Business leaders are facing increasing uncertainty and must adapt plans quickly as as supply chains, tariffs, and forecasts shift faster than quarterly planning cycles can keep up.
The conversation begins by explaining optimization and the types of challenges it addresses, before exploring how generative AI is making optimization insights more accessible to executives and where quantum computing becomes the engine for solving the most complex, constraint-heavy problems.
Mathematical optimization has enabled decisions in areas such as airline crew scheduling and logistics routing for decades. GenAI bridges the gap between people and models and helps them keep up with the increasing pace of business and complexity of decisions
Quantum computers use superposition and entanglement to explore massive solution spaces almost simultaneously, rather than working through possibilities one at a time. That makes them especially well-suited to the most computationally complex optimization problems that were not possible to address previously. For example, in a multiple-vehicle routing scenario, a quantum optimizer found a feasible solution in 3.5 seconds compared to a classical approach that couldn’t find one in two hours of high performance compute.
Just as organizations rushed to apply AI to everything when it emerged, the same pattern is starting with quantum. Good-fit problems involve lots of interlocking decisions under constraints, such as routing, scheduling, network design, production sequencing, and inventory positioning. Pure forecasting, image recognition, or sentiment analysis are better suited to AI and machine learning, not optimization.
The biggest gap for businesses to leverage this technology currently is not a lack of data. Without a clearly defined decision, even the best model or quantum solver can't help. This means leaders play a crucial role in understanding and communicating the business's variables, parameters and critically, its risk posture.
- Introduction
-
Helen Fang (00:05)
Hi everyone and welcome to our From AI to ROI podcast. Today's episode is going to be on optimization, GenAI and quantum and how these three domains can be brought together to solve some of the biggest challenges business leaders are facing today. My name is Helen Fang and I'm joined today by three of our CGI experts, Parimal Kulkarni, Mariolys Rivas, and Curtis Nybo. So maybe we'll start off with some short introductions. Parimal, do you want to go first?
Parimal Kulkarni (00:37)
Hey everybody, Helen. I am a Director of AI, data science, and machine learning with CGI. I've been in the consulting practice for about seven years now. I support our clients in public and private sector for data AI supply chain, whether it's for logistics, whether it's for healthcare, finance. This happens to be one of our areas of focus for innovation around AI and quantum. So happy to be chatting about this today.
Helen Fang (01:14)
Thanks, Parimal. Maybe Mariolys, you go next.
Mariolys Rivas (01:19)
I'm Mariolys Rivas, Senior Data Scientist at CGI. I spent most of my time on the technical side of things, doing research and code. I've been working on different projects involving tools like Generative AI and data analytics and I'm very happy to be here too.
Helen Fang (01:35)
Thanks, and Curtis.
Curtis Nybo (01:36)
And lastly, me. I'm Curtis Nybo. I'm a Director of AI and quantum computing at CGI. I've been with CGI actually for 10 years now. So mainly spent a lot of time, as Mariolys and Parimal have said, as well, in the technical space doing actual hands-on development. And alongside a few colleagues, I lead our global quantum practice at CGI, where we try and apply quantum computing technologies to our clients' challenges today, and optimization is a very, very big part of that. So excited to be here.
- What is optimization and why it’s more relevant than ever
-
Helen Fang (02:09)
Thanks so much. And it's really exciting to be talking about this topic with the three of you. It's super exciting, really practical, and I think not enough people know about it yet. So hopefully that conversation today will change that. So off of what Curtis has just mentioned on optimization, I think when maybe IT professionals or people in the wider industry think about optimization, maybe they start to think about IT optimization or cost optimization or anything else around optimization. Maybe Parimal, you can start off and explain a bit about what you mean when you guys talk about and work on optimization.
Parimal Kulkarni (02:50)
Yeah, that's a great clarification. So, I know we're a technology company and people immediately think, oh you're going to talk about IT optimization, you're going talk about how to get our servers to perform better or more performance out of code or system tuning. That is not the optimization we're talking about today. That's IT optimization.
What we're talking about is mathematical optimization. It's a field that specialists call operations research. And all it means, to me, and if I were to define it, is a business scenario, a business setting has millions of decisions to make. There are variables, there are constraints, there are risk factors to consider.
And optimization is really the science of making those decisions possible, the best possible decision when you have a lot of these moving parts and all of these dependencies.
And to make it a bit more tangible, think of a shipping company or a logistics company, for example. They're trying to send orders across the border. They're trying to decide which truck to use in their fleet, which driver, what are the tariffs, what are the fuel prices, what are carrier contracts. I mean those combinations for making the best possible decision can run into millions. And human planners are amazing, but they cannot evaluate that in their head.
And so this field of operations research and optimization has been around for decades. It's not new, but in some ways it's been invisible, because either, you know, the supply chains were more reliable, costs were more controlled, but airlines have been doing this for decades, right? That's how they schedule crews, that's how they schedule flights. So in this setting, optimization really is I'm going to build a mathematical model of this problem, I'm going to define the objective. What am I trying to maximize? Is it how I use my trucks? Is it where I put my inventory? How I reduce my tariff costs, whatever it might be. and then what we're talking about here today is really how do we make optimization more visible?
This is this is the optimization of improving business decisions and we'll talk about this in a little bit, but really what has made it visible is AI in some ways and the fact that people interact with models in a more natural language. And so that's the sort of optimization we're going to talk about today.
Curtis Nybo (05:31)
And I think the distinction of that it's been around forever is important. A lot of folks assume that since AI has come around, that's when we're starting to see more optimization. But like Parimal said, people optimized things using spreadsheets back in the day. And then cloud compute became available and we started to apply machine learning algorithms to improve optimization processes and be able to take in more data. And now it's just become a larger beast where we can tackle huge optimization problems compared to back in the day and we can also get a lot more accurate results from it. And then as we'll talk about, pairing that with AI, optimization's not necessarily AI, but we can definitely augment it together.
Parimal Kulkarni (06:13)
And nothing against spreadsheets, right? We can still build optimization models on spreadsheets.
Curtis Nybo (06:19)
Yeah. Thank God for Excel Solver. Everybody knows that, but that's optimization. At the end of the day, you're maximizing a function or minimizing a function.
Helen Fang (06:26)
Yeah, and I think that's such an important point to make that a lot of, you know, like the foundations of the things that are possible or not possible today have been around for like a really long time. It's just maybe the tool sets or the ability for non-specialists to maybe interact with that information or utilize in their jobs is newer. And I think also to your point, the intersection of a lot of these different fields is kind of where you can get the real potential out of it.
And especially in today's world with so much happening in the world and how interconnected and how global a lot of the businesses are now, I think it's definitely like a really critical area.
Parimal Kulkarni (07:07)
And how rapidly everything is changing, right? Just the business environment is so much more unstable now. We're no longer at a point where, you know, we say we're going to make a decision for a quarter and just hope and pray and wish that that decision holds. Like we're having to replan, restructure, reformat overnight. And that's why I think the technology and the global economy are at that point where this has become very, optimization has become so very visible and relevant.
- How GenAI makes optimization more accessible
-
Helen Fang (07:39)
Yeah, and to that point, Mariolys, maybe do you want to elaborate a bit on Curtis's earlier point around how AI and Gen AI are helping to kind of advance what we can do with optimization.
Mariolys Rivas (07:54)
Yeah, sure. Just to keep it clear, GenAI models are not here to replace optimization models. In fact, LLMs cannot effectively do the math that is needed for optimizing operations. What LLMs can provide is a bridge between the user and the optimization model. That's because LLMs are great at speaking the language of humans as well as the language of computers.
So if you provide a series of constraints using natural language, then LLMs can translate these numbers that can be the input for optimization formulas or algorithms. And later, after those algorithms return the results, LLM can present them to you in natural language.
So I would say like the takeaways here are, first leave the LLMs at home, as Curtis said, what we chatting about before, until your optimization models are built. And then second, optimization decides while generative AI translates and communicates. There are two different engines with two strengths, and you need both of them. I don't know if you need to add anything to it, Parimal and Curtis.
Parimal Kulkarni (09:12)
Yeah, the one thing I will say just to really spotlight Mariolys's point there is what has made GenAI and optimization come together is this ability to talk to the models, right? Like this is no longer, you know, there's three mathematicians or operations researchers in a back room creating models. This has now gone up to the boardroom. Like there are executives who just want to ask questions of their models. They just want to say, what if I cap tariff exposure at 5%? Re-optimize. This is no longer the whole long feedback chain of I'm going to send a set of requirements to my analysts and then they're going to redo this and I'm going to get an output, right? That that pace and that workflow has changed because of GenAI. And you're right, they're two different engines, but work very well together.
Curtis Nybo (10:04)
Yeah, I think overall the barriers to entry to being able to build these optimization models, to utilize them, you don't have to necessarily buy expensive proprietary software to do a lot of this. You can and they do a very good job, but you can have your teams really access free open source tools that do a lot of this work at scale. And so the AI has helped break down the barriers to applying optimization tremendously. Whereas Parimal said, you don't need a whole room of mathematicians and stat statisticians to be able to implement these tools. You can have just your developers who are used to the usual mathematical problems that they might come across in computer science be able to tackle these. And I think that's pretty huge to be able to democratize access to these optimization tools.
Parimal Kulkarni (10:55)
Mariolys, do you remember when we were doing the fleet optimization one where like the scenarios that they wanted to throw at it was just like my frozen goods truck broke and so re-optimized. Do you remember that?
Mariolys Rivas (11:10)
Yeah.
Parimal Kulkarni (11:34)
And so it had to take that natural language command and convert it to the variable that said whatever FSE meant frozen trucks and then it had to reoptimize and that's the kind of power that Gen AI I think brings to the game.
Mariolys Rivas (11:51)
Yeah.
Helen Fang (11:28)
That's so funny. How would it work in the past, right? If someone was like, I want to figure out what happens when a truck when a truck breaks or doesn't work, would it just be that the team would write back and say sorry, that's like not worth our time or compute?
Parimal Kulkarni (11:47)
Yeah. I think thankfully the planning and logistics suites are fairly mature. I think they would eventually get around to the idea that I need to take a truck out of the fleet and replace it with another truck. I think what is different is the pace of redesigning those scenarios. And Mariolys, you can probably talk to this more about how quickly we were able to do that simulation and scenario analysis.
Mariolys Rivas (12:18)
Yeah, we're working on fleet optimization problem. As Parimal mentioned, you have trucks and you have orders and you want to find and you have some constraints as well like truck capacity and restrictions and cross-border travel and you want to find the best possible way to respect this constraint so that nothing breaks. So when people think about optimization models, they think there's a big data problem, but it's not really that. It's more about combinatorial problems. You're trying to find the best possible solution with certain interlocking constraints, right?
But yeah, so we work on this problem and the way we could interact with the model was way simpler because I will give it like a natural language question and it would translate this to the model and then we get the solution and then we get an answer in natural language and we get to understand it. But it's not a big data problem, it is mostly a combinatorial problem in which you can interact with the model. It's more accessible to you than it was before because now you have the LLM in there helping you communicate with it.
Helen Fang (13:59)
Thanks, Mariolys. Yeah, that's a really concrete example that helps a lot, I think. And then you can imagine it for a lot of different scenarios as so many industries have different types of supply chains. I remember even from a conference that some of our colleagues ran in Finland, right? A hospital CEO was super interested in hearing the manufacturing supply chain optimization examples because hospitals have patients and also decisions to make about their resources. So I think it's also increasingly applicable to lots and lots of different industries and areas. Maybe Curtis, before we jump into the next topic, where does quantum come in with all of this?
- Quantum supremacy? Quantum as the engine
-
Curtis Nybo (14:41)
Yeah, that's where Mariolys set it up for a slam dunk by basically prefacing it, that it's all about combinatorial size and computational complexity, it's not about big data. And so that's where not necessarily about big data anyway, but it really does come how complex the problem is. And that's where quantum computing has become one optimization tool that you can really use to solve really, really, really complex optimization problems that you might not be able to.
And what I mean by that is complex problems in the sense that there's too many constraints, like Mariolys talked about, that you can't fit it into a classical computer. Or there's too many possible solutions that the classical computer can't explore each solution effectively to be able to and in in a reasonable time to be able to find the best solution out of you know billions and billions of possible solutions.
So as you have more variables in a problem, as you have more constraints in a problem, that problem complexity starts to explode. And so, how can we tackle that? That's where quantum computing really comes in.
And so, in kind of 60 seconds, what quantum computing is to level set for everyone, it's really a new way of computing information. It's not an algorithm, it's not AI, it's a completely new physical machine that harnesses what's called quantum bits.
And so it sounds complicated, but really what it comes down to is a quantum computer can use superposition and entanglement where we can kind of interconnect these qubits. We can use superposition and entanglement to explore that complex solution space very, very effectively. We can explore it almost simultaneously. So we can do a lot of parallel type of computation very quickly and very efficiently to explore the solution space all at once. So what I mean by that is we can explore the set of possible solutions to a problem all at once versus a classical computer that would have to iterate through those sequentially in some form and kind of explore them one by one. A quantum computer can do that much faster by exploring the solution space kind of all at once using superposition or entanglement.
And so they're not necessarily just faster computers, they're better suited and they're faster at certain problems. And those problems really are those computationally complex problems, which is optimization.
We have our other simulation use cases, our forecasting use cases, but optimization is a very, very good use case for quantum computers because there's a lot of possible solutions and we can explore that solution space very, very efficiently and using less energy essentially, so we have better energy efficiency when we're computing and finding solutions to these problems. And so quantum computers can essentially help solve problems that can be practically impossible for classical supercomputers.
And what I'd like to leave with users today, if you take one thing away, it's that we can do this today. We can use quantum optimization today. We can implement it in production, we can implement it at scale. The quantum computers to do that are called quantum annealers. And they're available. We can access them, we can build the tools, we can solve these problems. And so when we run into very, very complex problems, there's often a use a case there where we could use quantum optimization, to be able take advantage of superposition and entanglement to solve those problems, basically find the most feasible solution for that use case versus a classical computer that might take much, much longer. So hopefully that gave a bit of an overview of what quantum computing is and how it applies to optimization. But again, that's really one of the most exciting use cases for quantum. And it's one of the ones that we can apply today. Quantum's still a developing technology and so let's apply it where we where we can get some benefits now that is really in the optimization space.
Parimal Kulkarni (18:29)
Curtis, I think the thing that made it most real for me was when you had the demo of the store planning. And I'm a classical optimizer. I tend to always use classical optimization. And Curtis had these side-by-side demos. And you could see how when you took away the slack constraints, the classical optimizer essentially, not that it reached in feasibility, just couldn't search the solution space anymore, whereas the quantum optimization actually gave you a solution. And I think that is so representative of what actually happens in a business, right?
Curtis Nybo (19:09)
Yeah, and what Parimal is referring to, it's a really good example of that, where we benchmarked a multi-vehicle routing with capacity constraints, so capacitated multi-vehicle routing is what the problem's called. Where we're routing vehicles from a delivery point to where we have to deliver the goods to stores. So we have a depot, we have to deliver it to a bunch of stores. Those stores are demanding a different amount of product. We're trying to route multiple trucks, six to seven trucks. We want to make those trucks run as efficiently as possible. We want them to drive for the least amount of time, we want them to use the least amount of fuel, but we also want them to run as full as possible. We don't want them to have any empty space and we don't want to be sending extra trucks out onto the road. So you can see that we're now we're trying to optimize several different parameters and say from you know 10 trucks to 80 different locations across a large city, that the possible solution space explodes.
And what we did was we applied a quantum optimizer where we were able to benchmark it also against a classical optimizer. And what we found was that when we reduce the constraint, when we when we increase the constraints that the trucks had to run as full as possible, we found that the classical approach actually couldn't find a feasible solution in two hours of high performance compute. Whereas the quantum optimizer was able to find a feasible solution in three and a half seconds. So it's not really an indication of quantum supremacy by any means. There's a lot more classical algorithms we could have used that might have done a better job.
But it does demonstrate that using the classical tool that we use, which was a standard tool that was used in production, it wasn't able to find a solution. And we were able to then apply quantum optimization to it and find a solution within seconds. So it is a really good use case where as we make that problem more complex, we can potentially solve that with a quantum optimizer. But again, you don't need quantum for everything. And so like Parimal said, that's where your standard optimization tools are very, very powerful.
Some of them are open source, some of them are off the shelf, and that's where you know just general classical optimization is really the key in in most scenarios.
Parimal Kulkarni (21:13)
I'm still stuck on quantum supremacy. I feel like that's the next podcast title. I like it.
- Choosing the right use cases and application of technologies
-
Helen Fang (21:22)
To that point I know there are a lot of talk about how people are using AI, a topic people are a little bit more familiar with, right? People are trying to apply AI to everything currently, where maybe you don't need the latest GenAI model or to have an AI agent do some very basic topics. So when it comes to kind of quantum optimization and GenAI together, are there any specific scenarios or any general scenarios where it would be a good use case for this? Like further examples? And are there any problems maybe where you shouldn't be trying to use all of this?
Curtis Nybo (22:04)
I think even step away from the quantum optimization side, because again, quantum optimization is just really good for those the most complex type of problems, which most problems you might not need that level of compute power to solve those problems. And so if we take a step away from quantum optimization, just look at our traditional tools that most organizations would be using, traditional optimization tools, that's where you don't it really doesn't utilize a LLM in the sense that that Mariolys touched on that we don't need to be able to…we can't plug the whole problem set into the LLM. We have to use specific optimization tools to do that.
Parimal Kulkarni (22:45)
No, I think you touched on it exactly and it's a question we get a lot, don't we, Curtis? Like where to use quantum, where can I just use regular optimization? And we have a framework for that, but like Curtis said, not everything needs quantum, but an optimization problem makes itself pretty well pretty quickly.
Curtis Nybo (23:09)
And that's a good point too. Not everything also needs AI. We have a lot of clients that are often wanting to apply AI and over apply AI, and they want to apply AI or LLMs to optimization, and that's where, you know, you don't need that level of…it's a different tool for that case. And so there's a distinction to be made, I think, there between when you need an LLM versus when you need an optimization solver.
Mariolys Rivas (23:35)
Yeah, I think what's going to happen is like same thing with AI. When AI came out, everybody was like, I need to implement it because everybody's doing it. It's going to happen like with quantum it is the same thing. People are going to start hearing the word quantum and then they want to say, “Oh, it's cool now, let's implement it.” But it's not necessarily true. You have to understand your use case, understand what the problem is and see if you really need it.
Curtis Nybo (24:07)
Absolutely. Yeah, that's a great point there.
Helen Fang (24:08)
Yeah, that's a great point, Mariolys. So typically, how are you usually approached around these optimization challenges? Or is it usually that you're already in an organization and you see an opportunity for it? And do you have any other examples on how certain industries or certain clients are already seeing some results from this?
Parimal Kulkarni (24:33)
Well, in consulting, the nice thing is people don't bring you a technical problem, right? They bring you a business problem. They're going to say something like, we have supply chain issues, we have orders that are not being met, or I'm in healthcare and we're thinking of wait lists and you know it's hard to manage these. So people always bring us a business problem, right? Which is where we get to sort of analyze it and say, okay, what do you need for this? Is this just, you know, making your data more visible? Is this very descriptive analytics? Are we talking about doing more predictive, prescriptive analytics? Or do we need optimization? Do we need AI?
Do we need quantum?
So most of the time that this makes itself visible, like I'll go back to the fleet optimization one, was they were just trying to understand how to route their trucks and structure their fleets better because there was a lot of cross border traffic and tariffs were kind of messing up, how they were planning their trip sheets and things like that.
So most people will approach us with, most clients will approach us with, I need to predict, you know, I don't know, harmonize the hierarchical tariff codes better or some sort of business problem around inventory and then we decide what kind of solution it needs, whether it's optimization fit or not.
Helen Fang (26:11)
Curtis, Mariolys, anything to add?
Curtis Nybo (26:12)
No, I think that yeah, pretty much hits a nail on the head.
Helen Fang (26:15)
Parimal, all your answers are just totally perfect. Are there any mistakes that you see organizations making when they're going into trying to solve these optimization problems?
Curtis Nybo (26:33)
One for me to start it off is just going back to finding the right tool to do that. Trying to apply a generalized AI model to try and solve these problems, which might not be the best tool to be able to do that. There's a lot of open source tools you can leverage. There's also a lot of off the shelf tools that are proprietary, like Gurobi, that are very specialized solvers and can solve very, very complex problems.
But the main problem I think, as Parimal touched on, it's really just identifying what is a really good optimization problem. So if you have a problem where there's a bottleneck in a process, something's taking too long, we're trying to, you know, you're using too much fuel in some context, you can optimize all the variables around that to basically minimize some sort of function or maximize some sort of function. That's really what optimization is all about. And I think the identification of that can be a little tricky and that's where we start from what is the actual business problem and what value are you trying to get out of it. And then we can formulate, you know, what variables and constraints do we need to take into account to be able to get there. And then how we can actually solve the problem to find a feasible solution or a better solution that you might have had, might not have had .
Parimal Kulkarni (27:49)
I'm going to add two points to what Curtis said, and one of them is actually what Mariolys said earlier. One of the mistakes I think I see around optimization is when we bring this forth as a potential solution, people think they need a lot of data. They don't need a lot of data. What they need is an understanding of the business variables and parameters.
Parimal Kulkarni (28:12)
And that leads to the second point is that I see businesses maybe not being able to explain their risk posture. Because risk is what essentially is translated to a constraint in optimization. So what is it that we can tolerate if the business doesn't meet a certain goal or you know, we don't get a delivery through?
It’s the thing that shows up on the P&L. It shows up on your bottom line and businesses can't articulate that. And then you cannot build a good optimization model and then the AI and the quantum part kind of becomes redundant. So those are two mistakes, I won't say mistakes, but those are two levers I would say people need to understand is you don't need a lot of data, you need a really, really good understanding of your decision of your decision flow in your business, of the parameters there, and you need to be able to articulate the risk in your business.
Curtis Nybo (29:19)
Yeah, I think that's a really, really good way to put the overall underlying benefits analysis, trying to see where those benefits are and how it affects your risk profile for that specific problem.
Mariolys Rivas (31:53)
I can add specific good fits for optimization models, like samples. We talk about routing, scheduling. There are also topics such as network design, production, sequencing, planning, inventory positioning. All these examples are good fits for optimization models.
Anything, and we have said this a lot, but anything where you're making lots of interlocking decisions under constraints, where bad fits, like pure forecasting or image recognition sentiment analysis, these are examples that are related more for AI machine learning problems rather than optimization models.
Anything that is related to questions like what is it and what will happen, those are problems more related to AI machine learning. And for optimization, you want to answer questions such as, what should I do? Those are the problems that are more related to optimization models.
- Advice for leaders and closing thoughts
-
Helen Fang (30:46)
That makes a lot of sense. So it sounds like the most complex questions that might not have a very straightforward or immediate answer with lots of variables, things that people would have a lot of trouble wrapping their heads around, let alone some of the classical approaches. Those are kind of the best scenarios or problems to tackle with this quantum, optimization, GenAI approach.
I guess on a practical level, is there anything concrete that every executive should take away in the form of action, in the form of maybe understanding things better today to best take advantage of the opportunity or to be ready for what's to come?
Parimal Kulkarni (31:34)
One thing I would suggest is if executives are asking these questions, right, they are trying to essentially in their mind plan and do a sort of a simulation in their head of what's going to happen next, because we all now know like, you know, there is no one best plan. It's more about if the plan doesn't work what do I need to do? Like if my forecasts are starting to diverge from reality, what is it that I need to do? And that's where the executives are sort of thinking about this. And they may not call it optimization, right? They might not call it AI-based intelligent optimization or quantum-powered optimization, whatever we want to call it.
So I think the one concrete step is if you're asking that question as an executive, you want to enable your organization to be able to give you the parameters and give them the opportunity to have your whole business sort of mapped out. It kind of goes to the very foundational work of do you understand all the decisions, all the parameters, all the constraints? Do you understand what buffers are there? What can you fund? What penalties are you okay to incur if your forecasts don't work or if your plan doesn't work? And I think that needs a very holistic understanding of the business. And to me, that is a foundational first step.
Is it you know, if you want to get value out of optimization, it's not so much about what models do we have or what tool we brought on, it's whether you have we done the hard work of defining a decision clearly enough for somebody to sit down and model it? And I think that is the piece that I would ask executives to enable.
Curtis Nybo (33:36)
And I would add, especially for the quantum optimization side, but it goes for both classical and quantum optimization, one concrete step is just building that literacy within your organization. So taking the tips that Parimal provided and spreading that around your organization for to help educate for how to understand your business, what level of detail you need to understand your business, what tools are available, how you actually implement optimization.
And building that literacy within the organization at every level can really help to build that understanding of where it can apply. And then once that's understood, it makes it much easier to be able to identify those areas and be able to identify those benefits that you might get if you could just get a 5% decrease in fuel consumption or a better, you might get better workforce satisfaction if you can schedule everybody a little bit better, but take into account another feature.
So when we're doing a lot of optimization, a lot of optimization use cases, you often have to make a lot of assumptions and being able to understand which assumptions to make and which ones you shouldn't just make an assumption for, and you should include the actual metrics, that's pretty key. And I think a lot of that comes back to just building that literacy within your organization and understanding where it can be applied, what tools to apply, and how to eventually go about applying it, and then how to use maybe an AI tool like Mariolys talked about to be able to interpret the results of that and plug it into a dashboard or some existing tool that you already have. So being able to optimize but also make use of those results as well in the end.
Helen Fang (35:19)
I think this has been a really interesting conversation. And I think you guys have spoken to both the technology and skill sets and what needs to be in place, but also the importance of making sure your organization has thought through the tough problems and you've enabled your people and your leaders to be able to make the best use of what's possible. So I hope that this will just be the start of many conversations to come.
I think there are a lot of additional aspects that would be quite interesting, like how do you work within an ecosystem, what are specific capabilities your organization should have compared to maybe your consulting partner plus maybe like a quantum provider or anything like that in the industry. So Parimal, Curtis, Mariolys, thanks so much for your time today. And I hope you guys will join us for a future episode on the podcast.
