UiPath, Inc. (PATH) Earnings Call Transcript & Summary

Thank you all for joining us. We'll kick off the final session of the day. So Daniel and Ashim, thank you very much for joining us from UiPath. My name is Keith Weiss. I run the software equity research group here at Morgan Stanley. I'm going to make Kelsey a little bit mad here. Having gone through all these sessions and talking with private companies, we've had some really good kind of foundational conversation. Like starting from the ground up to understand these companies and then kind of dig in from there in terms of sort of the stories around these companies, and so natural debates are going to emerge. So I want to do a similar thing with UiPath. Even though you guys are public and the story's a little bit, well, better known, I think it's instructional to kind of start at that bottom. So Daniel, maybe you could talk to us about sort of what was the fundamental problem that UiPath was kind of built to solve? Like what was the market opportunity that you saw? Like, listen, here's a huge inefficiency in the marketplace, and I think we could solve it with this.

Thank you. Thank you for having us, first of all. It's really fantastic to see you in person and seeing all the people here in person. Look, I think, I'm a software engineer, and I used to work for Microsoft. So in our world, everything was automated. So as a software engineer, I couldn't believe how many inefficiencies and manual repetitive processes are out there in the businesses. In the beginning, when we build our technology, we didn't think this is going to apply to this type. We thought it's more like an IT thing. It's not -- it doesn't have such a huge TAM. But this is how businesses build processes. Processes are built by business people, and they are not built for automated consumption. They are built for human users. And our approach that emulates people is the only one that can go after the long tail of processes. And they are never going to disappear. There is no ideal world where a company will be -- will stagnate in time when all the business processes will be perfect. This is not the reality. So they will always build these processes. Business people will continue to build processes. They will acquire companies. They will integrate systems. So this type of automation is the only one that really go after the long tail of opportunities.

Got it. So you look into a corporation, you see the reality sort of how they run their business and how they run their processes, and there's still a lot of manual task work. And manual task work that was based on human interaction because that's how the systems were built. So you create the bots to be able to automate those tasks and create efficiencies where there was these manual processes. And you're talking about like that huge TAM. Like where did you first see kind of the adoption? Where did the bots really resonate first? Maybe for Daniel. Then for Ashim, like when you extract that out and you think about that core RPA TAM or robotics processing automation TAM, what does that get you to when you're just thinking about that core solution?

Well, our first big process that we've seen automated was in supply chain. So it was a big computer manufacturer that have like an 800-people delivery center working there and automation was the really sweet spot. But from that one, we really went much bigger into the finance departments of various clients, especially in the financial services sector. And then our technology is pretty horizontal. So everywhere there are humans doing something, we can apply it. So it's financial sector, it's an entry point, but we work across all industries, particularly in health care, in public sector, transportation, retail.

Got it. And then Ashim, when you talk about the TAM for core RPA, how do you guys size that up? How do you guys think about sort of the market opportunity ahead of you?

Yes. So we actually triangulated it multiple ways and independently verified as well. So we've quantified the market at $60 billion-plus. And we did it very simply. We took our customer base, segmented our companies by the number of employees because that's relatively correlated to complexity and number of processes that are there. We took the top 10% of our customers and said, if every customer in the market could reach these levels of our top 10%, you do that simple math and you get to this number of $60 billion plus in terms of a total available market. Qualitatively, when you look at it, I live the TAM because I was a customer. So it was the only piece of software, when I was a CFO or a CIO, that could solve problems for legal, HR, finance, supply chain. So it's really relevant to every single department, every single process and every single employee. When you look at it both qualitative and quantitatively, you can see the massive market that it really is.

Got it. Got it. [ I wanted to do ] is a kind of market sizing exercise of like a bot on every desktop, right? That given that everybody is dealing with multiple systems, everybody has tasks, do you think that's still like a viable formulation of that, there can be a bot on every desktop in terms of understanding market opportunity?

I think so, but this is not the entire market. So our business is comprised of unattended automation that runs in data centers completely without any human supervision and bots that work on people's desktop, on employees' desktops and help them with their routine day-to-day tasks. When I say a robot for every person, we meant really half of our business, the attended automation. Together, they are combined in what we call the fully automated enterprise.

Got it. So when we looked at total available market, we looked at it -- we took your perspective and RPA today, $2 billion market opportunity -- or $2 billion spend today against a $60 billion market. So less than 5% penetration, right? When we looked at it, we thought about information workers, right? And 400 million people pay for office, right? They're 80% of the market, there's probably 500 million people out there who would be willing to pay for software, right, information workers. And if I run the math, and I think the average bot cost about $5,000 against that $2 billion, there's less than 0.5 million bots out there. So 0.5 million bots out of -- against 500 million potential information workers, even on that half of your business, would suggest even a larger TAM, if you could kind of hold that pricing and even a smaller penetration. Anything in that equation that you would significantly push back on?

Listen, we don't comment on pricing and how we would do that. So that's just 1 factor. But no, listen, everybody who we've talked about with the TAM, who's understood our software, from our customers to investors, all have kind of agreed that the market is probably larger than the initial sizing. We obviously have to put some mathematics to give that credibility. But again, when you look at the qualitative and you look at calculations like you did, even if you look at 1 billion knowledge workers that are out there and you look at the number of processes they deal with, the number of applications they deal with, and just extrapolate from there the number of robots that are in demand, you can see the market continues to be bigger and bigger from an opportunity standpoint.

Got it. So a huge inefficiency out there in terms of how people do their processes, way too much manual work, really large market opportunity that's well underpenetrated. The last kind of component is the effectiveness of the technology. So can you talk to us about proof points on the effectiveness of the technology? The type of ROI that your customers are seeing by deploying these bots? The type of efficiencies that could be pulled out by using a bot or using UiPath more broadly?

Yes. This is the type of technology that is really meant to get huge return on investments. This is not a kind of a personal productivity space. This is an enterprise-grade software that when deployed, provides enormous return on investment. We have customers like SMBC Bank that said that the return on investment is $0.5 billion. So this is why particularly we have not seen really pressure on the pricing of our software. With pricing in the grand scheme of total cost of ownership, it's only maybe 5%. So it doesn't really matter.

Right. Got it. So a very effective solution. Your customer's seen tremendous return on investment into the overall market opportunity. But the UiPath story doesn't stop at the bots. And I think that's a part of the broader story that's underappreciated, is that you guys have built on significant adjacent functionality both organically and inorganically, bringing on process mining and task management. You have StudioX, which gives you kind of low-code development platforms. Can you talk to us about the broader suite? Like you have the units of automation that takes place on the desktop or unattended in the data center. What's the suite that you built around it to create what's kind of units of automation to create a system of automation, if you will?

Well, I like how you said it, the system of automation. It's...

It's an equity research gig. If it works out, it becomes [indiscernible].

No, it's a good line. I -- we have built a platform because our customers requested it. And in what way? In a typical enterprise, there are multiple personas that are required in order to automate the process. So starting from the business users that do day-to-day this process, and they can run now the automations on their desktops. And then it's the process analyst that uses Process Mining and Task Mining. And then it's the people in center of excellence that understand, supervise the program. And the IT and the developers that implement this. So when you build a platform, you truly have to address the needs of different personas. So this is how we started in our thinking. And now our platform fundamentally has 4 big pillars. The first one is process discovery. And we are the only company that provides an integrated way between Process Mining and Task Mining. On Process Mining, it's the ability of understanding back-end system, logs, connect to the transaction log and figure out the processes. While Task Mining is really the ability to understand what people are doing when they have desktop in order to produce these transactions. Think about most of the tools you are using on your computer don't produce an audit log. Excel doesn't produce an audit log, Outlook doesn't produce it and so on. So combining these 2 technologies, you really understand the processes at a very greater level of details. So this is one pillar. Second pillar is the implementation phase, the run phase, the orchestration phase. And for us, it was very important that we have tools for professional developer. Put yourself in the shoes of a professional developer, you need to use the best tool out there. Your time is much more important and costly than the tool itself. So you need power. But put yourself in the shoes of a citizen developer. You need the easiest tool to use. You need something that address your immediate need. You have -- you've seen an opportunity and you see kind of automation, you need to implement it. And then we have the analytics pillar that help us -- our customers to measure the impact of automation and actually help them to reinvest back the dollars that they saved into. So it's really powering the flywheel of automation. And the last part, which I think it's really important and maybe not so well understood, is what we call the engage layer. And this, we believe that automation is becoming a new layer in the enterprise stack that will sit above applications. And engage for us is the ability to connect people to automations, to robots. So we built this user interface layer that makes it much easier to combine data from multiple robots, present it to people or having people commanding different types of robots to do task on their behalf.

Got it. Got it. So there's 2 areas that I want to dig into. One is on sort of the front end, the process pillar. Again, I think it's early days in terms of people understanding sort of Process Mining versus Task Mining. And I think there's perhaps an artificial kind of perspective of either got to be top down or bottoms up, right? You have like the Celonis view of the world, if you will, of hey, listen, really, it's just about creating that map of your processes. And then with that map, you could find the inefficiencies. But it seems like your perspective is you need both, right? You need the map of the top-down processes to understand where inefficiencies might lie. But really to understand how to automate those processes or where that inefficiency lies, you need to be on the desktop. You need to have an agent there understanding the work that's being done to really target that process. So it's really the integration of top-down and bottoms-up that creates the accuracy of where to point the automation.

Yes. Because in the end, what is the best outcome of Process Mining exercise? We believe this is automation. So you'll find automation opportunities, but in order to do the real automation, you need to understand what people are doing to create those transactions. Task Mining is giving you the view in what people are doing. We can, even from Task Mining, we can even create some kind of an automation skeleton that you can pass to the automation developers, and they will refine it until it works exactly like a human user.

Got it. So once these inefficiencies are found, our view on the market is that people are going to use the tools like the Process Mining and the Task Mining to find where these inefficiencies go with the goal of automating them. And there's going to be several technologies used for that automation. And RPA is going to be one of those technologies. API management is going to be another one of the technologies, workflow automation or low code development is going to be another one of those technologies. So 2 questions. Like, one, do you agree with that world view that there's a kind of a tool set of technologies used for these automations? And how does UiPath plan to sort of work, if you do agree, kind of work in that world?

Yes. We do agree that...

It would've been a very awkward question, if you didn't agree.

Yes. No, we do agree, and we are seeing consolidation across automation, API integration and RPA. And we play basically in all of the 3 categories, but with different approaches. And from our perspective, in order to -- if you look at the palette of use cases, there are different approaches for different type of use cases. If you look at the long tail of use cases, the only thing that works is emulation. If you look at the processes with a high volume of transactions, you need to do API integration. If you look at the new processes when people are building new approaches, new applications, this is where low code is. So I think this is quite important to think through the lenses of our world. Because you'll understand if you go to the long tail and if we talk about emulation, emulation is a combination of user interface automation, replicating what people are doing, but also API integration. Because some of the actions that you do, like sending an e-mail, are much easier to do via an API automation than just replicating what you all do. So we are all about emulating people the easiest way to do it, UI plus AI plus API.

Got it. And Ashim, to expand the conversation to you. From like a pricing perspective, the world was a lot easier when you were just selling bots, right? You have unattended, you're not going to tell me what the pricing is, but we have an idea. But now you have a broader suite. Now you're talking about Process Mining, Task Mining. Now you're talking about API management. How does that change like the pricing conversation and how you're trying to sort of match value with your customers?

Yes. One is, as we roll out new products, we are always evaluating what is monetizable and what should be a core part of our offering that is already packaged in that, frankly, widens and deepens our competitive moat, which is why our win rate is so high, why we continue to outperform in terms of all of our key metrics. In terms of pricing, we have -- we've actually kept it relatively simple. For things that are user based, we have user-based pricing. For things like Process Mining, we have server-based pricing that is there because it's not fully -- it's not user-based. It's around something broader. And then what we -- what I think has been an area that we've innovated very well on is our ELA structures. So when you look at it, there's a reason why we now have 1,000 customer -- 1,000-plus customers above $100,000, 100-plus customers above $1 million. ELA when they're buying the entire platform, that gives them a chance for the right economic value to take advantage of the full automation suite. And the breadth of our platform has allowed that ELA to become even more attractive to our customers.

Got it. Daniel, one more kind of high-level fundamental question for you. You talked about this abstraction layer that sort of automation presents above the applications where you have visibility into the actual work that's being done, the processes and sort of mapping into the processes that exist within the company. It seems to me that there's a tremendous amount of data that is coming from that visibility, that control point that you guys have that could be useful in optimizing those process or potentially in sort of ML/AI type scenarios to create insight on how to sort of better run those processes or better predict the outcome of those processes. Is that a future opportunity or a current opportunity for UiPath? And how do you guys look to address that?

This is clearly a big opportunity for us and plays very well into the new direction that we announced for our platform. And this direction is semantic automation and if you want semantic AI. But right now, the robots don't understand what they are doing. They don't understand the data. They don't understand the documents, not even the process, what they are doing. We aim to bring more understanding into -- and this is why when you have understanding of what data you submit, you don't submit like an array of raw data to your analytics. So you have to do a lot of work to figure out what it is. But if the robot understands it's submitting customer data, patient data and the job in the analytics is much easier. And they can build a lot faster machine learning to understand and predict how the processes are running.

Right. Got it. So that ends the easy part of the session. Now we're going to shift into the hard part of the session. So of course, that was a story that I tell to all of our investors, but I get pushback, I get debates. And I wanted to get the response to the debates that I get from you guys. And I'll just present them to you straight up. Of course, understand this isn't me talking, this is the investor debates out there. So #1 investor debate that I hear about is RPA is a transitional technology. It's not a fundamental technology that you should be building systems on. Or said another way, RPA is like duct tape. You use it to sort of fix problems, but you don't really want to build a house out of it. How do you respond to that critique?

Well, I have even worse, like lipstick on a pig. Look, I think the easiest way to debunk this is by looking at our numbers. So we have 98 dollar-based retention, which I think it's really amazing and it speaks volume that this is not a transitory technology. If we look at our early customers, at early cohorts, they grow at an amazing rate. Our net retention rate is one of the best in the business. And now let's go to fundamental. Why is it so? Essentially, really, our approach is the only one that goes after the long tail of automation. There is no other feasible way. Look, when you integrate 2 systems, let's say, you take Salesforce and NetSuite, like in our own case, we are big users of our own technology. You integrate kind of the biggest pain points between these 2 systems. But around the age is there are a lot of workflows that cannot be integrated. It's simply, you don't have enough resources to do it. So people are using, customers are using people around these ages. We go around these ages and we convert the manual work into software workflows. This is never going to change. Think about API integration or low-code, no-code. It's like I am building like a rail roads between 2 big cities, but when you go to the station, you need to take a car. We have the car in this situation.

And Keith, just having been a customer, what is transformational about UiPath's platform is that it can transition so quickly. How many customers spend $50 million, implement a big application and then the business process changes the next day? Or you do an acquisition, regulations change. There's very little software that can actually be implemented in 2, 4, 6 weeks, have that 6-month payback and be transitional to the point where companies are transitional. And that is a huge value add for our customers and why it's so sticky. Because the business environment is constantly changing, you need a flexible deployment or a flexible application.

Got it. It's a good answer. So debate #2 that I hear most often is the competitive debate. And honestly, it's not about automation anywhere. It's not about Blue Prism, the guys that are in your core market. It's about Microsoft, number one, that with their Power Platform, they're just going to run you over, they're going to crush you on price. Tableau is to Power BI what UiPath is to Power Platform. How does UiPath stay ahead of Microsoft? How do you maintain pricing and keep that innovation gap versus what Microsoft is able to bring into the marketplace?

Well, you know that I used to work for Microsoft.

Yes.

So -- and I have high respect for Microsoft, especially under Satya. It's a great company. And we don't compete really with Power Platform. Power Platform, it's a big animal. We compete within -- not even with Power Automate, really. Within this big platform, we compete with a product called Power Automate Desktop, which Microsoft bought 18 months ago from a Greek company. It's a product that was never really proven in large enterprise deployments. It's something that Microsoft sees more like in a personal productivity gain. So now, if you look at our business, how it is today, we have unattended space, we have attended space. For unattended space, you need a professional tool to -- for professional developers. This is not -- Microsoft doesn't even compete in this space. We've never seen them in a competitive situation for unattended robots. So this is half of our business. Now when you speak about attended automation, what this citizen developer tool that Microsoft has competes only for like 1 in 10 of the use cases out there. But what is even more important, when you go head-to-head in terms of products, Microsoft's strategy, at least until now, is to build something that works well within Microsoft ecosystem, like in-app type of automation. Microsoft Power Automate Desktop doesn't work well with SAP, with Salesforce, with ServiceNow. So it works within Microsoft Office technologies. But these are very few use cases, low-value use cases. In most of the processes that we are automating, we are seeing like 3 more, 3-plus different types, big categories of applications. So at this point, in reality, I think people are more scared about what Microsoft can do in our space than it's really the reality. And there is also something that is very specific about this technology, which is not specific in case of Tableau or in case of Slack. This is an enterprise technology that generates huge return on investment. License pricing, if you use the technology that costs you $5 million a year and you generate $100 million in return on investment a year, would you take a technology that cost you $1 million and will generate only $20 million? I don't think. So it's a very -- and we can make a very simple return on investment, total cost of ownership, time to value. It's kind of easy because look, when you go to a customer and I -- and they go to automation, they have to do something meaningful and measurable. So we have the -- use Power Automate, use UiPath, automate this process, put whoever you want to automate and if they see -- Microsoft sometimes can struggle 3 weeks to do something we can deliver in 2 hours. We've built this for many years. It's not an easy technology to build. So then the return on investment case, it's so clear.

Got it. That makes a ton of sense. Last kind of big debate that we hear. UiPath isn't a cloud company, right? In software today, we just love cloud companies. We don't like anything else. And UiPath doesn't have a cloud story. There's little asterisk that like I'm going to pin my question to. Like, I don't really quite understand if UiPath needs to be a cloud company because like especially on the attended side of the equation, so much of it takes place on the desktop. But what's the cloud story at UiPath today? How is it evolving on a go-forward basis? And do you guys need to be in the cloud?

We are a cloud-first company for many years. One of our pillar of building an ecosystem was to launch our orchestration layer in the cloud, in the multi-tenant cloud offering since the beginning of 2017. So we were really the first company that was cloud enabling, a multi-tenant solution, our offering. So it help us to build a huge ecosystem of practitioners and SI partners. Now we really deliver on the cloud. Every couple of weeks, we deliver a new release in the cloud. This is how all our new products are built in the cloud. And right now, what we have announced at our FORWARD event that happened last week was that we have achieved parity in deploying on-cloud and on-prem. So all our technologies that were built on cloud now are available on-prem. So we deliver -- our common denominator is the Kubernetes. So we deliver on the top of Kubernetes. Even more, we are the only company that is now capable of running automation on Linux containers. It's really the only RPA company that runs automations on Linux containers. You cannot be a cloud-native company without supporting running automation on Linux. And this is important because some of the mother companies that we are working with like CrowdStrike or Uber or Facebook are completely based on Linux, their infrastructure, and running on Kubernetes and containers is really the only way to achieve massive scalability and auto scalability.

Got it. Got it. Any debates that I'm missing that anyone wants to bring up? All right. We got one in the back.

[indiscernible]

Sure. So on the competitor side, I narrowed it down too much to Microsoft. Investors are wondering about ServiceNow buying Intellibot and Salesforce -- and their Celonis partnership. So now they have the RPA part of the equation, and they're partnering with Celonis on process mining. And then Salesforce recently made a small acquisition in the RPA space as well. ServiceTitan, was it? Servicetrace.

Servicetrace. At least until this point, we really don't compete with ServiceNow or Salesforce. We simply don't compete with them at this point. And to me, what they built, it's kind of they sprinkle an RPA term, but buying some kind of cheap companies. And it's mostly used for like in-app automation, which like in the case of Microsoft, this is -- this doesn't have too much legs to go over. So we -- in case of ServiceNow, it's also even more starking difference between our go-to-markets. They are primarily an IT shop. We sell primarily to business lines, to CFO, to operations. So I see clearly that it's not conflicting. We are actually customers to each other, we and ServiceNow. So we more partner than compete. We don't compete right now at this point in time.

And Keith, just one thing is RPA 3 years ago, people like you were talking, like you mentioned, didn't believe in it. Now just like every application comes with a BI or dashboard or an analytics, nobody says I don't have analytics in my application. Every single company, every single application in their app is going to have some Workbench related to automation that will be called RPA. That is not the lion's share of the $60 billion market that we talked about. That is solving hard problems, taking things like moving from treasury codes, from the web, to Excel, calculating something, going to SAP, going to Kyriba, going and doing a reconciliation for an auditor. So in-app automation is a natural progression. And frankly, the trend that UiPath and others created, that's not a direct competition to us today. And then the Celonis partnership, I know we talk about, but observability, Daniel can talk about it, I think there's obviously use cases. But again, having process discovery linked with an automation platform at scale, that is the difference with our platform compared to any of the other areas that are out there, including the partnership that was created.

Yes, I think it makes sense for ServiceNow as well to use process mining technology for the sake of workflow automation like we use it for the sake of task automation.

And it makes a lot of sense for them to look to Celonis because ServiceNow is a top-down type organization. They sell to the CIO, and Celonis is that top-down perspective for the CIO as well. I want to make sure we touch on some of the partnerships that you guys recently announced because there's some really fascinating ones. Like the one with CrowdStrike, I think, is super interesting. You announced something with Snowflake. Particularly on CrowdStrike, security is usually such an afterthought for all of this. Like we're going to build this whole system. We're going to have it running significant processes, have really key data. And then afterwards, you're going to be like, oh, wait, we should probably secure it. But it looks like you guys are looking to sort of lock that down earlier in the process through partnerships like the one with CrowdStrike.

We work with a lot of customers in regulated industries, and security is paramount to them. Not that our technology is not secure. We build secure by default. But CrowdStrike gives the ability to have like this endpoint security also for robot-run processes, which I believe is really important because you get this runtime type of security. The robot opens some malicious e-mail and then this is when CrowdStrike comes into picture. I think it removes really some of the concerns of the -- of our customers. What if the robot runs into this malicious e-mail versus the person? So it was a really important move from us.

Right. And can you talk a little bit about the announcement you guys made with Snowflake?

Snowflake is a big technology partner for us. And speaking about Celonis, in particular, I think it addresses one of the few points where Celonis had the lead against us, and that was in scalability. Celonis was a technology that scale better than us on hundreds of millions of transactions, analyzing and ingesting them. But now we have re-platformed our process mining on the top of Snowflake. So we have achieved better scalability, better ingestion. And we combine the scalability of Snowflake with what was our strong point, and that was the visualization engine, which everybody thought it's superior to what Celonis is. So now we have better scalability and an amazing visualization engine. Plus, we are using Snowflake for our internal telemetrics, infinitely scalable and we are using Snowflake for our analytics pillar. So all, really, it's becoming really our data platform.

Got it. So look, we're over time. And I'm going to be on the hook for all the overtime we're paying these guys if we don't wrap up soon. Any last questions from the audience? I just want to wrap up on -- you just came out of your user conference. Some of my key takeaways, the partnerships I thought were super interesting. The big picture perspective that you guys presented in terms of like what I'm calling the system of automation, I thought was super interesting. Any other kind of key takeaways that we should take away from -- that's horrible -- key takeaways from the conference last year that you want to leave us with -- last year, last week?

Well, we realized once again that nothing beats in-person meetings. And especially in enterprise software, getting the trust of people is important. But also, we have been reenergized by the interactions with customers and partners. It's really powerful to see the energy of that, seeing, looking into the eyes of customers and seeing how they are using the technology is pretty amazing. We had Amazon on the stage saying, we have used your bots to hire 1 million people since COVID, 1 million people. This is -- also, it's insane when you look at the scale. And then we have a partner like Accenture saying, guys, we doubled down our investment in you. We are adding another 10,000 developers specialized on our platform. We are becoming the top 10 type of practices for many major SIs. So all of this, this -- again, I cannot tell you how important. I think it was a big, bold move from us to do it in person. We had our doubts, maybe what is going to be a disaster. But it turned out really, really solid. And I think it even accelerated visible of the pipeline for the rest of the year and into next year.

Outstanding. That was great. Thank you guys for joining us. It's been a great conversation, super exciting story. I'm super excited about following the progress over the next couple of years. And again, thank you so much for joining us.

Awesome.

Thank you.

Thank you.