# UiPath's Daniel Dines on Why Agents Do Not Mean RPA is F***** Why We Have Reached the Upper End of Scaling Laws · The Future of Work in an Agent World and What Everyone Misunderstands About Enterprise AI 20VC · Dec 18, 2024 · 58 min · 9,499 words Speakers: Daniel Dines, Harry Stebbings Source: https://www.996.fm/episodes/20vc--ep-77b4c0ad/ ## Cold open **Daniel Dines** [0:00]: This is a story that I never told anyone. I've wasted my late twenties, a big part of my thirties and forties thinking in this way. It's a totally waste of cycles and energy, man. I am a lonely wolf. I find life pretty lonely, I live mostly in in my head, thinking, analyzing, reflecting. This is how I spend my life. **Harry Stebbings** [0:30]: This is 20 VC ## Intro **Harry Stebbings** [0:31]: with me, Harry Stebbings, and I'm so excited to welcome back an incredible CEO and a very dear friend of mine, Daniel Dines, founder and CEO at UiPath, the $8,000,000,000 public company dominating the RPA space. And everyone says bluntly, RPA? It's pretty challenged in a world of agents. Well, today, we discuss the future of agents, how RPA and agents coexist in the future, and what is BS and what is not in the agent world. But before we dive in, what do Henry Ford and AI have in common? Neither could change the world without automation. In the future, there will be two types of business, those that have automated and those that wish they had. UiPath, the undisputed leader in automation, is taking us into the era of Agentic Automation. UiPath's new AI agents don't just follow rules, they think, they make decisions, they work along side the world's most powerful software robots already trusted by over 10,000 businesses. If Agentic Automation sounds new, just think of UiPath as more growth, not more overhead platform or happier customers, happier employees platform. Whatever you want AI to do for your business, Agentic Automation with UiPath will make it happen. Try UiPath's new AI agents for free at uipath.com. The future of automation is both Agentic and robotic, so don't get left behind. And speaking of transforming your business with cutting edge technology, let's dive into another game changer, Ateo. Ateo is the next generation of CRM. Setting up Ateo takes less than a minute, and in seconds of syncing your email and calendar, you'll see all of your relationships in one place, all enriched with very valuable data. Atio also lets you build Zapier style automations, gives you powerful reports, and works perfectly for any go to market motion from PLG to sales led. And Ateo is designed for the next era of companies like yours, and companies like yours shouldn't have to deal with inflexible one size fits all CRMs. So join industry leaders like eleven Labs, Replicate, Models, more to scale your startup beyond the next level, head over to ateo.com/20vc, and you'll get 15% off. That's 15% off your first year at ateo.com/20vc. And if Atio helps you stay ahead by streamlining your relationships and operations, today, I want to talk about a venture fund making waves with its unusual model. I'm talking about the Fundrise Innovation Fund, which is democratizing venture capital as a public venture fund. For example, most of the AI revolution is being built and funded in the private markets. Companies like OpenAI, Anthropic, Databricks, these are incredible multibillion dollar companies, but they're inaccessible to 99% of investors until they go public. Well, those days are finally over. Visit fundrise.com/20vc to check out the Fundrise Innovation Fund's impressive $150,000,000 portfolio for yourself. Carefully consider the investment material before investing, including objectives, risks, charges, and expenses. This and other information can be found in the Innovation Fund's prospectus at fundrise.com/innovation. This is a paid sponsorship. You have now arrived at your destination. ## Conversation **Harry Stebbings** [3:52]: Daniel, dude, I am so excited for this. Thank you so much for joining me in the studio again. **Daniel Dines** [3:56]: Well, I invited you basically for this podcast, so **Harry Stebbings** [4:01]: thank you for having me. Dude, honestly, our first show did so well. I got so many emails from young entrepreneurs, and so I thought I would capitalize on the success of the first show. And I wanted to start with something that you said to me before, which is at this stage of kind of the AI cycle, product matters more than innovation. Can you unpack that for me? **Daniel Dines** [4:21]: Yeah. Actually, I was thinking a lot lately. What's our story within the AI narrative with where we can really bring a lot of value? So over the last two years, we we've spent a lot of time really trying to fine tune LLMs, build around them, and to certain degree of success. But I've got really inspired by stories like Cursor AI, and my development team loves that product. It's a beautiful product built on the top of multiple LLMs, but just works. I recall actually how we started UiPath. There maybe there is a this is a story that I never told anyone. In In the beginning, we were always based on AI, but we were using a library called OpenCV, which among many other things provided a really cool feature to find a smaller image within a bigger image. And we repurposed that library for the sake of automation. So we can take a screenshot of an application, and if you wanna click on a button, we can take an image of that button and then find it during replay. Just call, you know, a function, find this image of the button. I will get the coordinates, and I will click on the button. But that was not the only thing that we did. We created, I think, a magical experience. So we let someone to record a flow on the screen. Just show. I need to click this button, indicate on the screen the button, and then we will generate, like, a very simple statement. Like, click that button with the image. Everything was stored. So you could have record an entire flow based only on working with images, even typing in an edit box. So we will capture the edit box image and then, like, a label, and we can we will find them during runtime. But from the perspective of the user was really simple. So I remember it was like 2013 when I showed this product to some guys that were really Blue Prism experts. And in order to do the same thing in Blue Prism, it would have taken like two days and the outcome would be not as reliable as in our case. I did in front of them this flow, like, in three, five minutes, press run and it worked. And I asked them, what do you think, guys? Total silence. They couldn't believe their eyes. And that was our first niche where people would deploy and would prefer us versus Blue Prism. And from that one, we have basically expanded, you know, into what we are today. **Harry Stebbings** [7:05]: So do you believe that we're then in that stage in the AI cycle stage, which is, like, ease of use, product simplicity, great UI is the driver of adoption, not technological advancements, sophistication of models, you name it? **Daniel Dines** [7:18]: Yeah. I personally don't believe that the models can innovate in a material way in a reasonable amount of time. I think they reach maturity in a way. I think I am pleased with what LLMs can deliver both frontier models, but also something like, you know, smaller models. Look, for instance, we are using GWEN, which is a fantastic model built by Alibaba. It's just totally open source. We are using it into understanding, like, a lot of our semi structured documents. It's a lot of product. **Harry Stebbings** [7:54]: Why do you use an Alibaba model versus the other models? **Daniel Dines** [7:58]: Because at this point, that's basically the best model for this particular job. We might change it. This is why I think the experience around the product will be so important because we do a lot. It's very difficult to use that model without the entire product experience, without helping people tagging documents and retraining the model on the fly. So it's it's an entire experience that we make it extremely simple, and we can exchange the model. If we find another model, maybe LAMMA 3.3 is better. And it's always a cost versus speed and versus accuracy equation that we have to **Harry Stebbings** [8:39]: Do you think we will continue to live in a world of many models which are specialized in many different areas, or do you think there will be one or two monolithic models similar to what we have in cloud today with AWS, Azure, Google Cloud? **Daniel Dines** [8:52]: I don't think models would would measure the cloud development. I think there will be there will be multiple models. Even if we look at the human brain development, we have multiple models. We have some general cognitive models, but we have a lot of specialized models that would do some tasks better than using my general model. One of the most difficult task to replicate in today is to take this cup and drink out of it. I am doing it's just perfect how I do it. But that's a dedicated model that we are training since we are very young. We train how to walk. I don't think in how to walk, so it's a dedicated model. So obviously there will be a world of few frontier models and a lot of dedicated models. And dedicated models will be built on the top of I think it's more likely to be built on the top of open source models, that closed source frontier models. **Harry Stebbings** [9:49]: With the recognition that Cursor is the right approach, that product is so much more important than technological advancement in that respect or product matters more, how does that impact how you view the next chapter for UiPath and what you do? **Daniel Dines** [10:03]: It has a big impact. It's a very big shift that we have actually made in the company in the way we've built software. It's not enough to think building incremental. Cursor was built from ground up, really, and in an AI world. It was an AI first product. This is how we should build. So we are we are building our agentic AI approach from the ground up. We gave up on some of our RPA stuff in order to come on the new technology, new frameworks, building from scratch because we want to build an AI first experience. **Harry Stebbings** [10:41]: What did you give up on the RPA side? **Daniel Dines** [10:44]: Look. We had an workflow engine, for instance, that was like Windows workflow engine, and we have perfected it over the years. But now we have switched to a more modern technology, that workflow engine, and we are building on the top of that. So I resisted for so many years, and engineers in UiPath will realize how many discussions we had for us. Should we bring another workflow engine? Should we not? AI finally convinced me that's the right moment where we we should bring a new workflow engine that is specifically designed for what we call agentic orchestration, specifically designed to facilitate very well the connection between agents and human users and other robots or other models, dedicated models, and some other APIs, some other entry points. **Harry Stebbings** [11:38]: Help me understand when we think about kind of this entrance into agentic orchestration. People are skeptical about how RPA coincides in the future of agentic orchestration and agent use in the enterprise. Why are they wrong to think that they are incompatible? **Daniel Dines** [11:54]: Because I think very few people understand what are the use cases where RPA is really used and valuable and why actually agentic doesn't work for those use cases. I can elaborate. Can we can we unpack them? So the sweet spot for RPA is to automate tasks that span multiple business systems and are of medium to high complexity. So they can span multiple steps. Usually, it can be even a hundred, two hundred of steps. But the key is they are rule based. The input is structured, and then the steps that you go are rule based. But they in a way, they capture the company knowledge within the rules. Even simple things like if the VAT starts with this particular two numbers, then you have to take this particular flow. But you capture it in rules. This is very important. But these automations are very reliable. They simply work until the underlying system changes. Now when it comes to agentic, LLMs are actually not good at following repetitive steps. You are not gonna have LLMs multiply to numbers. No. You are gonna follow an algorithm and you are gonna use crip language or you will program it. Right? This is kind of the same with automations. LLMs work relatively well when we are dealing with unstructured parts in a business process. It's sometimes the enterprise knowledge, it's difficult to express in rules. You can eventually. It's very difficult. There is a lot of tribal knowledge on the top of the public knowledge that human user is supposed to have. So when you cannot express in rules, then you can build an agent that will mimic what the user will do. But with the intent that will reduce the human input on that part of the process, you cannot really eliminate a task using Agentic. Because in a way, Agentic AI, it's about delivering something autonomously. **Harry Stebbings** [14:03]: But if we look at it as, like, rule based versus non rule based, because I think that's kind of the easiest distinction between the two, does that mean that enterprise customers will essentially buy from two different vendors for rule based versus nonrule based? And why would you not just go, fuck it. We'll just do nonrule based. And why will nonrule based not just go solid? We can also do rule based. **Daniel Dines** [14:22]: Well, this is a great question. Thank you. And I think this is the essence why we have it right at the Agentic AI table. The answer, in short, is a rule based and nondeterministic part actually sit within the context of a business process. It's like a long long business process like order to cash or procure to pay. You'll have nondeterministic parts and you'll have deterministic parts. It makes really sense to have the same technology and put them in the same framework. This is why agentic orchestration is so important. We have the technology that connects all the parts of the process, and we have the technology to automate those steps in the process. Think about as a metaphor, like, robots are more like low skilled employees while agents are high skilled employees. But you manage them within the same platform. You don't have two different workdays. **Harry Stebbings** [15:20]: And so you wanna be the platform that manages low skilled and high skilled? **Daniel Dines** [15:24]: We are the platform that manages low skilled, and we are adding something to manage also the high skilled. And another thing that it's not so easy to realize, it's not enough to have a technology that automates a single task. You will have to be capable of automating thousands of tasks and manage them, deploy them, monitor them, get analytics out of them, access control, who can run agents, who can run these particular workflows, what applications they can access. We have this. We built a key differentiator in UiPath platform. It's our ability to orchestrate robots. **Harry Stebbings** [16:07]: But do you have them? Because you have them in a rule based environment and rule based processes, and now we're moving into a completely different universe of non rule based ambiguity. **Daniel Dines** [16:17]: Well, it's not it's not a completely different universe. You know what's the common denominator? Both RPA and agentic imitates people. And when you imitate people doing a process, there are slightly different ways in order to deploy and manage because it's more fragile. You need to have a lot of exception handling into places. You need to have a lot of retries of different things. Maybe think about you lowered a website. It's variable until it responds back, or maybe it's a lot of things that you have to build within the technology to make it more reliable. We have the experience with robots, and we are taking this experience to the agents. It's not so difficult to build an agent as to make it working reliably thousands of times. You go there, you call it as part of an enterprise workflow, and it has to work. Otherwise, enterprises are not going to deliver them in an autonomous fashion in production. One story that I keep hearing from our customers is they prefer our workflows to fail than to be too smart because their risk appetite for this type of workloads, it's low. This is the mentality of our customers. That will be the same mentality with delivering agents. Agents will make recommendations. Agents are not gonna take actions directly. There will be a progression from agents making recommendations, going to a human user for validating, and then calling an action. **Harry Stebbings** [17:54]: And that and just so I understand, and that is because we are scared of them being too smart and making decisions that are wrong on our behalf? **Daniel Dines** [18:01]: Not because we are scared, because they are like idiot savants. Sometimes they can be extremely smart, sometimes they can be extremely dumb. You have no idea right now how to distinguish between these two scenarios. **Harry Stebbings** [18:13]: With respect, is that not a little bit like Workforces? Like, they make mistakes. **Daniel Dines** [18:21]: They make errors. They make errors of judgment. This is this is why most enterprises will create a lot of rule based workflows and a lot of precision type of enterprise workflows. They would not let people every time to decide, who should I call? Should I call Daniel or should I call Harry for this job? No. It's based on rules. **Harry Stebbings** [18:41]: How long do you think it will take to move from a recommendation based engine adoption cycle to a fully trusting agent? **Daniel Dines** [18:50]: As long as it's gonna take for the nice self driving cars that we have today to be fully autonomous and to drive, you know, sides with people on the streets and they will just work. **Harry Stebbings** [19:03]: Do you really think that long? Because I mean, that the self driving car development has been **Daniel Dines** [19:08]: I truly think it's gonna be that long. In the same time, we will reduce actually a lot the human input. We are going to a place where we'll have a semi autonomous agent that will do most of the job, and humans will just sit, monitor their inboxes. They will get tasks like, please validate this. Please validate this. What answer is this. I need this information. And then feedback to the agents and to the an enterprise workflow that will be orchestrating the work. And the enterprise workflow will be rule based. **Harry Stebbings** [19:41]: So when we think about this kind of landscape, we've got the orchestration layer on top. We have kind of rule based RPA, and then we have kind of non rule based kind of agent workflows. **Daniel Dines** [19:49]: But the orchestration is rule based. This is how work is done today. Will have a lot of rule based workflows that will connect different people to do do their job. **Harry Stebbings** [19:59]: Do you as UiPath also want to be the provider of agent based, non rule based **Daniel Dines** [20:05]: actions. But we also will integrate with agents built with other platform. We call a lot of APIs from different platforms. Right? An agent to me within the Salesforce platform will be just an API. I will just call an API, I will get an answer back and I will feed it to a human user, say what do you think about that answer? And then based on human response, I will call another action maybe in another or call another agent in another enterprise Because we will be the orchestration on the top. And there is also it makes a lot of sense to have an orchestration technology that is agnostic. We call it the Switzerland of the platform. Because an orchestration engine should provide equally good access to different platforms. What's the interest of Salesforce to provide amazing connections to SAP and vice versa? No, it's not. So they will focus on building agents that work specifically for workloads that stay within their platforms. Why will **Harry Stebbings** [21:07]: focus that respectfully, that's not true. Like, Ben because Benioff is saying, I wanna build customer support. I wanna build marketing agents. Benioff wants to build your full suite of agents. **Daniel Dines** [21:20]: But it's not about what he wants. It's about what customers will actually do. And why would they not do that? Because they they won't want to I will I will give you a real quote from one of our largest healthcare customer in United States. I was talking to the CIO, and he said, I will never put data from Epic into Salesforce in order to create an agent. Never. There is no chance to do this. It's as simple as this. They will they will prefer to use us to have connectors, feed agents only with the data they really need to make a decision, and then have the orchestration in an agnostic way. **Harry Stebbings** [21:58]: That's so interesting. So it's the data migration between the kind of core home of where the data is that they will not be willing to compromise on. And so what you're saying is we're gonna have a future where we have specialized agents within data repositories and an orchestration layer on top. **Daniel Dines** [22:14]: Yes. And also agents that will be on our platform for tasks that require connecting to multiple platforms. Most of RPA tasks today don't work with one single platform. They work across platforms. There are so many tasks that you will create agents for that require data from two or more systems simultaneously in order to make a decision. **Harry Stebbings** [22:40]: Can we just go into just, like, a little bit on the task side? Because we mentioned Bernie often. What tasks are we like we will meaningfully see game changing results in the next one to three years, and where will we not? **Daniel Dines** [22:53]: Instead of going to see what kind of roles do you have in your company, like you have a BDR person and so on, we go and ask what kind of processes do you run today? How does your procure to pay process look like? What are the rule based parts? What are the parts that are nondeterministic, that are good use cases for agents? And let's go and deliver those. It's the same approach that we did with RPA. That can lead to immediate successes because you actually look to smaller tasks. In a health care system, you are doing something small like processing denials or prior authorization. We will focus on these type of tasks, and we will create agents specifically for these, and we will connect them with the robots that do the the rule based task in the in this orchestration layer. **Harry Stebbings** [23:50]: What do you think is the biggest misnomer that people have about the nonrule based agent layer that they think, but they're wrong about? **Daniel Dines** [23:58]: The biggest misnomer is that agents will be good at doing rule based tasks, which they are not for sure. **Harry Stebbings** [24:06]: And that's because LLMs are creative, and they're not fundamentally good at rule based things like multiplication or **Daniel Dines** [24:11]: Yeah. Think about every time you you go to an LLM, you have a certain error rate. So let's say your success rate is 0.99, right, at every step. If you'll do a 100 steps and you multiply 0.99, you'll get to a very small number. That means that the error rate is huge for every step that you you are taking in the process. So they are not good at this. And LLMs will yield a different result every time you ask them even the same questions because they, in a way, they work like our mind works. **Harry Stebbings** [24:45]: If I'm blunt, if you already have the position in Switzerland, you already have the distribution. I think you mentioned the kind of product being more important than innovation. I think distribution is more important than product. True. If you have that position in Switzerland, if you have the orchestrator, why does Wall Street not appreciate you more? **Daniel Dines** [25:03]: Because it's early. The agentic, it's really early. And we have to deliver on the roadmap that we have. We have the orchestration engine, but we need to deliver the the agentic orchestration workflows. We have to deliver the agent builder. By the way, we have just launched it in in private preview. It's the early innings of a huge movement to adopt agents. GenAI was not so successful in enterprise so far. I think within next year, we are gonna see a lot more use cases, particularly in the Agentic AI space. **Harry Stebbings** [25:39]: Why was GenAI not successful in enterprise for the last few years? Because it's not predictable. And it will be in the next few years because? **Daniel Dines** [25:47]: Because we will put it in the context of an agentic workflow, and we will surround it with rules. And we will limit the non predictable aspect, and we will put humans in the loop to validate the outcome. It's not gonna be like a chatbot. It's gonna be the conductor will be this enterprise workflow that can be triggered by an external factor. Maybe a client will send you an email, you know, asking for a mortgage application. And then you will trigger an enterprise workflow that will call an agent that will process the application, will make a recommendation, is going to create a task that will come in my inbox. I will validate I can reply from my inbox I'm okay or not or anything. That will trigger back the enterprise workflow, that will resume and will go and will call a robot that will grant the mortgage and will do necessary changes in the banking system. This is a very different way of working compared to a chatbot. **Harry Stebbings** [26:50]: It's a very different way of working. To what extent are your conversations with customers, customer education versus selling? I think **Daniel Dines** [26:58]: we are **Harry Stebbings** [26:58]: Like, **Daniel Dines** [26:59]: do enterprises get it? I was never a seller myself, and I'm not into I'm not into selling. I'm into actually learning from customers and also educating them on how we see the world. What's the best **Harry Stebbings** [27:13]: thing you've learned from customers in the last year? **Daniel Dines** [27:16]: Well, I would say that you need to think end to end process versus thinking piecemeal, task by task. We are a company thinking, let's go bottom up. Let's identify a few tasks that are rule based. Let's put them in production, get quick return on investment, and then move on. But many customers feel you have to have the end to end picture of the process. And I think for Agentic will be truly important because I don't believe in so much in isolated agents that, you know, business user will chat with rather than these enterprise workflows that will connect agents. **Harry Stebbings** [27:52]: How does this change the structure of companies, do you think? When you think about the internal roles, functions, how people work within large enterprises, how does it change them? Do we have more people in compliance, less people in BDRs, sales, whatever? How do we think about how functions change? **Daniel Dines** [28:08]: People will become more productive. Roles will be changed more from doing things to more overseeing technology doing things, validating what technology is is creating for you. **Harry Stebbings** [28:24]: I'm sorry. Does that not sound really fucking dull? Like, we're just gonna become validation monkeys. Tick. Validate. Tick. Validate. Tick. Val fuck. I prefer doing it. **Daniel Dines** [28:34]: No. I I don't think so. These these validations will be actually mostly on the on the difficult cases in the longer term. So you ask me when agents will become, like, fully autonomous. I think there will be a progression to become comfortable with the output of an agent. **Harry Stebbings** [28:52]: Does the democratization of agents with agent builders, with the different elements that we've spoken about, does that mean that you can access a different part of the market? You are an enterprise company very much so today, does that mean that you can access SMBs in a way that you never have done before? **Daniel Dines** [29:07]: I'm not sure about it. The catch here is that building agents require some interesting skills. Today, we are at the point where we know technology works, but in order to make it working, you require, you know, very high skilled people in some particular areas. Building an agent require creating a prompt. Building a good prompt, it's actually more difficult than building a script. A script is more predictable. You have the requirements. You just follow the requirements using logic, algorithm logic. With the prompt things are not like this because slight changes in the prompt can make it working like you want or not. **Harry Stebbings** [29:52]: Yeah. **Daniel Dines** [29:53]: We put a lot of work right now into helping our automation developers building better prompts, suggesting them prompt, helping them build evaluation sets. It's much easier to test a script that works as intended than a prompt because you'll have to you the input data can make such a huge difference, and it's very hard to test it. **Harry Stebbings** [30:17]: Will we see the size of companies reduce in the future? You mentioned that we'll move to a more verification approvals based function for humans. **Daniel Dines** [30:26]: Jobs are changing over time. We are seeing in agriculture. If you think about a hundred years ago, 50% of UK or US population work in agriculture. Nowadays, it's only 2%. They are mostly people that supervise machines doing their work. Yes. Many of the jobs today will change, but they will be new jobs. Society evolves. It's you don't wanna be stuck in the existing society. This thing will create abundance. And the only way to keep with the economical growth is to increase productivity. It's with population aging and actually on the verge of starting to reduce like in many first world countries. It's only increasing the productivity. This tool is essential to increase productivity. Why we are focusing on the doomeday scenario when actually we live in one of the best world possible, when it's the rate of unemployment is the lowest in, I don't know, ever maybe. **Harry Stebbings** [31:30]: What? So I I mean, I I can pose a different argument back to you, which is that the speed of progression through this technology cycle will likely be much faster than any other technology adoption cycle. So you mentioned the farming industry. That was over multiple decades, actually, that you saw the adoption of farm machinery. Mhmm. People were able to transition. This transition **Daniel Dines** [31:50]: You underestimate the inertia of, you know, of corporations, honestly. Even with RPA. Our RPA technology, it's pretty good, and it's not fully penetrated, not at all. It's all about When do **Harry Stebbings** [32:04]: not fully penetrate, like 10%, 20%? **Daniel Dines** [32:07]: I would say it's probably less than 20% penetrated. The rest is it's not it's not an easy technology to deploy. You need to put an entire program, you know, behind RPA. It's kind of the same with the GenTKi. It's not that, you know, suddenly you have an agent that sits next to you and you show you do this and this and then you just go and they will run the job. No. It's not it's gonna take next five, ten years. With the current state of the art LLMs, it's gonna take next five to ten years to to get to very wide scale deployment of Agentic Plus Automation. If LLMs, of course, will go to AGI **Harry Stebbings** [32:47]: That is bluntly what many people are thinking, is that scaling laws will continue. We are seeing more compute lead to an equal level of performance increase and that actually there is a lot more room to run-in LLMs. Sam Altman says that we will see AGI in 2025. Well It depends. You have about fifteen days. I have **Daniel Dines** [33:05]: yeah. Look. My definition of AGI, it's a bit different than their definition of AGI. So AGI for enterprises would be when I have an LLM that has the capabilities of a guy with an average IQ like a 120 points, but predictably, Not a 180 in some fancy math jobs and 60 into, you know, other type of job, predictably. This is AGI. When we will get there, the jobs landscape will change completely, and it's not about RPA or automation. I think it's it's a lot. It's every industry will will be subject for a big change. I also believe that we need a new giant leap in order to get there. I don't believe that actual LLM's reason in the sense that I expect a guy with a 120 IQ reason. It's a stochastic engine in the end. There there is a big argument. Is the intelligence purely stochastic or it's something else? I believe there is something else in Why? I am stupid compared to LLMs in many tasks, but I am much better than LLMs in so many other. Why it's why LLM make logical mistakes that I cannot do while can do solve some problems like math, Olympiad problems that I don't have any chance to do. Why? This question has to be answered. It it shows me that actually the essence is different. It's a different type of intelligence. But that intelligence is not equipped to work in the context of business operations where you'll need reliability. **Harry Stebbings** [34:48]: I'm just actually thinking, though, you know, when we're talking about the transition of of labor forces and kind of what happens to people, Masser said that there's 9,000,000,000,000 of CapEx that will need to be invested to see AI agents fully deployed. That would then create 9,000,000,000,000 a year in GDP gains. **Daniel Dines** [35:07]: It's very small to invest 9,000,000,000,000 to create 9,000,000,000,000 a year. I think it would be an easy investment if there is a predictable outcome. But do you really think that just adding GPUs and with the existing algorithm to train, they will suddenly become become godlike intelligent? I don't understand. There are many signals that the training kind of plateaued at some it's mostly a bit of plateaued. I heard even Sundar saying recently that they are not seeing much more gains by training alone. So there should be something else. **Harry Stebbings** [35:42]: What do you think is the biggest challenge that you as CEO of UiPath have to embrace approach, like wrestle with in the next twelve to twenty four months? **Daniel Dines** [35:52]: My biggest challenge is transforming the company to be an AI first company, reenergizing our people. **Harry Stebbings** [36:02]: Did they need reenergizing? **Daniel Dines** [36:03]: Yes. We we had the rocky ride into the public markets. And probably in 2021, we were To what extent was that avoidable versus unavoidable? I think if I do a second IPO, I would be better equipped. **Harry Stebbings** [36:19]: What would you do differently if you did a second IPO? **Daniel Dines** [36:22]: I would look into into finance and go to market a bit differently. **Harry Stebbings** [36:28]: What does that mean, sorry, go into finance and go to market differently? **Daniel Dines** [36:31]: You can plan maybe a bit less aggressively but more consistently. I think it's better to plan and execute a growth of 30% year over year rather than eighty, sixty, thirty, twenty, 10. Because in a way, doing such a huge aggressive growth makes you maybe sometimes steal from the future, not knowingly, but you discover in time. It's an organic growth in a public market. It's better you worth it. **Harry Stebbings** [37:02]: Paul Graham wrote about founder mode and the incredible benefits of having founder led companies. You hired a CEO, and then you come back as CEO. What are your reflections on founder mode and the importance of founder led companies? **Daniel Dines** [37:16]: There are stages where founder's mode work better. When I hired Rob, he's actually a great guy. But when I hired him, I didn't realize that we were still at the stage where founder mode is essential. I I thought we are at the stage where we are more established, and we need we need an experience that is stage appropriated for something bigger than us. So I think for 1,000,000,000 in revenue company, it's actually it's not that established in a way. You need to get to a critical mass. I I think maybe that critical mass is Do **Harry Stebbings** [37:56]: you think that's it? I think the revenue is actually irrelevant. I think what's relevant is the technology cycle. And I think when you see Larry and Sergei, they are back at Google. Bezos is at Amazon every day again. Everyone is coming back home because founder **Daniel Dines** [38:12]: But Larry and Sergei didn't get back to the CEO job. **Harry Stebbings** [38:16]: No. But they've come I **Daniel Dines** [38:17]: I was try I actually, I never I never left the company. I ran product and engineering directly while Rob was co CEO and CEO. It's actually I always think that's harder with respect to you **Harry Stebbings** [38:30]: because you're kinda like in and watching someone else In run-in some **Daniel Dines** [38:33]: many ways, it's hard. Yeah. It's hard. I know it's a bit of a dance. Maybe I didn't speak enough to him, he didn't speak enough to me, and it could have been probably better. I agree. It's in a time when there is such a huge change in technology. I think the CEO pattern is so important because you you connect instantly product and go to market and marketing, and it's a powerful flying wheel that has to work. **Harry Stebbings** [39:02]: Both of us are not very good at fluff. I I hate empowerment. I hate alignment. I hate all the bullshit corporate words that people use which mean awful. How do you reenergize a team? How do you say, hey, you know, kind of Wolf of Wall Street style? **Daniel Dines** [39:15]: Look. My way is to to speak really transparent to the teams. This is where we are, guys. It's a lot of work ahead of us. We made these mistakes. We have to fix these things. It's not easy. It's gonna it's gonna require a serious, serious work. And maybe that's not a way to energize them, but it's a way to make them hurt. I think one of the things that as you grow bigger is that people feel smaller and smaller. I don't want this is this is hard. I don't I wanna work in a company where people have joy, people feel that they are empowered to do things, not in a bullshit alignment. But it's very important that if you wanna make a big change, go and drive a big change and have the voice to fight for it. If you are shut down, then it affects the moral more than the stock price, I think. **Harry Stebbings** [40:12]: How do you actually do that though? You say, hey, go take big bets. I support you. **Daniel Dines** [40:17]: Yeah. We are we are trying to reduce bureaucracy. We are trying to empower the regions to have more in their control. We are trying to get closer to the customer. This is how we build a lot of shit inside, taking cues from the customer up to the product. **Harry Stebbings** [40:35]: And what do you think are the biggest rules of management that are most bullshit? **Daniel Dines** [40:39]: People put too much importance on being disciplined, having regular one to ones. It just gives you the impression that you do work without actually doing work. I believe more in in understanding my priorities on a dynamic basis and do what is required to do. I don't believe in one to ones. I don't believe in one to ones with my directs. We have to call each other. If you cannot call me, if we don't have an honest, candid discussion, it's not it's not gonna work in a one to one. **Harry Stebbings** [41:13]: How many direct reports do you have? Nick at Revolut, whom we had on the show the other day, they both talked about having 45, 50. **Daniel Dines** [41:19]: I think that's a great model to have as many as you can handle. I think I have around 12. What do you think about the role of **Harry Stebbings** [41:26]: CEO you are not handling well enough today? **Daniel Dines** [41:29]: I have two ways of working with people. One, it's very direct and sometimes might be constructed as rude and tough. And one which is indirect when I avoid to really tell them what's wrong because I I don't wanna necessarily take the ball from the horns. But, actually, what what I'm telling people that I work with in a very direct style, that's a sign of appreciation. The moment you see that I'm, you know, working around and I'm finding my words, that's not really a good sign. Have you changed a lot as a leader? Man, in the essence, no. In understanding better what does it mean to run a company, of course. But in the essence, I have not changed since I was 17. And who am I? And really, my essence, I have not changed. I would have led this company as a 17 years old without the experience to run it, but may in the same way as a person. **Harry Stebbings** [42:28]: When you look back at the last few years, what did you do that you **Daniel Dines** [42:31]: wish you hadn't done? I hired for experience. I should have never done compromise on chemistry for experience. **Harry Stebbings** [42:41]: That's what you did that you maybe shouldn't have done. What did you not do that you should have done? Yeah. Should you have lent into kind of Agentic workflows two years ago or a year or sooner? **Daniel Dines** [42:52]: We should have landed into Agentic six months earlier. Yes. That that's for sure. Do you feel like you're behind? No. We are not behind. It's it's very early. It's very early, but we could have been even even faster. **Harry Stebbings** [43:06]: Ben Eel said on the show that went very viral, they are not hiring any more software engineers. **Daniel Dines** [43:12]: No. We are not hiring as well. We have repurposed a lot of engineers from products that we deemphasized into agenda. **Harry Stebbings** [43:20]: He also said that the existing software engineer cohort have been made so much more efficient from the tools that they're now using, mostly AI tools. How are you seeing efficiency within development teams change internally at UiPath? **Daniel Dines** [43:35]: Look. We've all respect to Salesforce. Our technology is much harder to build. **Harry Stebbings** [43:45]: That's like saying to someone, with respect, your child's very ugly. **Daniel Dines** [43:49]: Why is it It's in a company like ours where it's really difficult what we are building and it spans many different technologies. You will get some productivity improvement that I don't think will be gigantic. **Harry Stebbings** [44:04]: Can I ask one final one, which is a weird one? And it is personal too, it's just kind of advice, which is I invited you around to mine the other night, and you saw my apartment. You were like, wow. This is really nice, and you're 28 and, like, great life. And I actually have been thinking about that for the last few days because all I can think is, well, it's not as nice as yours. And it'd be nicer if there was a bigger kitchen, and it'd be nicer if this, and it'd be nicer if that. And I was walking in the park with, like, one of my oldest friends today, and I felt really guilty for this in my head. Because all I can think is it'd be nicer when. What do you advise me as a friend when you hear that? I've **Daniel Dines** [44:40]: wasted my late twenties, a big part of my thirties and forties, thinking in this way. It's a totally waste of cycles and energy. I was fortunate in life understand what does it mean to have it all. It doesn't fucking matter. Everything that matters is here, and it's how we think day by day. Thinking about a bigger kitchen will take cycles from reading, understanding world, understanding people, understanding AI. And you're doing we live in a pretty amazing world. Think about with almost zero capital, you can build amazing stuff, fucking incredible. I know it doesn't really matter. There is not a single thing that I possess that it's worth spending cycles wanting it, not a single one. Do you think this is the best time ever in history to be alive? I'm 100% sure. And it's the best time of my life. I feel way better than in my twenties. How so? You know, from a health perspective, from my my mindset, and not wanting things, it's so powerful. Do you feel freer not **Harry Stebbings** [46:00]: wanting things? **Daniel Dines** [46:01]: Yes. You feel free. That's that's the right word. When **Harry Stebbings** [46:05]: did when did you feel freer? **Daniel Dines** [46:06]: When you have peace and you don't want anything, but not necessarily in a zen type of mood, but in the sense that I wanna do my best, but irrespective of the outcome of my best. **Harry Stebbings** [46:20]: Is that I'm just trying to understand. Is that because the outcome doesn't actually change your life at all? Like, whatever happens now, you know, if stock price goes down 20% or up 20%, they'll have everything. **Daniel Dines** [46:31]: Yeah. But it's it's gonna affect employee morale. It's it's it's you you have responsibilities in the world. Stock price is important. Not for my, you know, wealth for how much money I can spend in this life, but it's important for what we can build, for the talent that we can attract. So it's we are part in a much bigger **Harry Stebbings** [46:55]: game. Final one for you before we do a quick fire. My brother is having a baby in a month, and I'm trying to offer him words of advice from great fathers. What advice would you give to my brother having a baby in a month for the first time? **Daniel Dines** [47:10]: I think the best is to enjoy the ride. It starts as very difficult, and it's becoming more and more enjoyable with every month that goes on every year that goes on. I said he could have two weeks off. Not not in the first two years. **Harry Stebbings** [47:31]: Listen. I'm gonna do a quick fire. You're gonna give me your immediate thoughts. Does that sound okay? **Daniel Dines** [47:35]: Let's try. **Harry Stebbings** [47:35]: So what do you believe that most around you disbelieve? **Daniel Dines** [47:39]: I believe in certain lack of discipline. It's very important in order to stimulate creativity. **Harry Stebbings** [47:46]: Where did lack of discipline hurt you? **Daniel Dines** [47:48]: Lack of discipline actually empowers me. That's an essential part of myself. Without that, I would have not created what I need. **Harry Stebbings** [47:57]: What is the single hardest element of being a CEO of UiPath that people don't see? The hardest? Yeah. Do you find it lonely? **Daniel Dines** [48:06]: The hardest part is to manage the unhappiness of people. Good news don't go to me. They are in day by day in their day by day jobs, but everything that is not working well goes up to me. So I have to deal. But that's not only being CEO of UiPath. It's being basically running any sort of company world. Do you find it lonely? I am a lonely wolf. I find life pretty lonely, man. So that's not that's not only about this job, but I I live mostly in in my head, thinking, analyzing, reflecting. This is how I spend my life. So it's not only about being CEO. Even when I was a software engineer at Microsoft, I felt You find that **Harry Stebbings** [48:56]: hard. But I actually find it hard to switch context. Again, I was on this walk this morning talking about it and talking about our conversation the other day. When I go to spend time with family, I find it hard to switch context because I live in my head, and then they're talking about TV and the weather. **Daniel Dines** [49:12]: Yeah. It's kind of hard, and I feel I am getting more and more disconnected. It's why it's hard, the more we get into our stuff, the more is difficult into providing bits of information. If someone is asking me, what do you what have you done today? I need to tell I need two hours to give you context of my job, and then I can tell you what I did today. So I don't have these two hours. So let's talk about TV. It's easier. **Harry Stebbings** [49:37]: There's I heard this brilliant statement, which is the heaviest things in life are not iron or gold, but unmade decisions. The big decisions that weigh on you, essentially. The big decision that weighs on me is should I have gone to San Francisco when I was 19, built a career there, become friends with Sam Altman and the biggest Silicon Valley founders, which I could have done, or was I right to stay here? That's the decision that weighs on me. What decision weighs on you? **Daniel Dines** [50:02]: You know, one is, should I have enjoyed life more? I didn't have enjoyable life in if you measure it by the nowadays standard. **Harry Stebbings** [50:12]: What does that mean? You weren't happy doing what you were doing? **Daniel Dines** [50:14]: I didn't live like a hedonistic life and for many reasons. Not necessarily that I didn't want, and I have a fantastic ethic about it. But it just happened that I didn't. So I had many years of anxiety. Just hard fighting with myself, fighting, you know, my way. I am wondering how it would have been to try to have fun in my twenties, in my thirties. And even today, I still I can give up on everything I do, and I can just go, you know, for a life of a lot more pleasurable things. But I don't do it, and I feel there is something inside me. **Harry Stebbings** [50:54]: Why don't you do it? **Daniel Dines** [50:55]: I I don't know. There is something inside me that is it's a more powerful aspect of myself that tells me, no. This is the right thing to do. And I always did. And it was so hard for me to work in Microsoft, especially the last couple of years, because I felt I'm doing as a compromise for myself just to be capable of building later what I want to build, but I didn't want to do it. I felt so lonely, so depressed, so far away from the world that I like it, but it was a constant fight within myself because I believe that was the right thing. So there is a quest for the right thing. **Harry Stebbings** [51:35]: When we see stories like UiPath where it's like zero to billions and everyone's like, wow, wow, fantastic. Zero to billions, that sounds great. There are times when you don't believe. When did you not believe and what got you out of that? **Daniel Dines** [51:48]: You know, it's I questioned the success most before actually raising any money. I I bootstrapped the company for ten years now. Those ten years were the most difficult. The the moment I raised some capital, I got some people trust in me. Something changed in my brain, and all the fear disappeared. I just went killing it. It never kind of stopped in a way. I mean, the most terrifying moment was, like, March 2020 when the with the onset of COVID when I was thinking, maybe the world is gonna stop. We will have to basically fire, you know, almost everyone in the company to survive. We were still bleeding money. So it was a very hard moment, but it it passed really quickly. We realized that actually it's **Harry Stebbings** [52:37]: How do you deal with moments of intense stress? **Daniel Dines** [52:39]: Look. I have my own sorts of coping mechanisms. I like to write poetry. This is a thing that makes me forget a bit the source of stress, and it dials into a creativity aspect. Stress is painful. You can put this pain into words. Finding the metaphor somehow alleviates for me this type of pain. **Harry Stebbings** [53:06]: God, that's a cultural thing to do. I was gonna say something else. We're doing damn better than shit. Do you do, man? Do you know what? I write poetry, actually. Yeah. No. I I don't write poetry. Okay. You can buy OpenAI at $1.60. Can buy Anthropic at 40, or you can buy x dot ai at 50. Which one do you choose to invest in at those prices? **Daniel Dines** [53:29]: Probably Anthropic. I think the upside is bigger. **Harry Stebbings** [53:32]: The upside given the entry price at 40? Yeah. **Daniel Dines** [53:34]: If I have only these three choices Yeah. **Harry Stebbings** [53:36]: You can be CEO of any other company for a day. Which company are you CEO of? Maybe Paystex. Final one for you. UiPath in five years' time, what would make you happy if I showed you a a chart of UiPath and what it is in five years' time? What would be happy? **Daniel Dines** [53:53]: Very few companies actually get to have a second act. And what I realized, it's it's it's actually kind of very hard. Unpack that for me. You you know, as a startup, you become successful understanding a space very, very well. Getting into another space for a second act, it's kind of very hard. It's not enough to have capital, to have good developers. It requires, in a way, even a high degree of luck to get the second act. And I think very few companies are very successful in getting a second act. I hope that we are we are getting our second act in the agentic space. It's seriously the biggest opportunity for that is in front of us, and it's gonna change the company. **Harry Stebbings** [54:39]: Daniel, I'm so grateful for the friendship. I'm so grateful to you for joining me again. I've learned from you in so many ways, and I really reflect on our conversation. So thank you so much for doing this, and I've loved having you. **Daniel Dines** [54:51]: It's absolutely likewise. Harry, it's so fun being with you, man. Not only on the podcast. **Harry Stebbings** [55:01]: I have to say, I just love that show with Daniel. He is one of the great CEOs of our time. And if you wanna watch the episode in full, you can find it on YouTube by searching for 20 VC, that's two zero VC on YouTube. But before we leave you today, ## Sponsor read **Harry Stebbings** [55:13]: what do Henry Ford and AI have in common? Neither could change the world without automation. In the future, there will be two types of businesses, those that have automated and those that wish they had. UiPath, the undisputed leader in automation, is taking us into the era of Agentic Automation. UiPath's new AI agents don't just follow rules. They think, they make decisions, they work alongside the world's most powerful software robots already trusted by over 10,000 businesses. If agentic automation sounds new, just think of UiPath as more growth, not more overhead platform or happier customers, happier employees platform. Whatever you want AI to do for your business, Agentic Automation with UiPath will make it happen. Try UiPath's new AI agents for free at uipath.com. The future of automation is both agentic and robotic, so don't get left behind. And speaking of transforming your business with cutting edge technology, let's dive into another game changer, Atio. Atio is the next generation of CRM. Setting up Atio takes less than a minute, and in seconds of syncing your email and calendar, you'll see all of your relationships in one place, all enriched with very valuable data. Atio also lets you build Zapier style automations, automations, gives you powerful reports, and works perfectly for any go to market motion from PLG to sales led. And Atio is designed for the next era of companies like yours, and companies like yours shouldn't have to deal with inflexible one size fits all CRMs. So join industry leaders like eleven Labs, Replicate, Models, and more to scale your startup beyond the next level. Head over to ateo.com/20vc, and you'll get 15% off. That's 15% off your first year at ateo.com/20vc. And if ateo helps you stay ahead by streamlining your relationships and operations, today, I want to talk about a venture fund making waves with its unusual model. I'm talking about the Fundrise Innovation Fund, which is democratizing venture capital as a public venture fund. For example, most of the AI revolution is being built and funded in the private markets. Companies like OpenAI, Anthropic, Databricks, these are incredible multi billion dollar companies, but they're inaccessible to 99% of investors until they go public. Well, those days are finally over. Visit fundrise.com/20vc to check out the Fundrise Innovation Fund's impressive 150,000,000 portfolio for yourself. Carefully consider the investment material before investing, including objectives, risks, charges and expenses. This and other information can be found in the Innovation Fund's prospectus at fundraise.com/innovation. This is a paid sponsorship. As always, I so appreciate your support, and I can't wait to bring you an incredible 20 growth on Friday.