UiPath Fusion 2025 | Gerd Weishaar, UiPath & Saurabh Aggarwal, Deloitte

Gerd Weishaar, UiPath & Saurabh Aggarwal, Deloitte

In this UiPath Fusion 2025 interview, UiPath’s Gerd Weishaar, GM & SVP of Product Management for Test Cloud, and Deloitte’s Saurabh Aggarwal, manager for AI testing and engineering, join theCUBE’s Dave Vellante to unpack how agentic AI is transforming software testing from a bottleneck into a business accelerator. The discussion traces UiPath’s testing journey – early traction, a rebrand to Test Cloud and adoption across thousands of customers – while detailing why agentic testing matters as AI-driven development expands code volume and complexity. The guests share concrete outcomes across the testing lifecycle: faster design, optimized execution and lower maintenance via self-healing agents, with reported gains including ~15��20% productivity improvements in test design, ~30–40% better execution efficiency through smarter regression/prioritization and ~25% savings in ongoing test maintenance. Aggarwal explains Deloitte’s Ascend Test AI platform and a pre-built library of 1,500+ automations that jump-start ERP testing (SAP, Oracle, ServiceNow, Salesforce and more) on day one – reducing upfront costs and accelerating time to value. Weishaar also breaks down flexible pricing for UiPath Test Cloud – traditional per-user licensing alongside an execution-based consumption model that aligns spend to test runs. The conversation looks ahead to truly autonomous, agentic testing: agents that analyze requirements and code, recommend risk-aware scenarios, generate data, run tests and maintain them – helping customers evolve beyond manual workflows toward intelligent, goal-directed automation in the age of AI.

Share this session

Clips
More from UiPath Fusion 2025

Gerd Weishaar

GM & SVP, Product Management, UiPath Testing Solution

UiPath

Saurabh Aggarwal

Manager

Deloitte

play_circle_outline Enhancing Software Testing Efficiency: The Crucial Role of Domain Expertise in Overcoming Manual Testing Bottlenecks in ERP Implementations

play_circle_outline Transforming Testing Processes: Harnessing Automation and Self-Healing AI for Enhanced Savings and Productivity

Info
Transcript

Gerd Weishaar, UiPath & Saurabh Aggarwal, Deloitte

Gerd Weishaar

GM & SVP, Product Management, UiPath Testing Solution UiPath

Saurabh Aggarwal

Manager Deloitte

In this UiPath Fusion 2025 interview, UiPath’s Gerd Weishaar, GM & SVP of Product Management for Test Cloud, and Deloitte’s Saurabh Aggarwal, manager for AI testing and engineering, join theCUBE’s Dave Vellante to unpack how agentic AI is transforming software testing from a bottleneck into a business accelerator. The discussion traces UiPath’s testing journey – early traction, a rebrand to Test Cloud and adoption across thousands of customers – while detailing why agentic testing matters as AI-driven development expands code volume and complexity. The guests... Read more

explore Keep Exploring

What are the benefits of incorporating domain expertise into software testing, particularly in the context of ERP implementations? add

What were the observed productivity gains and improvements in execution efficiency during the implementation of the system, and what savings were achieved through the use of UiPath's agentic AI capabilities? add

bolt Powered by CUBE AI

Gerd Weishaar, UiPath & Saurabh Aggarwal, Deloitte

search

Dave Vellante

>> Hi, everybody. Welcome back to UiPath Fusion. My name is Dave Vellante. This is day two of theCUBE's live coverage, our seventh year covering UiPath's big customer event. We've watched the transition from single product company to platform company, to hyper-automation, and now of course agentic. And we're talking ROI and business cases and testing is an emerging use case that is really gaining a lot of traction. Excited to have Saurabh Agarwal here. He's the manager for AI testing and engineering at Deloitte. And Gerd Weishaar, who's general manager and senior vice president for product management for UiPath, the Test Cloud. Gents, welcome. Good to see you.

Gerd Weishaar

>> Welcome. Good to see you.

Dave Vellante

>> So, what is the nature of your collaboration? Are we talking here about... We just had Deloitte on. A lot of stuff going on internally. I'm sure you're doing a lot of testing. A lot of stuff going on externally. Is it both? Maybe you could talk about the relationship where it started and where you guys want it to go.

Saurabh Aggarwal

>> Absolutely. So, Deloitte and UiPath collaboration began with a shared vision to transform software testing for our clients. As we all know, software complexity has increased tremendously and business needs have evolved as well. So, that is where we saw an opportunity to combine Deloitte's deep industry knowledge, as well as domain knowledge, with UiPath's automation capabilities to address this challenge that our clients are facing.

Dave Vellante

>> Interesting. So, software testing is not just this horizontal generic thing, you can take advantage of your domain expertise within industry. Can you add some color to that and maybe give an example?

Saurabh Aggarwal

>> Yeah. So, let me tell you traditionally how it's been done. Traditionally, testing always have begun a bottleneck when it comes to large ERP implementations. Manual testing is always slow, as we know, it is prone to error and it provides you less testing coverage, because of that, it slows down your speed of acceleration of your transformation program. While automation provides you these significant benefits, but it requires upfront cost. That is where with collaboration with the UiPath, we came up with comprehensive list of test cases and test scenarios that we built, as well as with UiPath's agentic AI capabilities, we were able to have a pre-built library of 1,500 plus automation bots that we can deliver on day one of any ERP engagement, whether it is SAP, Oracle, ServiceNow, Salesforce, et cetera, across different industries. That gives them a kick start to start their engagement, as well as reduces their upfront cost as well.

Dave Vellante

>> And I remember when you guys first announced your testing platform, there was a lot of excitement and still is. It took some time to get going, but talk about the journey of testing within UiPath.

Gerd Weishaar

>> So, like you said in the beginning, we started from scratch, more or less. It is an adjacent market to the classic UiPath market, but it was very successful from the first day, actually. The first half year we had over 340 customers. And from there, it was a continuous story where we were growing and growing. Until last year, in November we decided to strongly focus on the application testing space and that's when we decided to actually brand or rebrand and launch Test Cloud. So, Test Cloud is the most modern agentic testing platform now in the market and that's our current offering where we are. So, it was a long journey, but it was very successful, with over 3,000 customers using our testing capabilities, significant portion on the revenue side for UiPath, which makes us very proud, we contribute to UiPath and a lot of happy customers.

Dave Vellante

>> Yeah, because testing it was this heavy lift to brute force. You're doing a lot of regression testing, like you said, a lot of manual work. You've got enough evidence now, can you quantify what's possible? What's the business benefit? What's it a function of? Is it lines of code? Is it complexity of the code? Where do I get the most bang for the buck?

Saurabh Aggarwal

>> So, you have three dimensions. One is when you design your test cases then you execute and then you have to maintain it. So, during execution, what we have seen, we have seen a tremendous productivity gain of 15% to 20%. When it comes to execution, that is where we saw tremendous increase of 30% to 40% increase in your execution by optimizing your regression, as well as progression functionalities. Now, lastly, once you're done with your implementation, you still have to get into the operate mode and you have to maintain your test cases. By leveraging UiPath's agentic AI capabilities, particularly self-healing agent, we were able to save 25% of your effort that you spent once your implementation is done.

Dave Vellante

>> Okay, so on the execution phase, that is time. Where does accuracy fit in?

Saurabh Aggarwal

>> Yeah. So, there are two aspects to the execution part one is speed of execution and second one is around how much you should execute. Everything doesn't make sense to execute every time. It all depends on what are your core changes, what business risk it involves, all that stuff. So, by leveraging agentic AI capabilities, what we have done, we have optimized test execution significantly. So, you execute what makes sense for that particular release or the enhancement.

Dave Vellante

>> Prioritize and just avoid things that you shouldn't be testing, saving everybody money. Are these combinatorial benefits or are they bespoke? Do they build on each other?

Saurabh Aggarwal

>> In the entire testing lifecycle that you see... So, at each level you save, as I've mentioned, so you gain at every stage of the testing lifecycle by leveraging these features.

Dave Vellante

>> So, the ROI is so blatantly obvious, which is why it's been such a successful business?

Saurabh Aggarwal

>> Yeah.

Gerd Weishaar

>> Yeah. I want to say something, my biggest goal is to eliminate manual testing. I've been in testing now for 25 years and I think it's the worst torture that you can do to a human being to do manual testing. It's a brainless, boring task and we could do it through computers. So, with some customers, we achieved the reduction of 50% in effort on the manual testing. Increase automation, increase efficiency, like you mentioned, somewhere between 30% to 40%, reduce test data effort, right? Test data effort can take up several days per month that the customers have to actually use to create test data and that can be reduced down to a minimum by, for example, using agents for implementation.

Dave Vellante

>> I remember when I got out of college, I had a math and a computer science degree, so I got a job... Unemployment at the time was like 10%, so it was really bad. So, I took a job as, they didn't call it a developer back then, they called it a programmer and it was a testing. It was miserable. I hated it. I said, "I'm out of here." And I went to work for IDC as an analyst, so I've been an analyst ever since. So, I felt that pain. And I'm sure there's a growth path there, but eliminate that, start the growth path higher. How do you see then the future of agentic testing?

Gerd Weishaar

>> So, the future of agentic testing is actually significantly different what the past was, that's my personal opinion. I believe that agentic testing will revolutionize, will make the test way more efficient, will be able to prioritize automate much more. I mean it will get to that extent where we don't even have to automate anymore. We can implement computer use and it can run natural language tests without even building automation. Autonomous testing capabilities that explore applications, we have agents that analyze requirements or source code, build requirements and test software. It will be very different. And by the way, I think it's necessary that this happens because software developed now is using AI much faster. Vibe coding, for example. The developers are much more productive now in cranking out more and more and more code. If we would do this manually, we could never keep up with that.

Dave Vellante

>> When you test vibe coding, what do the results say? Is it pretty good? Is it room for improvement? I mean there was always room for improvement.

Gerd Weishaar

>> Well, there's a ton of reports out there at the moment, but when you look at that, 60% of developers are using some AI capabilities, you see already that there's a decrease in stability in the software because we are creating more code, it's more complex, it's still the same amount of developers that need to handle an increased amount of complexity now. The testers are lagging behind. So, only 30%, 34% of testers are using gen AI capabilities, I think they need to catch up really quick because otherwise the divide between testing and software development, the speed will go significantly higher.

Dave Vellante

>> It's a great fundamental for your business.

Gerd Weishaar

>> Oh yeah, it is. Yeah.

Dave Vellante

>> Can you talk about, Saurabh, the Ascend? Because Deloitte has an Ascend platform, explain what that is and then test is like a module inside of that, right?

Saurabh Aggarwal

>> So, let me talk about the entire lifecycle of how Ascend is helping us in the entire journey. So, Deloitte has redefined the entire testing lifecycle by embedding AI at every stage of test. With our Ascend test AI platform, during the test design and development phase, we have been able to automatically generate the test cases and test scenarios by looking through the user stories and context grounding through our industry knowledge and domain knowledge that we have gathered from almost 700 plus implementations that we have done, whether it is around SAP, Oracle and all these package solutions. What we have done, the outcome is now you used to spend weeks in doing the test design and development. Now, that will barely be hours, where you are just reviewing your test cases that's been generated by Ascend test AI. The next step that we have taken is we have built a library of bots, as I've mentioned. We have close to 1,500 bots at this moment, which we can deliver to any of our engagement, which provides them the kick start to start their automation journey, reduces their significant automation cost and accelerate their development efforts to make sure that manual testing you are reducing from day one of your engagement when it comes to the designing aspect. The second aspect that where we have taken a big leap is around test data management, as Gerd mentioned. Most of the time you have your test cases, you know what to do, but test data becomes a bottleneck. So, by using agentic AI within our tool, what we have done, we have built intelligent agents that generates the data for your complex scenarios that you have, whether it is around region, business unit, organization, it reads through your scenario and it gives you based on what your scenario speaks. What it does, it really makes each and every test case that you have more relevant and more robust. Now, you don't have the case wherein most of your scenarios are you're not able to execute because you don't have data. Thirdly, next step is on the execution. We have optimized that layer entirely by using AI. What we have done, we have optimized it by using the prioritization of our test cases during the implementation phase, as well as for our regression test suite that gets prioritized based on the enhancement scope changes that has occurred. What it has made, now you don't need to execute what doesn't make sense for that particular release. You execute what makes sense, it gives you value, it gives you speed to market from that sense. Ascend platform has also taken a big leap wherein now when it comes to the maintenance aspect of the equation, we have leveraged UiPath for self-healing agents, wherein we are able to now reduce the overhead that we used to spend once you're done with your execution.

Dave Vellante

>> Thank you for that. Gerd, I want to understand how pricing works in testing. Is it a consumption model in the cloud? Is it license? All you can eat? How does pricing work?

Gerd Weishaar

>> Well, so with Test Cloud we defined two models more or less, right? One is what we call the market-aligned model, which is the typical per user licensing that we sell and we need this because a lot of the customers are still in this world and that will happen for another several years.

Dave Vellante

>> And they're comfortable with that model and that's they procure?

Gerd Weishaar

>> So, I have the firm belief that based on the gen AI improvements, that sooner or later it will not matter anymore how many people you will need. It will be more about which technology, which agents you have. And from there, we also define the execution-based model, which is a consumption-based model because the real value is in actually executing these test cases. And so, we offer actually both to our customers and I see an increasing interest, especially in the Americas market, for our execution-based model. So, you get the tools more or less with the execution and you pay for the executions that you need per year and I think that's the way to go in the future, to be honest.

Dave Vellante

>> So, you would commit to some number of executions potentially?

Gerd Weishaar

>> Correct. Yes.

Dave Vellante

>> And it's kind of a use it, lose it, or roll it over type of model?

Gerd Weishaar

>> Yeah, we thought about this like a phone model or something. No, but here's the good thing. We meet at the execution, so we know exactly what it is. I think we're trying to make sure that we are not overselling a customer because you want them to actually use the executions they need. Working with Deloitte, for example, with some customers, we have determined how many test cases they will execute. It is a very good model and when we work together with a strong partner, like Deloitte, to estimate what they need. And then, at the end of the year you look at the consumption. If they went over, well, we are not throttling them or we are not stopping them, but then we would have a conversation about the next year and increased consumption.

Dave Vellante

>> Right, because if I commit to a higher volume, I get a lower price per execution, is that right?

Gerd Weishaar

>> Absolutely, yes. Of course.

Dave Vellante

>> Yeah, okay. But it sounds like you're reasonable at the end of the year. You're not like... I won't mention the vendor.

Saurabh Aggarwal

>> Just to add to what Gerd mentioned, it actually enables your entire organization by just providing everybody the license to use. Now, with the license-based model, you can have only set number of resources who are literally using the tool. Now, you have empowered everybody to use it.

Dave Vellante

>> And I know some procurement purchases that way, but it's the old SaaS model, which is I think increasingly becoming outdated. A lot of customers are frustrated with that model. It's kind of broken in my opinion. I think agentic changes not only the technology model and the operational model, but also the business and pricing model. So, you guys have been thinking about that and the customers are going to demand it.

Saurabh Aggarwal

>> Yeah.

Gerd Weishaar

>> Yeah.

Dave Vellante

>> All right. So, where do you guys want to be a year from now in this partnership?

Saurabh Aggarwal

>> I can go first. So, looking ahead, I think we envision that testing will become truly fully agentic and autonomous. At this moment, we are talking about that agents will be able to develop, execute, and maintain the test. But what we see in the future, agents will not only be able to do all of these tasks, it'll also recommend you the test scenarios. It'll also anticipate the risk before it comes. That is a true autonomy that we really want to achieve through this collaboration with UiPath.

Dave Vellante

>> So, when you say risk, like security holes or governance issues, lack of compliance, bad code?

Saurabh Aggarwal

>> Yeah, so I mean in different industry we have different risk. In financial industry, you have the compliance risk. By looking through your requirement using these autopilot features, you can address those risks upfront, that solves one of the problems. Same goes with the healthcare industry. You have GXP requirements, right? You need to maintain your audit trails. If you can build that traceability from start to end, that solves the big issue. In traditional world, we used to do that manually. Any step you miss to take during your execution, you have to re-repeat it. I mean, just imagine if somebody will tell you to do it again, which you have done it... Now, agents can do it, right? They will never miss it. You have trained them in a way, they will self-improve themself in a way and give you all the power that you really need .

Dave Vellante

>> At 3:00 AM even. So, Gerd, your thoughts on where you want to be a year?

Gerd Weishaar

>> In a year or two? Well, we are already on the way, so I believe that the test tools will change dramatically. I think that it will be more like a collaboration. So, what you can see already today with our Test Manager, we have an immersed chat experience. You don't really need to know anymore which tab you have to click to create a requirement or whatever. You will be able to simply communicate with your testing buddy. You would say like, "Hey, I need some test cases. Automate them. Run them for me," whatever. So, it will also change who's going to do the testing, I think to some extent. The risk lies with the business owner. So, if we could actually expand and include all the business owners in this whole... And it will happen, I believe strongly.

Dave Vellante

>> Well, congratulations. A great business that you have. I was always excited to hear this addition. It's like Uber and Uber Eats. So, great having you guys on. Really appreciate the insights.

Gerd Weishaar

>> Thank you for having us.

Dave Vellante

>> Oh, you're welcome. Okay. And thank you for watching. This is Dave Vellante for theCUBE. We're here at Fusion UiPath's customer event in Las Vegas. We'll be right back, live, right after this short break.