Building Trust in Agentic AI with Rajkumar Irudayaraj

Hey everyone. Welcome back here to Techstrong TV. We've had quite a day here at Alteryx Inspire 2026 in Orlando.

We've spoken to customers and partners and executives. We've discussed so much around Alteryx and its pivot, its regeneration, if you will, over the last 18 months as a new senior management team came in here. And really more than anything, responding to what the market has been doing over this last year and a half, two years.

AI is reshaping not only this business, but all of our businesses. Continuing that theme, I want to introduce you to Raj Kumar, and I'm going to make sure we do our best here, Rajkumar Irudayaraj. You got it right, Alan.

I practiced. I pride myself on it. Rajkumar, welcome to Techstrong TV.

It's great to have you here. Thank you very much, Alan. I really appreciate you having me here, and I want to thank Techstrong TV as well for giving me this opportunity.

Oh, absolutely. We're happy to have you. For those who aren't familiar, Rajkumar is actually the SVP for Global Technology and Innovation Partners.

So I was a business development chief strategy officer person for 20 years, 25 years before starting this. So I've run the strategy, I've run the technology partner programs at several companies. It's a great role, but it's somewhat demanding because you always have to be rolling with what the market is dictating to you.

Give us a little bit of your background, Rajkumar. I would love to. It's good to start with a little bit very far from where I am now.

My background started with engineering and shifted to product management and then go-to-market. So I actually have a very unique background, those two hands that aren't 180 degrees, if you will, giving me that unique perspective of how I think about the data landscape and AI. And I've spent my entire career working on data, databases, building cloud data platforms, and now launching go-to-market products.

Early on, I was building databases, working at Informatica, and then transitioned to Oracle. I was there for about 16 years- Really? building Oracle database integrations.

Mm-hmm. Then moved to product management and worked across the Fusion Middleware and Fusion apps of Oracle. Then OCI, the Oracle Cloud- Sure ...

was still nascent. I helped build one of the earliest products of BI and launched it on the OCI, when it was very nascent. Then went to a startup, and the last eight years before I joined Alteryx, I spent at Salesforce.

Mm-hmm. Again, led two transformations. One was to launch their own organic database called Salesforce Database- Mm-hmm ...

and help build the Salesforce Data Cloud. It's called Data 360 now. Sure.

We're a customer. Yeah. Everyone is a customer of Salesforce.

Yeah. And one of the unique features we built while I was at Salesforce was the zero-copy architecture, which allowed that seamless data access to be able to drive and derive insights off of the data, and built it to all these modern cloud data sources, if you will, the database in Snowflake and the Google BigQueries and AWS, and so on and so forth. That partner experience then led me to Alteryx, where I am today.

And in my role at Alteryx, I run all of the global alliances and distribution and resource, but also how we think about taking Alteryx and building that whole business logic as a plus one and an and story with the existing cloud data platforms which all of our enterprises use today. So it's a long background, but very rich in data and AI. No doubt about it.

Yeah, grounded in how I think about that engineering first product and go to market, all coming together to help our customers drive the value. You're the right man for the moment. That's what they say, right?

I've got a couple questions here, Rajkumar, that I want to jump right in on with you. It's really three questions, right? Certainly, we've entered in 2026 the age of agentic AI.

Generative AI is 2025. And everybody, small companies like mine, big companies, Alteryx themselves, we're all going on this agentic AI journey. Some are accelerating, some are taking a wait and see approach.

A lot of organizations that I speak with, they'll have pilot programs going on. Yeah. But a lot of companies have not crossed the Rubicon to production.

What do you think is blocking them from really turning this loose? Yeah. It's actually a really good and a profound question, if you will.

When you think about the customer mindset-Sure, there's all this technology transformation going on, and AI has been on a rocket ship journey, and agentic AI transformation is afoot today. But when you think about that customer, the customer has built their business with trust as their foundation. And trust manifests in different ways across different parts of the business.

If you look at finance, for example, trust means can I defend this number? Can I actually tell what these calculations are? Can I tell what these exceptions are?

Am I able to verify these audits and audit exceptions, and can I approve any of these AR or APs that are coming through me, right? That is a very concrete and a very specific way of thinking about trust. " Mm-hmm.

Right? And so what I've seen in my experience is, yes, you have all the data that you need, but the data itself, for most of these complex enterprises, is coming from very different set of data sources. Some are structured, some are unstructured.

Some are coming from databases, some are still in spreadsheets. There's a lot of enterprises even today that- Still run spreadsheets. Yeah, no, absolutely.

Yeah. And so that is only part of the problem. And then if you actually can bring all of the data into a governed trusted dataset, you'll hear often claims from your AI vendors, if you will, that that governed data plus a little bit of context is good enough and your AI works.

Well, far from the truth. Because context, again, if you start unpacking it, can mean many different things. It could be metadata, it could just be the data lineage.

It could be just metadata about where the data is or just some of the calculations that they have. But ultimately, it's not enough. And when that becomes incomplete, your AI almost always doesn't have enough juice for a business owner to be able to give reliable ...

They just cannot defend the answers that they see out of AI. That's the biggest challenge that I see is how do you bridge that chasm from having good data, having some context, but making that leap to having an agentic interface that can answer? Yes, in some cases, depending on how critical your answer has to be, probably okay.

But if you're a finance guy, probably is not a good answer. No. And this is a reoccurring theme we've heard today, deterministic versus probabilistic.

And I think for us to, again, cross that Rubicon, cross the chasm, we need our agentics to be more deterministic- Yeah ... in their conclusions. I don't disagree.

That being said, I do think that's coming. It's possible. " No, it's going to happen.

It's probably going to happen sooner than you think. But we do need that to happen. I want to move on to our next question here, and that is, this is specific to Alteryx.

Why is Alteryx uniquely positioned to help enterprises in this move from AI pilots to governed AI execution to real production? That's a great question and really one of the reasons why I joined Alteryx. With all the background that I had, one of the opportunities I saw with Alteryx is how they're uniquely positioned to bridge that chasm from not just governed data and context, but bringing in that business logic- Mm-hmm ...

that is so critical for AI to be grounded in the answer that your business can trust. What do I mean by that? If I'm a supply chain, I want to be able to see what the exceptions are for the supply chain.

What is the logistics that I need? Are they grounded? If I'm a finance guy, and if I'm doing SOX compliance, I want to be able to see all the SOX compliance rules, and I can defend it and explain it.

So if I'm able to see it, it's visible to me. If I'm able to explain and understand what I have, and if I can repeat and reuse this workflow and the specific business logic that I have, then I actually feel good about this. Right?

Now you have all the ingredients that makes me feel comfortable as a finance guy because I can actually go back and not only see it, understand it, explain it, and reuse it, but I can also audit it- Yes ... if I needed to be. " Well, I don't know what that means.

No. I have no idea what that means. And if I cannot go back and explain exactly where that came from, and here's the data, and here's the $2 million exception and why it happened, you as a CFO have only two choices.

Either you fix the data or you approve the exception and move on. Well, but one of the biggest problems facing us with AI is a lack of trust. I mentioned it earlier.

And I think part of the issue was early on, it was almost magicalYou would tell it things, tell it to do things, it would do things, it would come up with answers. We would worry, is it hallucinating or what have you? But it was a black box.

You couldn't connect the dots. It was dotted lines. Now, I think that's really the Alteryx special sauce here.

And the agentic transformation that we're bringing. So what we're doing is not only making all of the business logic come together, so you can take your ground data, have your business logic, and have that agentic interface so you can easily ask the right questions- Yes ... and you can get the right answers to trust on.

But you can trace back. Right. And that's what builds trust.

When you can trace it back to the source- Yes ... now you can trust it. You got it.

I agree with you. And I'm glad we saved yours for last because that's been the question of the day. How do we build trust?

And that is, if you could trace it back to the source, you can trust it. I'll give you an example. We had a customer very recently.

It's one of the biggest global pizza chains. Mm-hmm. And they came to us and they said they have Google BigQuery, they have Gemini- Mm-hmm ...

and they have all of the data available. And they have this complex billing use case, which is franchise billing. Mm-hmm.

And so if you are any of these QR stores, you know franchise billing. Sure. Thousands of stores and they have 6,000 stores across the world, and you have essentially a problem where each of the stores, whether it's a franchise store or a corporate store, has different ways to calculate what the billing is, depending on how the sale happened.

Was it through Uber Eats? Was it through DoorDash? Was it a direct sale?

Oh, yeah. Was it an online order? Were the rebates applied?

Were there some other type of commissions that had to be paid? All of that has to be taken out before you get paid. You get paid, yeah.

And so that was a manual process with this customer. At 6,000 stores. Spending more than 400 plus hours every month trying to...

Well, it's actually every year. Let me correct myself. Every year.

Now, that is a lot of man days per month. Yeah. Right?

We went in with Alteryx, automated the entire flow, including the SOX compliance for franchise and all the calculations for corporate stores, used Gemini to be able to quickly ask a question. Which are the stores with variances, right? Or explain the SOX compliance rules.

Boom. You get the answer. That was almost impossible if you have everything in manual spreadsheets.

And that's a game-changing. That is the game-changing thing here where when the office of finance sees that real proof with the real data using their own environment and constraints, they actually believe what they see, and they get excited about it, and they want to go all in. That's how you go from experimenting, exploring, to something that's trustworthy and a real proof.

And that's where we are. All right, I've got my third question for you. You ready?

Okay. Okay. How will agentic AI change the business outcome landscape over the next few years?

And I realize this is a crystal ball, and the way things are, whatever we say now, two days from now or the next release from now might change. But as we sit here today, how do you think this changes? It's a really good question and I'll share my perspective.

Agentic AI is here today. We already have many enterprises starting to use it in smaller ways, in places where it's easy to use and see the value of it, but not in all the complex business functions. Right?

And that is the transformation I think we will see over the next two to three years is how do you go from a place of doing little things, experimenting, feeling good about your technology to a place where you can take it to more complex functions. And think about the business outcomes as a way of a measure for the value the agentic AI brings. Not the activity that how many agents do you have is not the question, what did the agent actually do?

Right? Was it able to make a business decision for you? Was it able to automate some complex functions?

How many hours did you save? How much money did you save by doing this? Can it do it repeatedly?

Is it consistent through and through? Is that something that you will trust every single time? Now, when this becomes repeatable, when it becomes completely a trustable foundation of how you run the business, you have a new transformation.

That becomes the way you're going to run your business from now on. It's no longer an imperative. It is how you actually want to- It's just the way it is.

It becomes the new normal. It becomes the new normal. And where I think this is going and how we will see this transformation manifest is across these complex functions and industries, and the more investment is going into AI, the more of the value proposition is realized within the line of businesses.

You will see all of this take a foot. And a lot of complex problems are yet to be solved, but you won't solve it unless you know that you've taken that step. And we are heading there, and we're heading there rapidly.

And it's an exciting time for Alteryx to be in the space where we're helping our business users to realize that value and accelerate their journey to be able to realize that outcome sooner than later. I think it's an exciting time for all of us. I've been in this, as I said, you've been in this business a long time.

I've been in this business a long time. I don't ever remember it being this exciting. Ah.

Totally. Agreed. Thank you so much.

I appreciate it. Alan, thank you so much. Really appreciate having me here.

I appreciate you being here. Rajkumar, let me make sure I get that last name right one more time. Irudaya Raj.

Rajkumar, thank you so much. I appreciate it. That's going to wrap up our Inspire 2026 coverage here in Orlando.

We hope you've enjoyed it. tv, but you can get it on iOS or Android or Roku, Apple TV, Amazon Fire, anything with a screen, you can watch it on there. Or you can watch it on the Alteryx sites as well, as they're going to have access to it.

And then I'm sure we'll have cuts out on social media as well. I hope you've enjoyed our coverage. This is Alan Schimmel for Techstrong TV.

We're out.