JOpt is a flexible routing optimization engine designed to solve complex tour optimization problems with constraints like time windows and required skills. It can be integrated into logistics and transportation management software or standard ERP systems, either natively in Java or as a REST service.
In this episode, we speak with Jens Richter, managing director and chief software architect at DNA Evolutions. We discuss the company’s flagship product, JOpt, which specializes in logistics optimization using advanced AI algorithms. Jens explains the evolution of JOpt since its inception in 2005, the integration of AI technologies, and how the software helps companies manage complex logistics challenges. The conversation also highlights JOpt’s unique features, flexibility for businesses of all sizes, and real-world impacts on logistics operations. Jens discusses the complexities of urban logistics, the innovative solutions they provide for delivery optimization, and the importance of sustainability in logistics. He highlights the role of AI in enhancing efficiency and the significance of strategic partnerships in the logistics sector. The discussion also touches on the future of automation in logistics and how their software adapts to changing scenarios.
Watch the podcast here:
Listen to audio only here:
Learn more about DNA Evolutions.
Interested in appearing on the SourceForge Podcast? Contact us here.
Show Notes
Takeaways
- DNA Evolutions focuses on solving highly constrained logistical challenges.
- JOpt has evolved from a basic Java library to a robust REST API system.
- AI components have been integral to JOpt since its inception.
- JOpt uses a variety of sophisticated algorithms for optimization.
- The software can achieve significant cost savings for customers.
- JOpt is designed to be flexible and adaptable for various business needs.
- Integration capabilities allow for easy plug-and-play with existing systems.
- The onboarding process for JOpt is quick and user-friendly.
- Real-world examples demonstrate JOpt’s effectiveness in complex environments.
- Customer feedback drives the development of unique features in JOpt. Urban logistics face various challenges, including congestion and legal restrictions.
- Innovative solutions can optimize delivery routes by penalizing unnecessary crossings.
- Sustainability is a key focus, with optimization leading to reduced environmental impact.
- AI can enhance user experience by providing assistance in setting parameters.
- Strategic partnerships are crucial for customer success and feature development.
- Automation is essential for real-time decision-making in logistics.
- The future of logistics includes exploring quantum computing for optimization.
- Customer feedback is vital for developing effective optimization tools.
- JOpt aims to be a strategic partner rather than just a software provider.
- The logistics industry is rapidly evolving with new technologies and solutions.
Chapters
00:00 – Introduction to DNA Evolutions and JOpt
02:57 – Evolution of JOpt and Technological Shifts
06:01 – AI Integration in JOpt
09:01 – Optimizing Logistics with JOpt
11:57 – Unique Features of JOpt
14:59 – Flexibility for Different Business Sizes
18:00 – Integration Capabilities of JOpt
21:07 – Real-World Impact of JOpt
26:56 – Navigating Challenges in Urban Logistics
32:11 – Innovative Solutions for Delivery Optimization
34:08 – Sustainability in Logistics
35:50 – Harnessing AI for Enhanced Efficiency
43:15 – The Role of Strategic Partnerships in Logistics
45:11 – The Future of Automation in Logistics
Transcript
Beau Hamilton (00:05)
Hello everyone and welcome to the SourceForge Podcast. Thank you for joining us today. I am your host, Beau Hamilton, Senior Editor and Multimedia Producer here at SourceForge, the world’s most visited software comparison site where B2B software buyers compare and find business software solutions. Today we’re joined by Jens Richter, Managing Director and Chief Software Architect at DNA Evolutions, a leading provider of optimization software components designed to really enhance logistics in transportation management systems.
In other words, they help companies transport goods from point A to point B in the quickest, most efficient way possible. And a testament to their capabilities, the fact that their software is being used by industry giants like Amazon Fresh and Saudi or Ramco, among others. Our conversation will focus largely around their flagship product, JOpt, which utilizes cutting edge artificial intelligence algorithms to make these transportation optimizations possible.
So to talk more about the company and this product, let me introduce Jens Richter. Jens, welcome to the podcast. Glad you could join us.
Jens Richter (01:03)
Yes, hello. Hello and welcome. Thanks for having me. And yeah, so what do you want to know?
Beau Hamilton (01:09)
Yeah. Well, it’s I just want to get started and just ask you, yeah, if you could just tell us about yourself and what it is that DNA Evolutions does.
Jens Richter (01:17)
Yes, so actually you always gave a good introduction and so my name is Jens as you already said and I’m the technical director. So everything that is related to technical questions essentially I have to deal with it. And so DNA Evolutions essentially is a German company. So it was founded in 2005. So we are around for quite a while.
And so our main goal is to solve highly constrained logistical challenges, right? It’s called the vehicle routing problem. So you have some customers, you have some drivers or technicians, and then you send them around and you have to find a good way to send them around to schedule them. It’s called scheduling optimization.
And our flagship product called what is called JOpt helps companies to manage everything from route and tour optimization, to fleet management and workforce dispatching. So we can solve essentially every variety of the vehicle routing problem. And for this, we are trying to use cutting-edge technologies. And therefore, I think we can say we can ensure you that we have quite a cost-effective and nice product here.
Beau Hamilton (02:29)
Very cool. Okay. So you’ve been in business since 2005. And I guess that, at that time, you know, the internet was sort of in infancy. It was gaining adoption. You sort of, there’s the transition to the cloud, which was happening and really kind of growing in prevalence. And, you know, I would say the peak was in 2008.
So you’ve been through some of these like technological challenges or challenges or just shifts in technology. I guess, you mentioned sort of the inspiration for creating JOpt, but I’m curious just how it’s evolved over the years since 2005.
Jens Richter (03:03)
So essentially, yeah, it was addressed because logistics became more complex and complex and complex. And so the original inspiration was to have a tool to do the scheduling optimization. So the idea was not around cloud or whatever. It was just to having a tool to handle this raising demand.
And so we started with a basic native Java library, and that is also still in use. So you can still use it. And it’s essentially the backbone of our product. And in addition, we had a cross-compiled C# version. But see JOpt more like algorithmic dependency. So it’s not like an end customer product where you have a GUI or graphical user interface. It’s more like a dependency in your system that you can easily add.
And so later on, as you correctly said, there was a big transition to the cloud and more and more cloud services. We also added a really robust, standardized REST API. And so nowadays, you can, it’s totally flexible. So from a normal, like a dependency library that you run locally or on your server, now you have a really flexible deployment way. So you can run it locally, you can run it in the cloud, you can run it even on your laptop, if you like, but also on a big Kubernetes cluster.
And what we essentially also allow is our users to host that in their own infrastructure, even the REST API server-side technology. So meaning you can make JOpt nowadays really your product. And so I think that’s the transition from a native Java library, that is still intact, to a big REST API system that you can use in the cloud or in a Kubernetes cluster for example.
Beau Hamilton (05:05)
What about specifically with the artificial intelligence components and technology with your product? Because I know that this is something that’s been, it existed for a number of years with different terms and different iterations, right? Now we just refer to it as artificial intelligence. And we think like ChatGPT and Gemini and all these different chatbots. But I’m curious, like, were you always utilizing these sort of automations in your product or are these fairly new?
Jens Richter (05:35)
So essentially, JOpt is from the beginning on, it is using a lot of, without going too much into detail, it’s using some more or less random algorithms. But those randomness is the backbone of a lot of AI driven technology. And we already utilizing a lot of self-learning technologies.
And so nowadays you can really say that JOpt itself is and was AI driven, right? So we, because it is self, it is self-learning, it is tuning itself, so you have a lot of those AI components.
And, but also in addition, and maybe we can take, talk about that also later, so we also have some recent developments when it comes really to the classical large language models. So what people nowadays really call or what the average person would call an artificial intelligence is also something we are looking at, but more with the focus of usability and making JOpt even more user-friendly. Yeah, but to give a short answer in the end, so I would say that JOpt definitely has a lot of AI components and not, and this since the beginning on, yes.
Beau Hamilton (06:52)
Right, okay, yeah, seems like, I mean, the AI that I think of, that most of us are probably thinking of nowadays, seems like a perfect fit for your company, just sort of, yeah, trying to make efficiency, just make companies more efficient, transportation more efficient. This sector is all about that, really.
And I’m gonna talk more about this later in the interview here, but I wanna drill into the, the topic of just like how you specifically help companies manage these tricky logistics issues. And I think of like, there’s tour optimizations, have driver schedules, you have customer preferences. How do you, how does JOpt just help these companies handle all this, like these moving pieces?
Jens Richter (07:36)
Yes, yes. So in principle, it’s not one algorithm. So we’re using really a bunch of custom, really sophisticated, handcrafted algorithms that we have developed over the years. So we have, example, again, to give an example, we have something called objectives where you set a goal for your optimization. Let’s say, for example, minimal drive time. You want to improve the food consumption, not yeah so less food of course, and so here we use all those AI self-learning technologies.
But we also use some pretty standardized ideas like simulated annealing. We use genetic algorithms. So just in a few words, simulated annealing is borrowed from physics, where you have some cooling mechanisms to find optimal solutions and genetic algorithms where you really have some individuals that play against each other.
And so all in all, if you put all these algorithms together, and one word here, our customers doesn’t, do not necessarily need to understand all of that because that’s why they buy JOpt, right? And but essentially what our algorithms are doing is considering time windows, as I said, food consumption, skills, a lot of different preferences. And I would say that’s quite competitive because we do that in a really highly constrained environment, right? So it’s not like a generic thing. It’s really highly constrained what we’re usually looking at. Like, almost all our customers are utilizing a lot of skills, preferences, and right person, right job techniques, and also skills of different levels and so on.
So it can become quite easily really complicated, really fast. And I think our algorithms, the AI and also the modified algorithms and clustering algorithms, they have a lot to find a really good solution. And I would say we have quite happy customers and they have a quite cost efficient system yeah? And yeah, there were some numbers that I read from one of our customers that they reported savings up to 30%, right? When they were using scheduling optimization.
Of course, it’s from their commercial page and what they reported. So but, so we can see that, definitely. So just to give you a bike number, it has quite a big impact, yeah.
Beau Hamilton (10:15)
Yeah, well, I want to speak on the competitive nature of this landscape because there are more companies than ever trying to get sort of a bigger piece of the pie in this particular sector. So maybe, yeah, can you talk about what makes JOpt really different and stand out from the other logistics software on the market?
Jens Richter (10:31)
Of course, that’s an important question, right? And so I think there are two main points for me here. So first of all, it’s, we are really focusing that it’s highly flexible and adaptable. So there are no boundaries in the way you run it. You can run it, as I said, in your local environment. We do not force you to connect to any server infrastructure from us. So you can also use it in high security environments.
Then, of course, you can, as we have REST API that is standardized, it’s your open API specification, you can essentially use every language that you want, right? And honestly, these are things that also competitors can do, right?
But there’s also second part. So instead of providing like a generic solution, we try to incorporate as much as possible from the business logic provided by our customers into the solution process. So for example, we designing custom constraints. So a customer can really precisely define what kind of constraint the customer wants in an optimization. Just as a side note, a customer once asked me and asked me, what can you do? And I said, if you like, we can visit each city that starts with an A on a Monday. That’s totally worthless, yeah? But it shows that we can, in principle, define whatever we want. And there are even interfaces where customers can define their own restrictions.
And one thing I’m really proud of is some of our hard constraints. yeah? So hard constraints means something that cannot be broken in an optimization, like an appointment that you really have to participate at that time. So do or die, either you do it or you skip it. And we managed to incorporate that in the architectural, architecturally in JOpt. So meaning it’s not a cost optimized thing, it’s really something architectural.
And we have a lot of those little points that are of those features. We have also on our website, you can find a dedicated page like Unique Features. And so I would say that it comes from the fact that almost all our features are customer inspired. So usually there was a problem and we had to solve it with our customers together. So that’s usually the path we want to go.
And so lately you were talking about artificial intelligence and we are really trying with one of our customers to do, to go on that path. For example, one application they like is that we take a result and we analyze the result with AI. So essentially, really now large language models. And the result then to get a deeper understanding and even to get some what if scenarios. So for example, hey, on that day there is a problem with that technician, right? So, or please hire another technician or those companies you should rework contracts and all this stuff, right? So those are some points. I would say that, to my best knowledge of course, makes JOpt quite unique, and of course, the team behind JOpt. So we are easily reachable, and we definitely help our customers. And if you don’t share that, we also, we are not always writing an invoice for every question we answer.
Beau Hamilton (14:07)
Okay. Well, yeah, there’s a lot of great things to like about, about what you just said there. I think what stands out to me is, yeah, the fact that you kind of cater your solution based off the customer’s needs. I think that’s huge. That’s crucial, right? And then also just the flexibility aspect, like the fact that you work with, you know, big fleets, you I mentioned the Amazon Fresh and Saudi and Ramco.
And then, you know, I imagine you also work with a lot of much smaller operations. I almost think of like a moving, moving company where it’s like me and some roommates are like two brothers moving, you know, or you have like a few trucks. We’re trying to get logistics sorted out. I imagine you could be helpful for them as well. So can maybe you can talk about how do you work to stay flexible for companies of all different sizes?
Jens Richter (14:51)
Yeah, actually, so first of all, indeed, we have customers in both worlds here. So some customers run like thousands of optimizations per day with even hundreds and thousands of appointments, whereas others only a few, right? So essentially, we have those both worlds and we are, we can, I can tell you before even telling you why it’s working, it is working, okay?
And so, the way that JOpt is designed, it’s quite a modular system. And you can, I think I already mentioned that you can run it on a Kubernetes cluster, right? So high performance computer. And of course you can even run it on a small laptop, yeah? And as we have in standardized API, you can essentially also interface it with the exact language that you need.
And so we believe that because of this API-based attempt and then this modular architecture, you can essentially easily scale it. And this is really what we also see. So people are using it for all different kinds of problem sizes. So even really a small problem size. So some customers do only a one-day optimization. So that could be like thousands of jobs with just a few nodes. So we call it nodes to appointments, and others are trying to crunch optimizations with a horizon of almost a year, yeah? And both works, yeah.
Beau Hamilton (16:30)
Hmm. Okay. Yeah. I think that’s really cool how you can run this algorithm and your software on a bunch of different types of equipment. You don’t need like a particularly advanced piece of hardware to run this. I think of the different AI models that, and large language models, that are sort customer facing and are out in the marketplace nowadays, whether it’s like Llama or there’s a new company that just launched one that’s through a mobile app that you could run as a smartphone app, it’s called PIN AI, shout out to them. It’s just, yeah, it’s fascinating to see this world where you can run these really powerful solutions, results just from the current technology we all have, whether it’s a laptop or a smartphone or tablet.
I wanna sort of segue into the integration capabilities, because I think that’s one area that really makes your solution stand out. So maybe you could explain for us just how easy it is to plug and play JOpt into these existing systems and why that really matters.
Jens Richter (17:33)
Yeah, so from the beginning on, it was designed for easy integration. So first of all, we wanted to create a product that you can start and you get something meaningful out. So for most of the customers, indeed, it works out of the box with the settings, with our default settings.
And so meaning the onboarding is really fast, yeah? So you can more or less put it into your product and use it. And you get something meaningful. And so we even have a sandbox. And you can start, for example, JOpt and all our examples inside your browser. And you can play with it there. So and later on, of course, then comes the fine tuning. So and exploring a lot of the features because our customers need to define their business case. So they have all different business cases. want to have different, they need different settings in the end, or they want to fine tune it.
And for this, of course, we also offer, for example, assisted trails. We offer meetings and all other kind of help here. So I think this way, most of the customers have a pretty good return of invest here. So because you can start easy, you can start fast, and you can even start anonymous, right? So you can just now download it. You can only solve small problems, but you can play with it. You can explore it. And if you find out, that’s a cool thing, then you can even contact us. We help you setting it up, right? And then you can see if this is the right fit for your product. But I would say that it’s easy to start with it, right?.
Beau Hamilton (19:25)
I love that. Yeah, that makes it a lot more user friendly and less intimidating. What? I want to talk about some of the specific examples where you’ve helped make an impact on the company. I mentioned some of the maybe industry giants you’ve worked with in the past and some other companies that might be a good fit, but maybe you can share some other examples where JOpt’s capabilities have just made a really significant impact on the business.
Jens Richter (19:49)
Yeah, so I can share essentially a story that happened last December. So it was quite recent, I would say, right? And I think it shows really how advanced our features are or how fast our development cycles are. So there was a customer that approached me and told me, hey, Jens, we, so it’s customer that is selling a system to schedule, do pick up and delivery. So essentially supplying groceries with goods, yeah? So you have trucks, you put the groceries in there and then you drive to the stores and you give it to them. And so the challenge was one of their customers, right? One of the end customers, had grocery stores in New York City. It’s also a big chain that you probably know. And so we had to meet time windows, we had to meet the preferences, we had to do all this right person, right job, everything that we offer.
And so then the customer that approached me, we know it’s quite a long time already, and he approached me and told me, hey Jens, this works perfectly in the middle of the US, but that does not work in a dense area of New York City. And I asked him why. And he told me yeah, you know, New York City has rivers, yeah? You have the Hudson, you have the East River. And so our end customer was complaining because they understood that the solution looks good if you ignore the rivers. And he told me it’s not acceptable for the drivers to pass the East River and the Hudson multiple times a day, right? Because see from that perspective, if you have one for the optimizer, if you have a node on one side or the other side, it just takes the bridge if it makes sense from a distance perspective. And b ut it does not make really a sense here. So my customer asked me, hey, what can we do? You know?
Beau Hamilton (21:55)
Yeah, that does sound challenging. What, what, like what particular issue is it because of the bridge? it, or is it because of like the bridges might be like up for, for boats to pass through, or is it just because it’s a different part of town? It’s you have to travel a little bit further?
Jens Richter (22:10)
It can be different things. It can be just like because of the way or you just simply not allowed to use sometimes you have high traffic times. It could be a personal preference of the driver, right? Because of the driver simply does not want because the roads are a little bit smaller, right? On a bridge, yeah, and so there are a lot of reasons.
So of course, there could be legal reasons. There could be personal preferences. There could be all different kinds of reasons why the bridge crossing is not desired. And yeah, so that was the reason essentially they shared with me. And so yeah, we had a meeting, I think a day after, with all the development team from them, and we thought about how we can solve this. And of course, it’s not like we want to forbid the crossing because you should go there in the morning and come back in the evening, right? It’s not a good solution if you tell them you are just not allowed to cross the bridge, right? It should be just not attractive to cross the bridge, yeah?
And how can you convince an optimizer that something is not attractive? You put a cost on it, right? So you tell the optimizer this solution is worse than the other because by crossing the bridge, you gain a cost and if you do that, you acquire a cost. And if you do that too often, you get more and more costs and the solution is not good anymore, ok? And so our idea we came up with is now imagine you have something like territories, right? So what is quite easy to imagine, right? On one side of the river, you have one territory and on the other side of the river, you have the other territory. And now you put each node into such a territory. So it’s as easy as giving them a number. So one is in territory one, the other one is in territory two.
And now you simply define a new feature that we crafted that we penalize a territory crossing. So if you go from one territory to the other, it’s generating a cost. If you stay in the same territory, it’s not generating a cost, ok? And this feature was able really to reduce the number of travels to the minimum that was really desired. And I think, and at same time, we also were able to still get a nice time window and distance solution. So meaning we’re still hitting all the time windows, we still had an optimization result that we’re focused on, for example, so distance, preferences, and so on. So the restriction just comes on top, right? And so I think that was a pretty good outcome of the whole thing.
Beau Hamilton (24:54)
Yeah. Okay. So you have, that’s a really interesting description and breakdown there. So you have these different zones and your software kind of like calibrates the, the, how efficient you can get to these specific particular zones. I’m curious, like what is this? Like, was this a custom tailored approached for this geographic area or does the software sort of handle this automatically? Essentially. I mean, I imagine you’re constantly, you’re obviously having, you know, developers and technical hands on, workers kind of manipulate the software accordingly, but is this something that can be translated to other geographic areas pretty easily?
Jens Richter (25:30)
Yes. So essentially, you can even assign one node to multiple geographical areas. That makes, for example, sense if you also, so the feature you’re talking about is essentially our territorial feature that we also essentially utilized here. So it’s a follow-up feature of the territorial feature.
And so, for example, you can also add multiple, one node to multiple areas. That makes sense if a certain resource is only allowed to visit a certain area, a territory. That is also possible, right? So you can easily define those territory and zones the way you want, yeah? And we had one customer in the past that was utilizing such, this tool by using postal codes. Because it makes sense, right? Because poster codes usually are a territory definition in itself.
Beau Hamilton (26:31)
Yeah, and they’re more they’re more narrowed specific than obviously like a city or county or they’re more like on a neighborhood level, essentially.
Jens Richter (26:39)
Yes, exactly, exactly. Yeah. And even the bridge feature could work with it because usually you could also have two different post codes on each side of the bridge here. What was not the case here, to be honest, but that was also something I proposed.
Beau Hamilton (26:53)
So how, so maybe you could talk about how some of these solutions translate into these real world results for the company.
Jens Richter (27:01)
Yeah, so essentially for that company, it was pretty important to solve that issue because it was a new customer of them and they wanted to get that customer. As I said, it was a pretty big grocery chain and in the end, we are not working for free, right? So it’s like they wanted to get that customer and that was one of the key points why in the end that customer signed the contract and they were able to, to have that there. And so, and yeah, they were really, really friendly and told me, hey, Jens, that we’re great and they were really, really happy.
Beau Hamilton (27:44)
Nice, that’s awesome. Yeah. And it kind of goes back with the competitive nature. Like obviously there’s other companies you’re competing with. So having this feature that really stands out as attractive, this is a good sort of case point. Yeah, just like you had this unique feature that really showed a lot of promise and real tangible assets. And that’s what kind of got them to sign on. So I love it.
And just generally speaking, I love like how, to hear about how technology like JOpt is, is tackling some of these real world problems, especially problems that I never really knew existed in the first place. As an end user who just wants to make sure my package arrives at my door on time, I never really appreciated or really recognized just how many variables are kind of involved to make that delivery possible. So this is great. This is really insightful.
And then also just from being from the Pacific Northwest surrounded by nature, I love the fact that, you know, this has a really sustainable aspect because the efficiency gains for businesses, transporting goods really has a direct impact on the environment. So I think that’s really important to stress. And I think it really shines through with the story you provided. So that’s great.
Jens Richter (28:58)
Yeah, so we’re also focused on that. For example, you can even set optimization goals to, if you have to go longer distances with the truck or something like that, you can say, OK, try to not be too full and all this stuff. So to say full, right? So you can really try to to become sustainable and to to to do something green, right? Because we cannot completely avoid it, right? But we can definitely improve it here. And there are a lot of things where you can read in the literature how, what optimization can save you in terms of distance. And if you save distance, usually you also save food here and do something green, yeah, that’s true.
Beau Hamilton (29:42)
Yeah. Well, yeah. And then just highlighting, just kind of highlighting the, the environmental aspect, the environmental impact of the options you have. Like I think of my national, my navigational apps and how, like a lot of times when I plug in a destination, it’ll say, you can get there faster, like a minute faster if you take this route, but you could be a little more environmentally friendly. If you take this, this side route, or when you’re like booking a flight or for example, you can see all the emissions that are result from this one flight.
So I think things like that, that really makes a lot of sense. kind of helps to kind of get us in the right direction of thinking about these, these impacts and sort of side effects.
Jens Richter (30:22)
Yeah, it comes for free, right? It comes for free. You get that without improving the engine, without doing something. You just get it as you just optimize what you’re doing, right? And yeah, that’s good point.
Beau Hamilton (30:36)
Now looking ahead, are there any new updates or features in the pipeline for JOpt that you’re particularly excited about?
Jens Richter (30:44)
Yeah, so coming back once again to some artificial intelligence. Essentially, we had some kickoff weeks some time ago, and we thought, what could be meaningful ways of really utilizing, for example, large language models? And we came up with that because see, a large language model, no matter what other people want to try to sell me, is not solving a true optimization problem because it’s just simply there are too many variables, it’s not designed for that, right? Maybe one day it can program a good solver. I’m not questioning that, right? So we thought what are good things we can do with artificial intelligence, especially with large language models.
And so we thought, first of all, we want to provide some assistant, yeah? So an artificial assistant that can more or less help you setting up some parameters. As I said, usually it works with default settings, but the more and more it gets complicated or you want to hit your own end customer requirements, it makes sense to have a helping hand, right? And of course, such a system could analyze what have been parameters before. What are parameters the end customers are especially happy with? So some to identify even some patterns, yeah?
And so such an input helping AI could make JOpt a little bit more user-friendly. I’m not seeing this to go to the end customer directly, the AI assistant, but maybe could be even intermediate layer for our customers. Because as I said, JOpt is a library and is put into another bigger system. And but, so what else? So what is really promising is also analyzing the outcome.
So for example, you have some historical data. You know, you had comparable problems, optimization problems, and you got out, you got a result, and you know what could customers complain about? Because customer optimization sometimes is a nasty thing, because if I show you a result, you can immediately tell me what you don’t like. But if I just give you the problem, you are totally lost. You cannot come up with a good solution.
But if I show you the solution, you can, usually you tell me, why is that? Why is that? And so there, we want to help our customers. So analyzing the result, understanding the result more or less better, and also communicating that back to the customers. And so here, of course, AI could help. So because it could use historical data to better understand the end customer and even suggest some actionable strategies here, like, for example, adding another resource or another technician or driver, and or, for example, adjusting contracts with nodes, right? Because you simply see that one of your end customers is a little bit too cheap or too expensive, right?
And these are all things you could do in terms of deep analysis from a result with helping and using also AI for that. And maybe even doing some intermediate things like analyzing the, so during the optimization run itself, from time to time checking if everything is in a good shape, yeah? And so these are things we are recently working quite a lot on, and especially also this what-if scenarios.
So to better help our customers to sell the solution to their end customers. Because that’s, to be honest, usually, and everybody in scheduling optimization will agree on that, that’s usually the hardest thing, to sell a person the solution and show it that’s really the best solution.
Beau Hamilton (34:48)
Yeah, very cool. I think there’s a lot of exciting things that, we talked about. I would just kept thinking about, you the way you’re integrating seamlessly with other existing APIs that businesses have, I think that’s a crucial point. And then also, I can just imagine if you’re able to have your product, your algorithm tie in with existing LLMs, so that it can kind of tap into the features that your specific algorithm offers. I think that would be huge and pretty, pretty, yeah, just like, it could allow for a lot of different possibilities.
But I’m just kind of spitballing here. I’m just thinking, you know, cause a lot of times there’s different APIs or LLMs that companies integrate, whether it’s in-house or, you know, they’re working with a third party. So being able to integrate with that would be, I imagine, be very, very useful.
Jens Richter (35:36)
Yeah, actually, the integration, of course, we were experimenting with a lot of different LLMs. So we’re not coding our own large language model, but we are utilizing it. We are training it. We even published some of our training questions, right? So you can even train your own AI. It’s in a standardized format. So you can download it from our website and even train it on your site if you want, yeah? So we are totally open to that.
And so for us, of course, could, yeah, hosting a large language model on your phone is quite challenging. But that could be a third-party service that we then provide. That you can, but we will also show you how to set it up in your own infrastructure, right?
Beau Hamilton (36:26)
Well, and also it’s very resource intensive to develop your own LLM too. So that makes sense that you’re kind of utilizing what that’s already in development.
Jens Richter (36:36)
Yes, yes, yes. So we also, as you already mentioned, like some Llamas things from Meta or also there are also a bunch of other AIs around that you can utilize. And also famous ChatGPT is giving you a lot of ways to train it, right? So and so on and so on. So we are not just mentioning those two without, yeah, there are a lot of other ones, of course.
But honestly, they are doing a lot of stuff for clustering it and so on. So it becomes less resource intensive, right? So there have been great publications around that topic, especially in the recent, yeah, almost days, right? So it’s really fast. It’s moving unbelievably fast. And honestly, we are not only providing the stuff, you’re also using AI for our own every day, every day, you know?
Beau Hamilton (37:35)
So okay, now if I were a company working with you, maybe listening to this podcast, maybe feeling slightly overwhelmed with all these different options and features and integrations you have, what kind of support would you offer to help take advantage of all these things you mentioned and just really maximize my investment?
Jens Richter (37:52)
Yeah, so first of all, it’s a personal thing maybe also about me and but also about DNA Evolutions. So we see ourselves not only as a provider of software, right? We are, but we see us more as a strategic partner. And I really believe, honestly, it sounds really pathetic, but our customer success is also our success, yeah?
And I’m totally open to have some meetings, to talk to the people, to learn something. And as I said, almost all our features are customer inspired, right? So telling you that you’re seeing us as a strategic partner, I think is totally fair. And it’s really satisfying if a customer calls you and tells you, hey, Jens, we have solved this and it’s working really well here. And also the fact that if you imagine like thousands of optimizations a day, a single customer, right? So how many people are getting affected by JOpt?
And honestly, I strongly believe that you also already got affected by JOpt without knowing it, right? Because a lot of our customers are in the US. And so, and I would say you can see us as a partner and we really try to be close to our customers. What is a little bit easier because we’re not dealing with end customers, we are usually dealing with, with customers on a level like smaller companies, of course, but development companies. And yeah, so usually we are really a strategic partner and try to help you with any question you have here. And we have a lot of different channels also for that where you can reach us.
Beau Hamilton (39:38)
Fantastic. Yeah, I love it. I love the, the human connection there, especially when we talk about all these AI automations. It’s like at the end of the day, you’re still having, you know, human conversations. You’re really working with other people to help, figure out the best solution for them, yeah.
How do you, how do you see like automation just changing logistics and the role that JOpt plays in that future? Where do you see this? Like where do you see everything going, you know?
Jens Richter (40:03)
Yeah, so I think it’s quite obvious that automation will continue to eliminate inefficiencies and also enable real-time decision making. So automation, we’re not going back. And I think JOpt is, and we’re really proud of that, it’s really on the forefront of this shift. So we are offering solutions that can also adapt dynamically, right?
So, to changing scenarios, so could be as easy as somebody is calling you and telling you, hey, I’m ill. It sounds quite normal, right? Because everybody has done that. But just imagine you have a big problem set up and now somebody is calling in and you have a company with 10 people scheduling the whole day. And suddenly somebody is calling in and says, oh, I’m ill. Yeah, that could be really a problem. And with, of course, with automation, it’s like clicking and it’s done. And that’s one easy example.
But I think for us, it’s important that we are really open to new technologies. So it’s not only AI. So we also have done some proof of concepts for quantum computing. So we really ran JOpt on a quantum computer. And that was really amazing, just pressing this button that it ran on a quantum computer. I love that because I’m a physicist. Also, I studied astrophysics. And that is why, of course, for me, it was really a personal thing.
And yeah, I would say it’s really exciting to be slightly ahead of this curve. And also, pushing boundaries here. And then, of course, you understand what can be maybe possible, especially in logistics and dispatching in general.
Beau Hamilton (41:52)
Yeah. Okay. Yeah. Yeah. How do that physicist, physics background. I think I could just, and I could hear it in your voice, you just get really excited about this topic. And I imagine there’s all these different, like, yeah, like seeing all the different sci-fi movies that we’ve seen, sort of predict the future is just so fascinating to us to think about. I feel like you probably, you know, stay up late at night just, just ponder all these different outcomes and solutions, right?
Jens Richter (42:19)
Yeah, yeah, so I really like what I’m doing, honestly. So it’s, I’m not joking.
Beau Hamilton (42:25)
Yeah, no, it shines clear through this interview, I think. And just, yeah, trying to predict all these different problems that arise from specifically with this, transportation industry, talking about like the congestion on bridges or, yeah, if people get sick, call in sick last minute, or I just think of like, you maybe a driver gets in an accident, God forbid, you know, and have to have to all these outcomes accounted for. It’s, there’s a lot of moving pieces and it’s good to have, it’s good to hear that you’re, you know, somebody’s thinking about these.
Jens Richter (43:00)
We could do a second interview just on the topic “What can happen?”. It’s really interesting. That’s why it’s so important to work with your customers. Because if you’re just designing such an optimization library without talking to a customer, there’s no chance that it will be usable in the end.
So it must be a storyline and. You believe me, I had so many plans. I thought, this is a great idea. And later on, it turned out, no, there’s something different, this and that. So we tuned this library for almost 20 years now, right? And yeah, with the help of our customers. So maybe that’s why I like talking to them. They give us a good input on what we should do, yeah.
Beau Hamilton (43:48)
Yeah, no, makes sense. Well, we’re coming down to our last couple of questions, Jens. So I want to give you one more opportunity to just hammer home one final point to those listening. Is there one like really big thing you wish potential customers do about JOpt? And if so, like what would it be? Could you explain that?
Jens Richter (44:10)
Yeah, so first of all, they should remember that when you join us as a strategic partner, we really would care about you, right? We will help you to solve your logistic problem and your dispatching problem. And you can start easy, you can just try it. And whenever you have a problem, our team, my team will help you with that here.
And of course, so we highly appreciate every customers because our features are indeed customer inspired, yeah? So I think when you, and of course you should remember, you can just try it, just try it. If you don’t like it, forget about it, yeah? But I promise you will, you will like it, yeah? So, so that is why you can, you can simply, you should start, you can simply start and try if you have your, your business and you want to do some scheduling optimization and you’re facing some logistical problems, then you should give it a try with our help.
Beau Hamilton (45:16)
Well, and for those looking to give it a try, where can they go? What’s the best way to get in contact with you and your team?
Jens Richter (45:22)
Yeah, yeah. So essentially, of course, we have a website. So not really surprising. dni-evolutions.com. And but we’re also around on LinkedIn, on GitHub, on Docker Hub, and many more. And of course, we can also keep it old school. You can write an email. We can have a personal meeting, an online meeting.
So essentially, I think the best way is simply go to our website or to our LinkedIn profile, and from there, I think it’s quite easy to contact us. Or GitHub is also quite easy to find us, yeah.
Beau Hamilton (45:57)
Awesome. Yeah. So they should be able to find just, you know, go to dna-evolutions.com or just plug it, plug the company name into Google. You should be able to find them.
Fantastic. Well, thanks, Jens. This is Jens Richter. Thank you so much for taking the time out of your day to talk to us about all these optimizations for DNA Evolutions. I really appreciate it.
Jens Richter (46:18)
Thank you very much for having me.
Beau Hamilton (46:20)
Thank you all for listening to the SourceForge Podcast. I’m your host, Beau Hamilton. Make sure to subscribe to stay up to date with all of our upcoming B2B software related podcasts. I will talk to you the next one.