the reason I picked this slide is actually just for the ML dot in the middle of the artificial intelligence. So when people think about artificial intelligence, some people will think that what that means is something that is using machine learning.
Now that is not by any any actual definition what in artificial intelligence means at all.
The reason people think like that and the reason that actually I do a bit as well is because the things that I recognize as being exciting in the world of AI normally have machine learning within them. So machine learning is just an AI technique and the reality of the really exciting stuff that’s going on at the moment with agents and conversational stuff and language models and all of those things is that what’s underpinning them is a machine learning model. So machine learning is one of the most sophisticated elements that might be part of AI And so sometimes people get confused what’s machine learning versus AI? Machine learning is just an AI technique, and so I thought that was worth kind of throwing in there because that one often gets people a little bit kind of tangled and confused.
So I’d like to move on to a maturity model, and you’ll find some of this some of the content that’s in the webinar today, you’ll find in different in different PROS, white papers, and content online as well, and you’ll find this maturity model within within some of those.
So I quite like this maturity model, but I’ve got to kind of put one caveat on it right at the beginning, which is on the far left, there is a type that is labeled Copilot. That does not mean Microsoft Copilot. Right? And we’ll we’ll come back to that in in in a moment.
This is not saying Copilot is right at the beginning of that journey. Not at all. That’s just a type of AI name, not a brand in this in this context. So let’s explore this a little bit.
Right? So the idea is that you can think of when I look at something that somebody is saying, hey. This is an AI agent.
How do I kind of think of that on a maturity scale? So on the far left hand side, you’ve got things that really are just information retrieval. Right? So and most of what people are thinking of as agents today leverages some sort of a language model so I can say, hey.
Find this for me, and the system finds it for me. Right? Really straightforward. Get this.
Yes. Here it is. There you go. Thank you. I’m fully in charge, and it’s just responding to what I instruct it to do.
Right?
And so that name of a of Copilot existed as a sort of a first generation agent concept. It’s also a brand name, but we’ll come back to that.
So you start to think of then, okay. Well, where do we draw the line and say maybe this is an agent versus some sort of a sophisticated retrieval language model type concept?
And so you can start to think of how is the agent beginning to do more than just retrieve information?
Can it actually do things for me, not just retrieve? Can it take an action for me? Retrieve information and then act on that information for me. Purely under my instruction perhaps, but can it do something and not just retrieve information? K.
But then we start to see different, let’s say, advanced capabilities on agents. Now we’ve got them in a maturity model here, but we we can see some of the most mature components appearing before some of the least mature. It’s not always a beautiful step one to the next to the next to the next.
But we start to see an autonomous agents where they are beginning to do things not just based on my instructions. Perhaps it can review information on its own schedule and either suggest actions or take actions on my behalf. So beginning to have more well, maybe agency is a good word. Right?
But beginning to be able to decide when it should do something and take action if I’m if I’m comfortable with that. Right? We’re beginning to think about things that we might want to be more involved in deciding how happy we are to have something just running and and making decisions for me. And then we think about orchestrating agents and agents to agents.
So orchestrating is really saying that the the concept of an autonomous agent maybe is moving into a world where there are there are multiple disciplines involved. So instead of maybe being a pricing related agent, it might be a sales and marketing agent that is managing pricing as one of the things amongst a portfolio of of disciplines that interact with each other. Right?
And then agent to agent is really going beyond just orchestrating and saying, I can’t do everything on my own even with multiple data sources and multiple focuses. I actually need to interact with other agents and hand off jobs to them to do and worry about and and so on and so forth. So you might get a different agent systems interacting with each other and and working to kind of achieve your your goals. So because that first one was labeled Copilot, we kind of have to address Copilot a bit because it certainly isn’t right on that far left hand side.
It does the copality stuff. It definitely does the agent stuff. It certainly has elements of autonomous and orchestrating, and I’m even beginning to see some components of agent to agent. So don’t for a moment think that I’m putting it at the beginning of that maturity model.
So I as I start to think about what what is a model, because for me model is something else, not just the co pilots and the agents, what is a an AI model? What is an AI agent?
And for me, what is an autonomous agent? So I’m not really that worried about anything to the right of the autonomous agent. Once it’s autonomous, it’s either works on its own or it’s multiple discipline or it coordinates with others. I’m putting them all in the same bracket of it’s an advanced agent.
How do I determine when I look at something, what is it from those from those three categories? And really there’s three simple questions and so the first is is it a point solution? What do I mean by that? I I mean is it designed to do one thing, one thing and one thing alone and do that really really well?
If the answer is yes, it’s probably an AI model.
We will come back to that a little bit in a moment as well, but an AI model, something that has a particular objective in mind and it’s obsessed with that and pretty much nothing else. Right? The advantage, of course, is that it can be really, really tuned and bespoke to that particular capability. Right?
So then then we look at saying, say, can it take actions? Does it have agency? Can it do something beyond just chatting with me and answering a simple question? Can it actually make a change?
Take an action on my on my behalf? That’s the real trigger for saying, yes. It’s an agent or not. There’s, again, hundreds of definitions, but for me, I’m going, okay.
Yeah. Now it’s more than just a model. It’s an agent. It can take action, not just not just do its kind of point solution behavior.
And then the last one is, can it actually do this on its own? Can it sit there worrying about, Thursday, I’m not happy with margins, maybe we should be thinking about this. Is it is it able to go beyond just taking an action, and is it be able to start operating on its own with guide, with review or not, and that’s kind of that final trigger for me of, okay, it’s that next level of sophistication. So in my mind, it’s models, it’s agents, it’s autonomous agents, and while those definitions are interesting and understanding it from that maturity landscape is interesting, don’t assume that an autonomous agent is better than an agent or a model.
Depending on what you’re trying to achieve, it might be that a simple model is actually the best answer for a particular problem or a particular situation.
Now as we evolve, we’ll start seeing these things coming together a lot, lot more so that an autonomous agent is choosing when to leverage an agent and when to leverage a model. But for now, we can put them in those categories and we can kinda go, okay. Just and now I understand what they are, second step is, okay. Where’s the most value?
What’s the most directly applicable for my business situation? So why did I stick that don’t look behind this box panel on there? I put that on there because as I was trying to draw this out, I had another decision that I was thinking, where does this fit in? And that decision was, I can I talk to it?
Does it understand me? And and I suppose in a way what what I realized as I was trying to draw this little flow, and I’m not a flowchart expert, so you, you know, if you understand these things, you’ll know my flowchart’s pretty rubbish, but that’s more about how do I communicate with it and is more specific to kind of language models and stuff like that. I want that, but I might want that as my front end to a simple model or an agent or an autonomous agent. That’s just how I communicate with it, not is this an agent, is this an autonomous agent, is this a model.
So I thought I’d kinda leave that on there as an interesting kind of side note. The way I communicate with it is just that and nothing more.
Okay.
So if we think about, now moving from, a model, something that does a a point job and moving into the world of of agents, we can start to think about the things that they are doing beyond just that model. And we’ve explored a little bit of this already, so we’ll move through this and and the next slide, reasonably quickly, I think. But when we think about it becoming an agent, we can start to expect it to do some reasoning and you see this already in things that you think of as a language model. Right? So we think of using like Copilot and ChatGPT and things like that. Right?
You will see them already beginning to take a question and turn that into a series of steps or questions or activities, a reasoning component. How do I not just respond to that piece of information, but how do I address that in a structured logical fashion? What is the best way to answer that question or to take the actions that that I’m that I’m being given? And you’ll see that already because, you know, a lot of what we start what started as models are blurring into the world of agents and autonomous agents. Right? So reasoning. How do I break that down?
Tool use, I love that phrase. It’s it’s it’s a bit like, you know, early man who maybe we became agents when we started picking up tools. I don’t know. But but tool use is is when we think about, okay, can it take an action for me? How is it gonna do that?
You know, in language model, I can chat with it all day. It needs to be able to use tools to do something for me before it’s an agent, so that means it needs to be able to access something else probably that can take that action. So I need to be able to trigger a call to some other system that does that, that makes the change that I want changed. Settle my discounts to ten percent.
Okay? Brilliant. Yay. I understand you perfectly well, but I can’t do it for you.
If I can, if I’ve got the ability to trigger that, then I’ve got that tool and I can use it and I might have a library of tools and capabilities that are supported and and accessible and usable by my by my agent. And then that last category that we’ve already explored a little bit, but the idea that that the that this becomes autonomous and that I am able to evaluate, determine, and trigger those actions all on my own potentially. Right? That’s where we begin to get a little bit frightened and scared.
Well, we might. We shouldn’t. We might. Right. Okay. So once we’ve got into that world of autonomous agents, then again you can start to think about how do we break down their capabilities.
We’re now in that world of maybe I’ve got an agent and it is fully autonomous. So what does that mean? Well, it it means that it needs to be understanding the world and observing stuff. Right?
So what what is that? In the pricing world, that probably means it needs to be looking at the prices that we’re achieving, the sales data that we’ve got. It probably also needs to be looking at some economic indicators or some competitive information or whatever is relevant to the to the to understanding are we doing a good job in the world of in the world of pricing.
So then can it plan? Well, planning can take all sorts of forms, but it could, for example, in the world of pricing, take the form of I understand what’s going on and I think there’s a better way or a better strategy.
I can start to plan that strategy and to suggest, well, I would apply this strategy for these types of customers in these situations. Let’s not do our traditional cost plus approach. For this sector, what we probably should be doing is aligning with our competition or using some value strategy or whatever else. And so an autonomous agent might be looking at your situation and trying to plan out what would be the right the right approach.
And then, of course, the last thing is coming back to that kind of tool use, which is the ability for it to enact that, to make that strategy change, or or at least to put it all in place subject to your kind of approval or blessing or whatever else. Right? I mean, just because we have an agent or an autonomous agent involved doesn’t mean we have to remove oversight and approval. I mean, you can think of these agents as like well, this sounds terrible and scary, doesn’t it?
Replacement employees. Right? But you can think of them as a as an alternative to a human doing something. We don’t just let humans do whatever they want without any oversight or any approval.
Most of them don’t anyway in most pricing situations.
So we might not do that anymore or less with with an autonomous agent, but it’s still able to do all of that work for me and let me make a a final decision or a determination as to whether or not I’m I’m kind of happy with that. Right?
Okay. So if we bring this now to a moment of kind of like, okay, so let’s illustrate some of some of what we’ve been kind of exploring and this is the moment again to just tune into that models versus agents versus autonomous agents concept. So if I look at at some of Pro’s capabilities and and I am very focused on the b to b space, that’s my particular discipline. So so we can talk a little bit about some of the travel stuff, but but my focus, as I said, is the kind of the b to b. So I’ll illustrate more by talking about the b to b stuff.
But but we have models, so models now, not agents, that are looking at things like, if you are going to negotiate a price with the customer, how do I help you work out what the right price should be? So it’s a it’s a point solution. It’s an incredibly sophisticated point solution and model and it knows better than you know what number you’re gonna end up with and what number you should end up with. It’s that good. But it’s still a point solution. So it’s a model. Right?
And then we have a model or models that look at things like take it or leave it. So list price type situations. Right?
There are models that are designed outside of b to b in the travel space that try to work out what the price should be.
And they’re balancing sell today versus what would I get for what what could I charge if I don’t sell it today. Right? So they’re introducing that sort of dynamic.
There are similar dynamics in some b to b situations that we call capacity constraint, but we have a library of AI models, point solutions, very sophisticated, but point solutions. Point doesn’t mean they’re not brilliant. Right? They are. Anyway, but they’re models.
They operate. They do a job.
Then when we think about moving into agents, we have agents of different levels of of sophistication. So we might look at, for example I’ll start with the bottom the bottom here and work my way back up actually, rather. I should have put them in the other in the other way around on the slide, shouldn’t I? Anyway, so if I think about the insights agent, what that’s doing is helping me to understand what’s going on in the world of pricing. So I can create charts and views and do stuff like that with analytics and reporting. Hopefully, can do that in anybody can do that with any reporting solution on any set of data including pricing.
But what the pricing insights agent is doing is understanding my pricing concept and context when I talk to it and translating that into analytics and information that’s answering specific questions that I might then want to have as a regular chart or view or might just be part of my investigation, if you like. Right? So for me, as an agent, but it’s actually at a sort of a low end of agents. God, they’ll probably tell me off for saying that, but but what I mean by that is it’s understanding information and it’s retrieving information.
And what’s brilliant is it can retrieve anything to do with pricing against that pricing dataset. Right? And we don’t say, hey. This is a hundred agents because it can answer a hundred questions.
We don’t limit the number of questions. The only number of questions that limits it is the sort of data that you have available. Right? So if you have your pricing data broken down by, you know, huge levels of detail, then the AA agent is able to answer any of the questions that are built on that on that data foundation.
Right? So for me, it’s an agent because it understands the data and is able to intelligently interpret, access, retrieve, and deliver. But it’s not going in and changing something. It’s not taking an action yet.
Right? That’s why I have a kind of that kind of early stage evolution.
I’ll go to the to the CPQ assistant next, and we’ll look at these and see them in reality in in a moment.
And what that is doing is helping me so CPQ is all about quoting. Right? So how do I actually get the prices?
And and the CPQ agent is helping the salesperson to work out what should I be selling? How do I find the right products and services? And how do I put them into a quote for my customer and maybe adjust the the pricing? So it’s sort of a combination of a little bit of pricing and but it’s mostly focused on how do I understand the catalog of products and services that we have available.
So again, agent because it needs to understand that data structure, and that data structure is unique to each and every situation in business. Right? So it’s gotta be able to key into that, understand it, interpret it, access it, but it’s beginning to take actions as well. So, okay, I found something for you.
That’s brilliant, but maybe put it in the quote for me. Actually, add it to a quote. Find me a selection of stuff. Put a quote together for me. It’s that building block to the end goal, which might be just create the whole quote for me, frankly.
But it’s how do I do that first step? Find stuff, add it to the quote, alter things, manage it for me. Right? And then the revenue management agent, which is why I actually know the least, so I’ll be a little bit careful about what I say because my focus, as I said, is b to b, but my understanding is that again is working in an agentic fashion and is doing a little bit more of the autonomous concept where it’s beginning to monitor stuff and make suggestions based on what’s happening in real time in the world.
So again, whenever you look at a company like PROS or anybody else because every software vendor is talking agents, understand, okay, you’re saying that’s an agent, what’s it really doing? How sophisticated is it? Is it using tools? Can it work on its own?
Where do I and then what value is it is it adding? And if I look at this picture here, the AI agents are are exciting, but I can tell you right now the models deliver absolute measurable, indisputable value in a way that the agents are quite frankly wishing they could at the moment. Right? They are they’re gonna be the most sophisticated version of themselves when what they’re doing is accessing the models to deliver information.
Right? The models will be the foundation for the most intelligent agents that are that are out there. They’ll be the tools to to a degree that the agents are using and leveraging. So you can see these sort of building blocks coming together that are going to deliver and and in my mind this becomes a little bit of a, you know, a stone rolling down a hill and building up moss and becoming bigger and bigger and bigger.
As these things start to move, you start to get a momentum and you start to see incremental value being unlocked as things talk to each other and become enabled. So let’s have a quick look at one of these a couple of these agents.
I’m gonna focus we’ll skip past this. Let’s get straight into it. So here’s an example of the Insight agent. Don’t worry if it’s teeny, we’re gonna zoom in in in just a moment, but what we can see here is an environment, it’s a PROS environment where I might be looking at analytics and reporting insights.
How do I see a dashboard? How do I view stuff? How do I understand things? And the agent is that pop up on the right that’s saying, hey.
Okay. You’ve got some charts and dashboards. You’ve planned stuff in the past, but maybe I can help. Just tell me what you wanna do.
Right?
And and so I was playing with this a little bit earlier, and so I put together some little screenshots of a conversation I had with with the agent. So you’ll notice that this agent starts by saying, well, here’s some suggestions. These are things that people quite often ask me for, but I’m not locked in. I can go in any direction I want, and so in this case, I just said, okay.
Well, yeah, let’s chart sales by region. Let’s go with that. Let’s find out what we can do and so the system came back and said, right, okay, here’s your revenue in a chart by region and I looked at that and I was like, okay, yeah, but it’s a bit messy, it? Simplify it.
Give me my top five.
So then okay. Yeah. I can I I’m using the language model? I understand what you mean by top five.
I did I even say by revenue? I don’t think I did. I didn’t tell it I needed it by revenue, just top five. So it worked out for itself. You probably mean revenue. Right?
And put that together for me in a nice visualization and then I said okay, well that’s nice but I really want to worry about the margin by by revenue as well, so show me both of those things.
And I can go on from here in any direction. I might start to say, okay, well, what is the discount in my top regions?
And I might see and I look for is there a correlation between that discount and my revenue or my margin and my performance. Right?
I want the agent to be able to answer any of my questions to do with pricing and my pricing data. So it’s a it’s a nice way to jump start my understanding and if I find something I like, I probably want that as a chart I come back to you on a regular basis. Right?
The other agent that I wanted to show you again because it’s b to b, so it’s the stuff I understand and love is the is the quote helper. So again, we’re starting small, but we’ll zoom in we’ll zoom in. Don’t worry if you can’t read what’s on the far right hand side. What we’re looking at here is a quote that has been put together almost exclusively by the agent.
Right? I this was a blank quote, and I’ve used the agent to fill it in and to and to find stuff. So again, the conversation I had with the with the agent. Here, I I had a really simple conversation and I started perhaps a little bit naively by this is an industrial example.
Right? So I what did I type? I put something like, yeah.
I want some industrial hoses. So okay. Now the agent looks at that information, and it’s doing some planning and understanding. It’s looking at the information and saying, okay.
Industrial hoses. There’s a lot of things that count as an industrial hose. What are you after? Help me here.
What sort of hose are you after? What’s it for? Or what sort of materials? Or what sort of length?
Give me some clue to help me narrow this down because there’s a lot of industrial hoses within my database. Right?
And it’s actually suggested things. You might want the material, you might want the size, just give me something. So I responded and I said, just give me hydraulic hoses and give me various lengths various lengths and materials, so don’t worry about materials, give me a selection. So it did, it came back and in the middle you can see it’s listed out some different lengths and different materials available for those for those lengths.
So what did it come back with, like, six items and then it’s and then it’s saying, do you want me to add them to the quote for you? It’s suggesting the tool use. It knows it can do this and it’s saying, hey. Don’t waste your time adding them one by one.
I’ve found them for you. Surely, you want me to put them in the quote. Right? And so that’s that’s what I did.
I said, yay. Add it to the quote.
And then I went, oh, yeah. But I could do with a bit more than one of each. Right? Give me ten of each.
So then it updated the quantities and and, you know, you can you can see how from this starting point, you can start to build an environment where that agent becomes more and more capable to change the quantities, to set the discounts, to whatever else, not just find the right products and services and and you can see how this can take another step towards automation where rather than me saying, hey, I want some hoses, you could imagine the agent starting to say, okay, you’re putting together a quote for this type of customer. I can suggest the sorts of products and services that they should have and the quantities and the right prices and then review and adjust rather than start with instructions.
Right? You can see how that journey unfolds.
Okay. So we said we’d get to the the the future of pricing and and AI and autonomous agents and all the rest of that kind of stuff.
So some simple thoughts and then a and then a kind of a don’t worry too much message. So for me, when I think about the near future, and I do mean near, I mean, you know, within the next few years, not the next ten, twenty years or anything crazy like that. Right? I’m I’m looking forward to the point where I can have pricing strategies, pricing logic that is completely created by an AI agent. It understands my data, we can see that already, it understands my history, we can see that already, and it is able to bring that together to say I think this is the right strategy for this situation, Here’s why and would you like me to do it for you? Would you would you like me to adjust that? Right?
I think we can start to see agents that do and in the very near term, we can start to to think about agents that do particular tasks for me, that focus in that single discipline. Things like promotion management, for example, is an obvious quick win for agents.
So is setting up and suggesting rebate models in the world of pricing. Quick win for agents. We’ve got one of them already, by the way, that’s I didn’t put on the slide.
But you can see places where there are specific activities that an agent can really help with. Okay. You want a let’s talk rebates for a moment. I you want a rebate program that is going to drive growth for this type of customer.
Brilliant. I can suggest this structure, this logic, these thresholds.
Do you wanna do it? Right? And that’s based on historical performance and what works and everything else. So you can see those point solutions in the really, really near term.
Point solution agents, I should say, not point solutions. But but in the end, I’m expecting in the not too distant future to get to the point where the agents are if I if I was a new customer of a pricing software vendor, I would expect to just give it all my data and have an agent say, right. I can set up a system for you, and it will suggest what the strategies are, and it will implement them and do almost I say almost. Actually, no.
Let’s be honest. I think we will get to the point where these things are pricing consultants, and that’s where the way to think about them in in the context of a business. So okay. That might sound a bit scary.
Hopefully, doesn’t, but I do I do I do want you to not panic. Right?
The for me, the reason why these things are not that scary is because we will always be in charge and and if I think about some sort of a top of this hierarchy for a moment, some orchestrating magnificent business consultant agent that looks at my data, that makes decisions, that makes changes, and all the rest of it, that isn’t just going to appear one day and go from nothing to that and just trust it.
You’re starting at that foundational level. Have I got models that will do things that I can validate, that I can trust? They’re the tools. Right?
And they might be AI models like we’ve explored. They might be simple logic concepts that either I’ve created or AI has created for me, but I’m building up a library of capabilities and tools that are kind of validated before I think about, okay, now I have something that can leverage and orchestrate those. So this isn’t a sudden wallet. We’re in a world of everything’s happening and I don’t know what’s happening and how, but more a a layer cake.
We’re gonna build up towards that with those tools, those point solutions, those initial agents that are leveraging them and we will naturally become comfortable with and know how to own and control and manage these models before we’re in a world of sit back, relax, and watch the cash come in. Maybe that’s maybe that’s overstating it. I don’t know.
So so I would say so thank you for your time. We we spent a little bit more than the half an hour that I was kind of expecting.
I’ll have a little look now and see if there’s any obvious questions.
Let’s have a look. Okay. So maybe the question bit wasn’t wasn’t working, but I can see a no, I’m not seeing questions and therefore I’m gonna allow the the system to to close. Obviously, PROS is available for questions and answers, there is material on our website, there is a link in the chat right now to a white paper that’s covering some of this and actually it was written probably by one of our like crazy expert guys, crazy expert guys, one of our professors in AI pricing and you can find a lot more content and of course we are always available. I get wildly excited about talking about this stuff so anybody that wants to know more, I’m always happy to spend time with you and to kind of cover this to whatever level of detail you want But thank you all for your time today, and I hope it was, to some extent, useful.
Thank you.
