Обединяване на хардуер и софтуер: автоматизация за устойчив бизнес растеж | Ивайло Караманолев

This talk discusses the evolution of sustainable business models, particularly in the software and logistics industries, as illustrated by the speaker's experience at eBay. The speaker emphasizes the importance of efficiency and controlling both hardware and software to achieve a competitive advantage. By implementing a sophisticated assembly line in their fulfillment center, they have succeeded in improving order delivery rates significantly while optimizing labor costs. The talk also draws parallels with successful companies like Amazon and Apple, showcasing how their integrated approaches lead to remarkable efficiency. The speaker concludes that true sustainability in business comes from adapting and combining physical processes with software innovations, ensuring long-term resilience against competition.

[music] Hello, nice to be here. And I'll start with a story from when I was entering the industry. Back then, it was just becoming fashionable, and pure software startups, the so-called SAS, and cloud had already been established, so it was very easy. Anyone who can write software, who feels brave enough to start a business, to start a startup with some idea that might be easy enough to

implement. There are no major technological difficulties. We can write software, sit down, do it, launch this startup, and look for clients, as they say, hustle, and start accumulating monthly turnover, 10, 20, 30, 40,000, let's say, and we live a great life. Back then, I thought that this was the dream and that this was the direction in which I wanted to develop. Naturally, the flow of life took

me in all sorts of other directions, and so I never reached that dream. Ah, but in the meantime, what we're seeing more and more these days is that maybe that's not actually the dream. Maybe not the most sustainable business model. Yes, at some point it may do well, and then competitors may emerge. Purely software startups are very susceptible to copying a business model, to working through artificial

intelligence, which is what we talk about these days. And the rivers of life brought me to eBay, where I am right now, and I'm very proud to be there, because I see a business model that has been around for many years, becoming more and more established. And the philosophy of our founder Ivan inspires me with a lot of confidence for the future and calms me down in

times when perhaps many of us wonder what lies ahead, what is happening, what it means for us. And, working, right, implementing various projects, over time I make some observations about what actually makes us a sustainable business, gives us a competitive advantage, and gives us peace of mind for the future. And observing how it is with us, observing how it is with other large, successful players in the

industry, I have found some parallels, let's say, in my opinion useful lessons for business and development, which I want to share with you today. So, what we're going to talk about is what dangers, for what types of businesses, appear with the introduction of more and more artificial intelligence. We'll talk about one of the larger projects we've recently implemented at IEK, a large assembly line that encompasses our

entire warehouse. We will talk about the effects on the business and the difficulties we had to implement it and we will also draw parallels with some other large well-known companies to see exactly what these things are that coincide and are not in any way specific to the company, but can be applied to other sectors, to other companies and to other types of companies. Speaking of sustainable business

growth, it might actually be a good idea to first say what sustainable business growth is. And in our fast-moving industry, efficiency is perhaps one of the most important qualities. We are very proud to be one of the few, if not the only, companies that are making a stable profit. We have an awful lot of competitors who have big growth, big turnover, but actually lose a lot of

money every month, every year, while this is not the case with us and we achieve it through relentless pursuit of efficiency at every level. Software, hardware, working with people, transportation, everything. And another thing that is perhaps becoming more and more fashionable is so-called exponential growth. In English, it's called hockey stick blood. And we start slowly and we accelerate, we accelerate upwards in our growth. And our philosophy

has always been that this is great, very nice, but it is not sustainable, in the sense that it cannot last indefinitely . The increasing growth rate must reach some natural ceiling, above which nothing can, right, grow in this way for many years. This is very important to us. That is, we don't want to explode at one moment and at another moment wonder how to actually adapt the

company that is used to 300% 1000% growth on an annual basis and then after three or four years to actually hear how the entire company, the entire structure, how we need to adapt it so that it can function in a mode of 50% growth per year or 30 or even zero. And these are very difficult processes for a company, while with us, a constant controlled pace of

development can be maintained for a much longer time and in the long term is much healthier for both the employees and the company than some kind of, so to speak, explosion. And it is very important to be resilient to competition, that is, to have an advantage in business, in the labor market, and in many other areas. And so that our business is always one step ahead of

others, and last but not least, we must be able to adapt to changes. That is, what are we doing so that, while observing external factors, we can continue to be competitive in the long term. Artificial intelligence is coming, we all see it in almost every lecture you 've listened to so far, it 's one of the main topics. It won't be a main topic here, there are

many reasons for this. For me, artificial intelligence is a help, but it is not a leading thing in sustainable development. But how, what effect does this have on our businesses? Purely software, purely intellectual endeavors will perhaps not be replaced, but they will certainly be greatly accelerated by artificial intelligence. When it comes to physical processes, logistics, working with people, everything that happens in the real world, it's much

more difficult. The effect there is much smaller. This means that if our business combines the two, that is, works with physical objects or physical processes that are optimized through software, it is much more sustainable. Yes, we can improve our work with artificial intelligence, but ultimately everything happens in the real world. And the big language models still operate purely virtually. What's happening is that we all interact with

a lot of hardware. Our phones, laptops, and cars have a hell of a lot of hardware. Ah, but this hardware that we interact with is actually a very small part of the whole thing. And in our business, as in many other businesses, a lot of what happens is actually behind the scenes. And in the picture you can see that in one of the previous lectures we talked

about the machines that produce modern microprocessors, everyone says how they are produced by TSMC. It was mentioned that one of the companies providing the machines in question is SML. She is European. In the picture of the machine, you can see that it is actually terribly complex, terribly large, very innovative, and apart from people who have been interested in or are involved in this thing, few other people

have any idea what actually sits behind this infinitely small and very powerful microprocessor that is in the phone in our pocket. And this means that what people see, being a small percentage, we control, so to speak, the iceberg that is under the water, that is not visible, and it is not visible to customers, but very often it is not visible to our competitors either. That is, the

whole thing that we do behind the scenes, the magic that we do to make the end result happen, and that's where the difficulty lies and that's what gives us the competitive advantage. And accordingly, it is very important to invest in these things. They should work smoothly in our case. And the customer simply receives their order at home, hopefully on time, hopefully with quality products, but we have

done an awful lot of work to make this thing work. this specific example and project from our business that I want to share today is about the conveyor line, which has been successfully operating in our fulfillment center for several months now. And currently it consists of 640 linear meters in a straight line. It's on three levels. It used to be much smaller. Over time, we have expanded

it quite a bit, also in the context of the efficiency that I talked to you about, and I considered it. We hadn't been interested in this perhaps slightly strange metric until now, but for every linear meter of accurate line that we have in our fulfillment center , we deliver 6.5 orders to our customers every single day. This, in my opinion, was a pretty interesting way to measure

how efficient we actually are , how much hardware we buy, and how much money we invest per unit of work done every single day. And for comparison, one of our competitors in the United States of America, who has a fulfillment center in New York, and they have also published data, and they have a lot of kilometers. They don't talk about anything under a mile there. And they

have the same indicator of 2.07, that is, more than three times fewer orders for each linear meter of line. And this is a kind of loss in a sense of money, an investment. This thing uses electricity every day, needs maintenance, and being so efficient, we're very proud of what we've achieved. And as a direct measure of what effect this assembly line has had on our business. Compared

to a year ago, we are delivering 44% more orders, and only 6% more man-hours have been worked in our fulfillment center. That is, the efficiency with which we work, even compared to a year ago, is quite high, and we are improving year after year. like a maximum load in one hour, just to illustrate to you, right, how much stuff actually goes through that assembly line. And over

1600 cassettes every hour at peak load pass through the assembly line and that's 2.6 items per hour, and 2.6 items per second every second, second after second, within an hour and quite often within a full working day for us, 16 hours. The assembly line itself is a fairly well- established technology. Conveyor lines have been around for many decades. What makes ours different, more efficient, and what actually

gives us the competitive advantage, because we are here to talk about how our business can have a competitive advantage over all others, and not just be one of thousands, tens, hundreds of thousands of companies that have a precise line as part of logistics. For us, this lies in the fact that we have hardware that is similar, not to say the same as many others, but a way

of managing this hardware through specialized software that we develop ourselves, that's where the key lies. And purely software-wise, we, being a business that works with physical products, can achieve very little. There are optimizations that we can only make by adjusting the software. The client site may get better, but ultimately it's about physical processes. But software is what sets us apart from everyone else. This is what we

develop ourselves and this is something we believe that the more we invest in, the more value we can bring to customers. Ah so aah what advantage in this case is the software. something like a universal guiding star that we use to make all kinds of hardware integrations, because in fact it's far from the only hardware project we've done. We work with scales, deposit machines, barriers, sensors, scanners,

many, many, many different types of hardware. But what we always try to do is when we integrate new hardware, we get an API or a way to control the hardware at the lowest possible level. What does low level mean? As close as possible to what actually happens as a physical process. When it comes to the assembly line, what is the low level? The low level is the

path of the cassette through the relevant sensors, encoders, barcode readers, and so on. And if it's about, for example, a smart locker with opening cells, the low level is a person goes to it, scans the chip, and we tell the hardware to open a door or multiple doors. What is the alternative and what is the way that many of our competitors operate and many of these systems

are actually sold that way. They are sold as an integrated, complete solution with ready-made software, where, for example, you load all your orders for the day, the system takes them and starts to execute them. This is a very big problem for us. Why? Because every system comes with some understanding of what an optimal process is. it comes with some ready-made software. It must simultaneously be flexible enough

for different companies, for different ways of use, and also more or less fixed in time. That is, they developed it the way they thought best and we use it that way. For the assembly line, this could mean that we hand them all the data for all the orders, and they balance the load between our cashiers, between our sectors, as they see fit and according to their algorithms.

We actively fight for the opposite in every project we implement. That is, give me control over the hardware, over the routing of the cassettes. We implement everything else. At first, we may not be better than what is available on the market, but over time, by observing our processes, by observing our business, we can become better and better. We are uploading software now and for the past many

years several times a day. That is, at any given moment, the latest version is a few hours away in production, and this gives us the opportunity to iterate super quickly and constantly improve our efficiency, with the ultimate goal being to be much, much better than all the ready-made solutions available on the market. One of our competitors can go buy the ready-made system, integrate it and see that

they have the same assembly line or the same locker, but their rates or results are, say, 30% lower than ours, 50% lower than ours and they ask themselves , okay, we do the same thing that eB does, but we have an assembly line, we have suppliers. What are we doing wrong? Where is it from? Where do they get this 50% higher efficiency from? The answer is difficult,

and that's good for us. Why is it good for us? Because when the answer is difficult, it means it is also difficult to repeat. It's difficult because it's made up of thousands, thousands of small parts that, working together and working in sync, give us this including the I'm saying is that we upload software several times a day for the assembly line, given that a tape passes every

few seconds, and that's quite a challenge, because the tapes don't have to stop. They stop and there is a traffic jam immediately. People no longer have anywhere to leave their products and so on. So, what we do is not easy at all . One of the colleagues who deals with the software on the assembly line and uploads software several times a day is here with us. Aa

chuckles a little because she knows what this whole thing is about and how quickly she gets the phone call when the line goes dead. Ah, but using this software with low-level access to the hardware is actually very challenging, because many of our partners are not used to working this way and look quite strange when we ask for this kind of API. Sometimes they have it, but they're

not used to giving it away for direct use, and sometimes they don't have it at all and we want them to develop it. So we have had a lot of difficult conversations with partners to get to this point, but the flexibility we allow ourselves afterwards is very great. And this is a screenshot of our internal system for managing the assembly line and our overall productivity in the

fulfillment center . We won't go into detail, but many people in our company spend almost their entire day looking at this, as well as many other screens that measure many metrics. But the purpose is illustrative, to show you how many different things we actually monitor in order to achieve this high efficiency. Ultimately, to realize this competitive advantage and have this sustainable business, because without a competitive advantage,

we will not survive long in our industry. as I mentioned earlier, and the lessons we 've learned, I think, are applicable in many other areas. I'll start with a field that is very close to ours and a familiar name: Amazon. And Amazon, we are talking about a commercial business. It is very similar to ours, because after all, they also have a large online store. And they are

a little different because the types of goods they offer are quite different. That is, they don't work, say, specifically Amazon, without their divisions, with food products, but for them, and they are known for this, is that in the same way they chase enormous efficiency and for them, every hundredth of a penny allocated to a process, multiplied by millions and billions of orders, makes a very big difference

for them. Our assembly line and software control project may seem comically small. They operate on a completely different scale. And they currently have over 750,000 robots working in their fulfillment centers. And one of the main ways they can achieve this is by purchasing Kiva Roboics and many others, but Kva in particular currently has the very large number of robots in question. And in the second picture, you

can see one of their arches that automatically scans cars when they come back from delivery runs, which allows them to skip an entire, say, five-minute operation where a person goes with a camera or a phone, takes pictures of the car, looks underneath it damage, and so on. And this is entirely their internal project. This is something that is quite complicated to implement. And something that is not

offered as a commercial product and anyone who wants to reach Amazon's efficiency is another thing they have to implement, at least as well as Amazon, in order to be able to compete. What does this mean for all of Amazon's competitors? It means billions, tens of billions of dollars, that they would have to invest to be able to compete effectively in Amazon's core business. That makes it very

difficult and means that this whole Amazon infrastructure is very resilient. Do they have competitors, is it difficult for them, yes, most likely. Ah, but in terms of price and they are very difficult to achieve. What are Amazon doing? They have the hardware, it's theirs, they have the software, and the whole thing is completely hidden from the user, who ends up getting a package delivered to their door.

And that's a very strong point. And moving away from the online retail sphere, we might see Apple. It is a very well researched company. Everyone has heard of it, perhaps many of us use it daily. What are we seeing at Apple? And they make their own hardware to a large extent, not 100%. Many of their components are manufactured by other companies, but what we see year after

year is that every component that was, let's say, completely bought from an external company, and as an example I can give both the processors that they bought from Intel and put ready-made Intel processors in, now they put M processors from M1 and so on. And the processor is not manufactured by Apple itself, it is manufactured But Apple is getting closer and closer to that lowest level, which

is like getting closer and closer to the bottom, right? They used to buy the entire processor, now they the architecture, design the whole thing and just give it to the implementation. Maybe no one would be super surprised if in 10, 20, 30 years they were even closer to power than they are now. And before and maybe still for iPhones they put ready-made 5G modems, which eventually create

problems for them such as energy efficiency, how it works, speed, and so on. One of Apple's big projects right now is to make its own 5G modem. They probably won't produce it themselves, but we might see them control absolutely the entire user experience of the end- client software. to some point and step by step they are moving towards complete end- to-end integration. Each such link in this

chain, which is within Apple itself, gives them another competitive advantage. There's not going to be any huge huge difference in many of these steps, right? The processors were quite revolutionary, but many of the other steps give a paltry 1%, right, if we count the ultimate customer satisfaction, maybe hundredths of a percent, but step after step, dozens after dozens of steps, given that each step is quite difficult,

that means they have a competitive advantage that is very, very, very And as a final example that I would like to talk about, is Garmin. As I mentioned, I am a big fan of mountain biking. I've recently started running, and that means I spend hours a day with Garmin fitness tracking devices for cycling. Accordingly, this company is very close to my heart. I have been using them

for many years. And they come from a context where they have a very long history in the production of aviation hardware. That is, imagine the aircraft control screen with the map, the so-called avionics and also for navigation, and sonars, navigation modules and so on. As you can guess, in these two industries, the requirements for hardware quality and durability are very high. That is, Garmin knows very well

how to make something sturdy, waterproof, and capable of withstanding many years of quite serious demands. But since I've been using them for a long time, and for many years, there was no better alternative, and my software was extremely, so to speak, wooden. Syncing with the phone didn't always work very well. You had to restart it. Sometimes it hits. And the experience was not good. But over the

years, they had to do a thing or two. Or let competitors overtake them, and they have competitors. There are already at least two or three companies that produce similar watches, bike computers, and so on. Or they should seriously fix their software. What does this tell us? It tells us that great hardware without a good software experience doesn't give us enough added value. And in the photo at

the bottom you can see how in the long term, year after year, they maintain their competitive advantage over ahu and I think Hammerhead, which produce similar quality. since the ecosystem consists of many different products that work together, let's say a bike computer, a phone, you can have a radar on the back, you can have lighting, a lamp on the front, they are all connected in a single

network and communicate with each other , if it doesn't work well, you have a competitive advantage and it's hard for others to replicate it. If the software didn't work, the companies had a real chance of falling behind everyone else. We've already talked about it, I just mentioned it to you, but we have a very big problem with all the ready-made solutions, ready-made administrative panels, ready-made business processes,

because they greatly limit our efficiency and don't give us a competitive advantage, they take it away from us. They take it away from us, because when we integrate such a solution, our entire business enters into realities that we then have a hard time getting out of. And all people get used to working with it, and processes are organized around it. That is, the more we introduce it,

the harder it can be to break away from it. But at the same time, any competitor, any other company can go and buy this. He's gone today. They do a project in a month or two, integrate, configure, and do the same thing. Does this give us a competitive advantage? No. When we do it ourselves, when we do it better, even if it's 20% better, and often it's

more, often it's 50, 80, twice as good, what does that mean? It means that we are not dependent on an external supplier. It means we can iterate very quickly. And with us, an idea is often born on a given day, and especially if the implementation is faster, by the end of the week it may already be a fact and introduced as a new process in the order

fulfillment center. In other words, every ready-made solution actually has the opposite effect, even if it improves our And in English there is a very important term for me: mode. And this is originally the moat around the castles. a, which prevents the enemy war from attacking and capturing the castle. And in movies you might have seen a big wooden, a big wooden bridge that goes down or in

our case we don't want it to go down. And the whole lecture is actually focused on how to build such a mode, a competitive advantage, that would prevent our competitors from taking over our territories. And to reiterate what's important, it's important to manage both our hardware and software, right? We write our own software , and we have low-level access to the hardware . And it's important that

we are very good at making software, because if we aren't, it will have the opposite effect. We will have hardware that we use inefficiently. Oh, and in fact, each of the processes should be our own. we invented it and if we do it well, that means we are better than everyone closing message, the future of those who control both atoms and bits, that is, everything that happens

in the physical world and everything that happens in the virtual world, hardware and software. And of course, there will continue to be companies that only make software, there will only produce hardware, but for us, the truth is in very effective businesses that combine the two in the most efficient way. This is from me. Thank you everyone.