The Exchanges

Q Maybe for the last part of this conversation, switching to the go-to-market side of things, um, Uh, how important was the partnering aspect of this? You, uh, you seem to have built very strong relationship with a couple of partners, and they seem to have been a big part of the acceleration. What was, uh, that story and any lessons learned there?

A Last partnership was instrumental. You, we couldn't have, uh, entered, uh, big enterprises doors without partners. It was really No, absolutely no chance back in the day. But you know, what I also realize around 2016 is that partners are not gonna fight for your own destiny. They are very opportunistic in the end, and they want to make money. So, uh, the same competitor that I talked to in the context of the India phase, they made the early on a decision to go completely indirect in, uh, in their go-to-market. I think that was a huge mistake. So we, we went Indirect, they are partners, but we build the, and I'm a big believer in a direct sales force that will fight for your own destiny. In our case, partners are not so important for us for reselling the technology. Partners are important to open the door and to implement our technology. But to win the deal, it's us. It's rare that the partner can win the deal for us. So this is, I, I've seen that that's the winning combination. To put your direct salespeople, to put the partners, Working together to win deals.

AI assessment note: “we couldn't have, uh, entered, uh, big enterprises doors without partners.”

Q And then another thing I, I read while prepping for this, uh, was the, um, AI trust layer, which was announced, uh, just recently. What, what, what is that?

A We discovered that many of our customers actually use multiple LLMs. And, ah, they, ah, they wanna have some governance and security layer. So, like an abstraction layer in front of the LLMs. So maybe they start with GPT-IV, but who knows, maybe code is better for a particular task or not. So instead of creating, again, a little bit of spaghetti in the application code, you rather use this abstraction layer, and then you configure the abstraction layer for different calls. What kind of LLMs are you using? And we can provide also better services like anonymization of data. We can do smart things, even if you send an invoice. To, to GPT-IV, to extract some information. What we can do, for instance, we can, we can digitize the document, and we can replace all the, all the sensitive information with, I don't know, with synthetic information, get it back, and then redo the translation when we go to the customer. So it's important to have this in between layer. Helps with a lot of things.

AI assessment note: “they wanna have some governance and security layer. So, like an abstraction layer”

Q And we're talking mostly about services slash consulting partners?

A Yeah, services consulting partners. So the technology partnership largely didn't work for us. And I, I think it's extremely difficult to make a technology partnership working. It, it has to be such a huge alignment at the, at the leadership level between two companies that It's, it's, it really works. That's my learning. It's, uh, this is why, you know, we, I'm also investing in a few companies, and this is what I tell them. Even recently, I'm talking to one of our portfolio companies that is on the verge of doing a partnership with a much larger company, and I, my advice to him is, uh, Build a strong relationship with the leadership team, and especially with the CEO of that company. This is a piece, a piece of paper doesn't make a partnership. A marketing, most partnership are just marketing announcement hype. But if there is no personal connection, it's not will on both parties to make it happen, it's not gonna work.

AI assessment note: “Yeah, services consulting partners. So the technology partnership largely didn't work for us.”

Q Why was it choice? Yeah. Was it choice of New York versus San Francisco somewhere else?

A San Francisco was, it's, it's too, it's too far away. I think for our global business, timezone doesn't help you when you do global business. It's not really so well connected globally, I think from, in, from flying perspective. And it's, we are not so big in tech. At that point, we are much, our interest was much more into the financial service industry, so I think New York and East Coast was, uh, was a good choice. And why New York? I think New York, it's a great city to, to be here. It's a lot of, you know, it's intense. It's, uh, it's fun to be. I don't regret my, my choice to put A little bit unusual, especially in 2017.

AI assessment note: “our interest was much more into the financial service industry, so I think New York”

Q So how does that manifest? Does it, is that a question of, um, Uh, customer support. Is it a question of like how your product roadmap evolved?

A It's a, it's a question of the product roadmap and not even the product roadmap, but it, it's, it was a question of, uh, fixing the bugs in the product. Developers, especially when you run so fast, developers are attached of building the next shiny thing. So, and many times companies have the temptation to bring a feature, but not finish it. Go to the next experiment, so you end up with an ugly mess, especially in an enterprise, when you have a long list of features, not in consumer. So, uh, this Japanese client actually with, uh, with, by their presence and the way of asking, so they commanded certain respect, and especially to our developers. So it was the first time when I went to them, when I said, guys, this time I don't take any bullshit from you. They said something. You go, though, I deliver. I don't want to enter into endless debates of what's priority was not. These guys are paying us our salary, so let's stop all this, you know, futile.

AI assessment note: “It's a, it's a question of the product roadmap and not even”

Q AI company through computer vision, uh, and, um, now AI seems to be everywhere in the, in the product. Uh, so there is, so use AI for analysis of, of communications, communication mining, and, um, You were, you, you just said you use AI, generative AI to start replying to stuff as well. So what, I guess, what is the latest in terms of like building AI into the product?

A Generative AI, it's, uh, it's, uh, in a way the missing piece in RPA and even broadly in automation. First of all, it lets you understand way better unstructured content. So it's, it helps closing some pieces in a process they couldn't automate before. So this is, so therefore that was the first use for us of GNI into our document understanding piece. And we have a, we have a generative extraction technology right now that help extract information from unstructured documents. Can even help to build better models to for semi-structured documents. In one of, one of our fastest growing business is this document understanding business. And, uh, we combine specialized models with generative model to basically deliver and an end product to the customer that improves over time. Because one of the thing where Gen AI, it's not, doesn't work well today, it, it doesn't learn in all fairness. All the fine training, it's more context grounding, large context windows, but it's not learning. They don't, you don't go to open AI and you modify their weights. This is not how it works. But with a specialized model, That's possible. So what we are doing for our customers, for instance, they want, they, they want to process invoices. I know it sounds trivial, but some of the invoices are so complicated, even a human user would find it difficult to understand all the relationships and to extract inf…

AI assessment note: “we combine specialized models with generative model to basically deliver and an end product”

Q And what's the operation layer? Is the analytics and the testing and to make sure that the system performs that scale?

A Yeah, we, that, that was, uh, Actually, really one of the breakthrough that we understood in, uh, when we went, uh, in, uh, in India at that, you know, that initial customer. Before that, our focus was to deliver one automation that, that was working well. So one developer, one product, build an automation running on that machine. We're scheduling on one machine, but So we focused so much on making it very reliable, and it was actually a good approach, because if you start Too big of a system. Day one, you'll end up with the spaghetti and nothing works. We mastered this piece and it was world class best ever. And it's still the best in the industry to build one automation, run it reliably. What we understood there that this is not enough. What, what you, if you have to build a hundred automations and they have to, they have to depend one or the other. And they have to run at certain times, and they have to run on different types of machines, with different types of software, different types of security permissions. So then we've built all this, a product called Orchestrator, That is capable of scaling 100,000 of automations, or thousands of runtimes, and make the communication between them via queues, and you have, it's a pretty sophisticated system.

AI assessment note: “built all this, a product called Orchestrator, That is capable of scaling”

Q And, uh, what's your strategy for this? Is that something that you're, uh, developing internally? Is that something that you're looking to partner? How do you go about it?

A We have a dual strategy. We have something called project foundation. So we are building internally, uh, You know, this dual model model that understands, uh, both text and images that we want to put at the foundation of many of our, uh, task-based automation. So it's aimed to replace document understanding, computer vision, the way we interact with the screens, But we would like also, we, we are partnering with one of the most interesting startups right now. You know, the one we are both investing in, in to build really this next digital agents that will be capable actually of combining knowledge of a task. Or a process maybe later on with the capability of acting and replicating what the human user would do to, to complete that task. So in this way, we will have to learn how an accountant works, how an auditor works, and That paired with some of our technology that is capable of acting on user interface, and their ability to plan, to have, I have the knowledge of a task, and then I can plan how that task should be completed. I can create a list of steps And then carrying on these steps using UiPath technology, I think it's an amazing combination.

AI assessment note: “We have a dual strategy. We have something called project foundation. So we are building internally”

Q So, um, where did the initial idea come from?

A Well, it's, we, uh, we had some, uh, experiments on the product side, but they were mostly in the consumer part of, uh, so we tried consumer. I thought it's way easier than to just go into big enterprise. I had no idea how to enter. So we, we built a really Small tool initially that was like, uh, a dictionary, something providing quick access to online resources from your desktop computer. So like, if you long clicked on any word on your computer, we would, uh, understand what you are doing, extract the word behind the mouse cursor, and then offer a pop-up to search, to translate, to do some stuff. It's actually way more difficult than you imagine to To understand the context of every application and how they draw text and everything that's there. So we built pretty sophisticated, uh, it was a system programming to intercept all the Windows APIs that an application is using, and we were rendering the text, you know, in memory, and we figure out. So it was pretty, pretty interesting. This is what we are doing in the first place. First, we have started with this idea. Let's intercept. API calls, and then let's build a product around it.

AI assessment note: “First, we have started with this idea. Let's intercept. API calls”

Q Did you go direct or through a partner? That's a usual question. So Japan being the second largest software market in the world.

A We went, uh, we went through a partner in, in Japan initially, but, uh, I was lucky to meet, uh, a Japanese, uh, guy that has become a bit later the one running our business in Japan. Și a fost instrumentul să ne puteți în front of large customers. Și să te raciți un anecdote, care nu cred că am mai vrut să-l spune. So I was, uh, I was in Tokyo for the first time, and I was meeting, uh, that, uh, uh, you know, uh, that guy from SMBC. His name is Yamamoto-san. We've built really a good partnership and friendship later on. But when I met him first time, I didn't have cards. So I, going to Japan, I had to, to print business cards. So at that time I had only my Romanian phone number, so I thought it's not a problem to put. Turn out that this, uh, this kind of scared him. How can I do? I'm a, I'm one of the mega banks in Japan. How can I do business with a company based out of Romania? And then what I did the next day, I, I went to Skype and I bought a K number to have to put it on my business cards. But anyway, it's, he, we started to, to get his trust. With our product, and then with our dedication, and we hired people in Japan, and started to, to really offer them a lot of free support. Always, our, our politics was really complete customer centricity. And, you know, it's, it's very easy to say this, but it's not so easy to practice it.

AI assessment note: “We went, uh, we went through a partner in, in Japan initially”

Q So obviously a, uh, natural next evolution of all of this is the acceleration of AI agents. Um, so we have processes, but then you build a glue between the processes of agents making decisions based on, uh, whatever the input is. How do you think about this?

A I think this is gonna be maybe the biggest transformation we are seeing. In the, in the enterprise space that would be based on LLMs. To me, you know, let me start with the generic, my, my generic thinking about where is, where are LLMs today? They are, they are, they are, they help a lot with the increasing productivity of clerical workers. But they are not transformational. Why I think they are not transformational? Because you cannot know it's, it's a user-based, it's a tool. It's a user-based technology. You cannot use an LLM in an enterprise production in an autonomous fashion today. It's simply impossible because of the hallucination aspect and the lack of reliability. In a way, I think for, to get to an incredible transformation, we don't need smarter LLMs. I think the LLMs that are today are smart enough. We need them to be predictable. Reliable. Because in, think about, even if we, even if you, if you put in the shoes of an enterprise, then you hire a person. I don't think many people would shy away from hiring, uh, like, a genius that, but it's, that is extremely unpredictable. They show up to work only, you know, one day a month, and they have all the, it's, for most of the job, that doesn't work. You will need, you will need smart enough people, but reliable, diligent. I'm not an expert in the field, but I believe that this is gonna, we need, uh, Another giant leap …

AI assessment note: “We will get to autonomous agents, but by going through, first, we will have”