UiPath Fusion 2025 | Dana Forfa & Hitesh Ramani, UiPath

Clips
More from UiPath Fusion 2025

Dana Forfa

VP, Gobal Head of Procurement, Real Estate & Travel

UiPath

Hitesh Ramani

Chief Accounting Officer / Deputy CFO

UiPath

From RPA to agentic AI, enterprises push automation’s next frontier

Robotic process automation, or RPA, is evolving beyond repetitive task automation into a new era of agentic intelligence, where bots and agents work together to enable enterprise-scale automation strategy.This next chapter builds on the foundation of RPA. It extends into judgment-based work that demands governance and oversight, according to Dana Forfa (pictured, left), vice president and global head of procurement, real estate and travel at UiPath Inc.UiPath’s Dana Forfa and Hitesh Ramani talk with theCUBE about the evolution from RPA to agentic AI.“I think something interesting is that what I’ve heard … is that a lot of people were concerned about technical debt with RPA,” she said

play_circle_outline Introduction of guests Hitesh Ramani and Dana Forfa at UiPath Fusion 2025.

play_circle_outline Transforming Procurement: Agentic AI Strategies, Customer-Centric Approaches, and Automation Hubs Driving Innovation in Real Estate Management

play_circle_outline Hitesh’s role in agentifying the entire UiPath corporation’s financial processes.

play_circle_outline Challenges faced with agent reliability and the need for governance.

Info
Transcript

Dana Forfa & Hitesh Ramani, UiPath

Dana Forfa

VP, Gobal Head of Procurement, Real Estate & Travel UiPath

Hitesh Ramani

Chief Accounting Officer / Deputy CFO UiPath

In this UiPath Fusion 2025 segment, Dana Forfa, VP & Global Head of Procurement, Real Estate & Travel at UiPath, and Hitesh Ramani, Chief Accounting Officer & Deputy CFO at UiPath, join theCUBE’s Dave Vellante and Rebecca Knight to unpack how UiPath is evolving from RPA “act one” to an ROI-driven “act two” of agentic automation. The discussion dives into how proven robots (the “doers”) now pair with thinking agents orchestrated by Maestro to transform end-to-end processes with human-in-the-loop governance. Forfa explains how procurement prioritizes ideas that... Read more

explore Keep Exploring

What is the agentic strategy discussed by Hitesh Ramani and Dana Forfa during the UiPath Fusion 2025 event? add

What approach is taken toward prioritizing procurement in relation to agentic AI and organizational impact? add

What transformation strategies are being implemented across departments in relation to automation and manual processes? add

What is the potential future role of agents and bots in the workplace? add

bolt Powered by CUBE AI

Dana Forfa & Hitesh Ramani, UiPath

search

Rebecca Knight

>> Good afternoon, everyone. Welcome back to theCube's live coverage of UiPath Fusion 2025 here at the Wynn. I'm your host, Rebecca Knight, alongside Dave Vellante, my co-host and analyst. I would like to welcome two guests for this next segment, Hitesh Ramani, Chief Accounting Officer, Deputy CFO at UiPath. Welcome, Hitesh.

Hitesh Ramani

>> Thank you. Thanks for having me here.

Rebecca Knight

>> Dana Forfa, VP Global Head of Procurement, Real Estate and Travel at UiPath. Thank you so much for coming on, Dana.

Dana Forfa

>> Thank you for having me.

Dave Vellante

>> Thanks guys.

Rebecca Knight

>> I want to start out by asking you about your agentic strategy, but also within your swim lane. Dana, why don't we start with you?

Dana Forfa

>> Sure.

Rebecca Knight

>> Talk a little bit about how you approach agentic AI, particularly when it comes to how you prioritize your procurement.

Dana Forfa

>> Sure. Well, I lead procurement, but also real estate facilities and travel. I am fortunate to have the trouble of having so many ideas from my enthusiastic team and we excelled at RPA. I'm happy to say that. When agentic came into play and the opportunity was there, they were hands raised, we actually had the first idea submit in automation hub when it was open for UiPath ideas. Happy to say that. With so many ideas, you really have to look at the same thing and the same way I prioritize RPA, to be honest. What would have the most impact across the organization? In procurement real estate and travel, the common thread is that we are customer service oriented. We are here to support the organization. We wanted to look at what makes it smoother for UiPathers, what is simple and what allows the team to really partner with the org. Let's take what we did with the mundane, take that to the next level, and really come up with some great solutions for the org.

Rebecca Knight

>> Okay. Similar question for you, Hitesh, in terms of how you prioritize for finance and accounting.

Hitesh Ramani

>> Great. By the way, I also have a larger responsibility at UiPath. I'm responsible to agentify the entire corporation.

Rebecca Knight

>> Oh, okay. Just that little thing. Okay.

Hitesh Ramani

>> Yeah, exactly.

Rebecca Knight

>> Hopefully you cross that off your to-do list after this.

Hitesh Ramani

>> The interesting part was in our act one, most of our automations were done as part of, I call this necessities mother of invention or innovation, but in our act two, our agentification is based on the ROI. We are really focused on those areas where some of our processes are still manual in nature, just because the technology was not ready for us. Now, that the agentic solutions and the orchestration through Maestro is available, we've expanded that and across the corporation we have engagement from legal, from marketing, from finance, even our customer service. We are seeing that we have opportunities across the board. In terms of prioritization, we are really focused on which are the natural human manual touch points that we can transform. That is where we are really focused on. Within finance, we are seeing it across the board.

Dave Vellante

>> Dana, you talked about the similarities between the RPA rollout and agentic. What are the differences? Because RPA in the early days, a lot of back office, and then you guys established proficiency.

Dana Forfa

>> That's right.

Dave Vellante

>> Then it was, I'm sure it was a good example for customers. Probably the same with agentic, but AI seems to be much more ubiquitous throughout the organization.

Dana Forfa

>> That's right.

Dave Vellante

>> Moving much faster, maybe less of a pull and more of a push from top down. I wonder how that dynamic has affected, not so much internally, but externally when you get called on salespeople or peers to help them out.

Dana Forfa

>> Yeah. Sure. I think something interesting is that what I've heard, not just from customers but from my fellow procurement leaders, is that a lot of people were concerned about technical debt with RPA. Is RPA going away? My team has used RPA as the building blocks for agentic. Our agents are actually using our robots. We have just continued on the goodness that we've already built, which I think is really exciting. To your point, what is different now is that I can give you an example, we had a new agent join our team yesterday. This agent is helping with one of the worst tasks my team had, which is the goods receipt process to clean POs at the end of the quarter. Hitesh knows about this because someone from my team is often bothering him about this. Now, the next step is that the agent is actually going to give recommendations to the business. Whereas, RPA was doing, we say robots think, and so they will now go to Hitesh and say, "Hitesh, you need to complete this. Here's what I recommend." Based on all of the information it has, it can say whether or not it's a good idea to approve that good receipt. If Hitesh says, "No, thanks, I don't want to do that. I want to dispute this." It will give him prompts of how it can solve for that and take care of that for him, and then Hitesh can just pivot back to his next meeting. I think it's really exciting to be able to talk to customers and offer these solutions to consider and say, what's the feedback you still have from the organization about things that they aren't seeing in the background perhaps, and how now this agent can take what we've already built and bring it, make it more real for our internal customers.

Dave Vellante

>> What's interesting about that is you talked about the robots as the doers, but agents, the very word agency, people think of giving the agents the right to do or the ability to do, but they don't trust them. What I'm hearing is you trust the bots because they're secure-

Dana Forfa

>> They're proven. Absolutely. Yeah.

Dave Vellante

>> They're proven and they're good doers, but the agents can think. Then, is that how this is playing out? Is, the agents think and then the doers do. Will the agents do more overtime or will the bots actually be the worker, be doers?

Dana Forfa

>> I think it depends how your needs change. One of the benefits to looking at a enterprise solution like UiPath and the AI trust layer that we provide is that we are proven. We know it works. We use UiPath ourselves, not just because we work at UiPath, but because that is the best solution for the company in this way. We are just pushing forward what is next, and UiPath is the best solution for that.

Rebecca Knight

>> Hitesh, talk about you as customer zero for this and describe some of the finance processes. You just gave a very compelling one too that provided the biggest lift and what metric you're turning to, that is actually doing what the tagline of this conference is, which is agentic AI meets ROI.

Hitesh Ramani

>> Yes, absolutely. In fact, when I think about processes, in fact to also answer the question, an enterprise process has a part of process which is deterministic in nature. These are those mundane, repetitive tasks which does not require cognitive skills. Those type of automations we can transform using automation, and the same process may have a portion of the sub-process which requires cognitive skills, that's where agents come into play. But it's a probabilistic technology. In a probabilistic technology, you'll always have an outcome, but you want to make sure there is absolute governance and trust layer. You want to incorporate human in the loop, and that's where the orchestration comes into play. Now, just to give you some example. In the context of accounts receivable, every finance function goes through at the month end. They have to evaluate whether they need to create a general reserve or a specific reserve, if the customers are not paying you on time. The concept of general reserve in that equation is a deterministic problem that can be solved using automation, traditional automation. But in the same example, the concept of specific reserve, where assuming a customer has declared bankruptcy or the customer had promised to pay, but they did not honor their promise. You need to apply some judgment to say, I need to create a reserve. In that context, an agent will be applied. Both when the general reserve automation and the agent that is focused on specific reserve will jointly propose an accrual amount to the controller, but the controller has to then ultimately make a decision. That's why a human in the loop becomes very important with the orchestration in play.

Dave Vellante

>> Will there be greater network effects with agents than there were with RPA, with bots? In other words, as you scale the number of agents, will you gain more operating leverage? Do you have visibility on that? Relative to robots, which were fairly prescriptive in nature.

Hitesh Ramani

>> Absolutely. When an enterprise embarks on an automation journey, the starting point naturally is, you want to focus on deterministic tasks. That gives you a certain degree of operating leverage, but you cannot squeeze it or maximize your ROI a hundred percent because you still have cognitive tasks you were not able to automate in the past. Now, when you combine those two, you are actually building a strong foundation for an autonomous enterprise, so it maximizes your ROI significantly.

Dave Vellante

>> Because there was a big conversation a couple years after the rocket ship took off, about scaling RPA. It seems like you guys had experience, UiPath had experience and your GSI partners had experience, but it felt like it was more of a heavy lift. Now, with agentic, it's almost like that was the hidden answer to scaling RPA.

Dana Forfa

>> Absolutely. Absolutely. In the example that I gave you, we had a bot that would chase people at quarter close to complete the goods receipt process, rather than individuals following up all the time. The difference now is that the agent is actually answering some of the questions the procurement buyer would've had to answer when Hitesh comes back and says, "Can you give me some more detail about this?" In your example, yes, absolutely, my buyers are now freed up from answering those questions and can move on to be more strategic partners to the business, to add some more value, to negotiate some hard contracts themselves.

Dave Vellante

>> This is a massive productivity lift that we always talk about. New services that are delivered through software that give 10X the productivity and probably cost some jobs, but eventually they'll create more jobs and profit.

Rebecca Knight

>> Right, but also I can see, going back to the human side, is that it's also enabling the human to have the real interactions that matter. But I want to know about when the bot misbehaves or the agent takes a little too much agency, shall we say. Are there any rollback functions or kill switches that you have?

Hitesh Ramani

>> That is why enterprises have to be very thoughtful in terms of how you are putting agents in production, and I'll give you an example. We have an agent in production, it's called financial insights agent. What that agent does is on the back end it writes a SQL query and probes a question into the data warehouse. For example, tell me how does the payroll expense for the last four accounting periods compared to the preceding four periods? It'll go and mine that information from the warehouse and get you the information, and it'll also provide you an explanation. When we initially put this in the UAP, it was hallucinating. It was giving us reasonings which were not substantive or authentic, and so we had to go back and rewire, re-prompt it, and give more context to it. That's where we were able to fix and improve the agent's score. Having said that, that is why when you try to put an agent in production, especially in the back office, in a company's internal controls or financial reporting framework, having a human in the loop along with that and orchestrating it together is always the right answer to start with.

Dave Vellante

>> I think a lot about experience curves. It applies in so many industries, manufacturing, certainly in semiconductors. Is there an analog in creating agents where the more experience you have, your marginal costs go down at volume? Are you seeing that type of dynamic in your business?

Hitesh Ramani

>> What we are seeing now is significant role transformations with the whole agentic. Because what has happened in past, we had this concept of digital FTEs, but they were only able to transform only certain aspect of what humans were performing in past. But now, if you pair the robots along with agents, you can actually transform an entire role. That is where I think enterprises will see, there are certain aspects. I think you mentioned that. Does that mean that we are potentially going to see that agents are going to replace humans? It's possible in certain areas like that where an entire role can be fully transformed. Having said that, you want to make sure that you have absolute confidence because agentic solution is a probabilistic solution. It can be easily applied in areas like customer service, potentially areas where you want to do coding. But when you think about enterprise processes on the back end, you have to think twice and say, "Okay, I want to potentially put a human in the loop along with an agent in the production."

Rebecca Knight

>> Dana, you started this conversation by talking about how you have an overabundance of good ideas that you want your team to implement yesterday. Talk a little bit about processes that you foresee being ripe for automation and maybe if there's a process that you automated that looking back you might not have automated.

Dana Forfa

>> So interesting. I think the number one thing I would say is that in procurement, whenever you do a big project, if you did it again tomorrow, you would do something differently. There is always the next thing. Looking back at SAP, I wish that we had the ability to use agents during that time and not just rely so heavily on RPA. I would do more, not less, in the future. Absolutely. What is ripe for automation? Most of it, especially in my world. Really what we want to get into next is really supporting, like I said, the organization from a service standpoint to make things smoother. I don't know if you've heard this rumor before, but usually, procurement is not exactly loved throughout the organization. When someone comes over to me at Fusion and says, "Are you Dana Forfa?" I'm not sure if that's going to be a positive or a negative interaction coming next. My goal is that procurement feels like a wonderful partner because my team is spending their time being partners, and these other regular repetitive tasks are taken care of by our agents and bots.

Hitesh Ramani

>> It's an interesting comment. Just to add what Dana mentioned here. She mentioned during the SAP phase we would have potentially been blessed if the agentic solution was already around. I think this is the trajectory we are experiencing and most of our customers are also talking to us about, is all SaaS applications where traditionally CIOs made a decision to go under the hood and do custom configuration, those are now they want to minimize and eliminate that technical debt. The viewpoint is, instead of going and doing hard-coding customization, can you solve that problem using an agentic solution?

Dana Forfa

>> Absolutely.

Hitesh Ramani

>> This option did not exist. If you were to go back and reflect saying, what is it that you would want to unwind or redo it in a different format? I think that's a perfect use case for that.

Dave Vellante

>> I deal with a lot of procurement and I would say that-

Rebecca Knight

>> You love procurement.

Dave Vellante

>> Sometimes yes, sometimes no. I would say that I would characterize UiPath's procurement as a strategic enabler, keeping the business moving. A lot of times procurement's a blocker. I've never had that experience with UiPath. It was always a very pleasant conversation. What matters to you? What matters to us? Okay, let's get this done and keep moving so we can drive the business. I think that's my interpretation as an observer as to what the philosophy has been versus just, let's grind them down, let's slow it down, put it in the back. Is that a fair characterization?

Dana Forfa

>> Absolutely. I can also say that we cut all of our SLA's in half or better. Hitesh knows because we actually automate the report that goes to him. Through the use of AI, all of our SLA's are cut in half or better, like I said. It's not just about meeting SLA's, it's about beating them for our team. We hope to actually bring down those SLA's because we're so consistently beating them, which this is the dream in procurement, to have someone say exactly what you said. I think that experience is only going to get smoother from here on out.

Rebecca Knight

>> What does the future hold for procurement and for finance here, when we're back together at Fusion 2026? What will you want to have say that your team has accomplished in terms of making this agentic vision a reality? Do you want to start out, Dana?

Dana Forfa

>> I'd love to be able to share demos real world. Have the team here, allow them to talk about their favorite agents. What's been the most impactful?

Rebecca Knight

>> Name their agents.

Dana Forfa

>> Yeah, exactly. We have , who is one of our favorite corporate card bots. What are the agent's names? I don't know. Maybe we'll crowdsource at the next Fusion. All right.

Hitesh Ramani

>> Yeah. For me, I guess next year, we in fact met as a team, leadership team a week ago, and we laid the foundation for autonomous finance function. As Dana mentioned, I would like to see as many proof points with a demonstrated ROI, that we can actually showcase to our customers, and hopefully they can go and rinse and repeat without thinking much about it. From my standpoint, that would be a great success.

Dave Vellante

>> Well, Dana, your Yankees are looking pretty good here. It's tight.

Dana Forfa

>> Hey, it's a good day.

Dave Vellante

>> It's a tight game. You're hoping that the Yankees win the World Series. Of course, I'm hoping the Red Sox, so we'll see.

Dana Forfa

>> Agree to disagree.

Dave Vellante

>> We'll see you here next year. Right? All right, that's what makes a market.

Rebecca Knight

>> Indeed. Yeah. You'll wear Red Sox shirts next year on the state.

Dana Forfa

>> I'm not sure about that.

Dave Vellante

>> I don't think that'll ever happen. Either way, we won't.

Rebecca Knight

>> All right. Well, Dana and Hitesh, a very fun conversation. Thank you both so much for coming.

Dave Vellante

>> Thank guys.

Hitesh Ramani

>> Again, thank you so much. Thanks for having us.

Dana Forfa

>> A pleasure, thank you so much.

Dave Vellante

>> Appreciate it.

Hitesh Ramani

>> Thank you.

Rebecca Knight

>> I'm Rebecca Knight for Dave Vellante. Stay tuned for more of theCUBE's live coverage of UiPath Fusion. You're watching theCUBE, the leader in enterprise tech news and analysis.