The Human Element in AI-Driven Procurement

With artificial intelligence reshaping nearly every function, procurement leaders face critical questions about the technology’s limitations, applications, and understanding the human-machine balance. At ProcureCon 2025, we interview Mastercard’s Pratik Patel and get candid insights on why AI’s rapid advancement doesn’t eliminate the need for human judgment, but rather amplifies its importance.

After seven years at Mastercard, managing five critical categories from Tech Services to Contingent to Staffing to Data, Pratik brings a truly global perspective shaped by negotiation experience across 18 countries. His unique experience brings valuable insight into how procurement must evolve with AI.

AI Is Only as Good as Its Training Data

The conversation around AI-driven hiring often glosses over a fundamental limitation: algorithms can only process what they’ve been trained on. This creates boundaries that many organizations often overlook when implementing automation.

“We’ve got to be a little careful because we’re only as good as the test data that the AI has…

AI has not been to 18 countries, for example, and negotiated in those countries, understanding the dynamics. They can understand whatever is in the test data.”

This limitation becomes particularly relevant when considering complex procurement decisions where cultural nuances, tacit knowledge, and relationships influence outcomes. While AI continues advancing rapidly, Patel suggests we’re still far from achieving the 99% predictability needed to fully automate high-stakes decisions related to people and suppliers.

Finding the Right AI Applications in Procurement

Rather than viewing AI as a replacement for procurement professionals, Patel sees opportunity in targeted applications that leverage AI’s computational strengths.

“AI can definitely help with executing RFPs. If you have a library of RFPs that you can go back and look at… now go and ask AI to build me an RFP. The AI can do that, but it’s going to be constrained by the test data again. But every single day that data gets better and better.”

This approach aligns with broader industry trends. According to a recent McKinsey study, procurement organizations implementing AI for specific use cases like contract analysis and RFP generation have seen efficiency improvements of 20-30% in those processes.

Where Patel draws the line: using AI to make final supplier selections. 

“If you’re going to use AI to actually select the supplier, now you’re getting into risk elements. Who knows the business better than the people that interact with them on a daily basis? And the AI doesn’t do that.”

The New Value Proposition for Procurement Leaders

As AI assumes more routine procurement tasks, the role of procurement leaders evolves toward strategic oversight and governance, becoming more central to enterprise decision-making, not less.

“The value prop with AI is going to be better cost-effectiveness, quality, and fast. It’s going to be able to do complex computations, different scenarios, much faster than a human can do it.”

This shift demands that procurement professionals develop new competencies focused on AI governance, strategic thinking, and relationship management. Rather than diminishing procurement’s importance, automation may elevate it by removing administrative burdens while emphasizing higher-order decision-making.

Managing the Three Core Risks of AI Implementation

Any organization integrating AI into procurement and hiring processes must navigate several critical risk areas. Patel identifies three pivotal concerns that will determine success or failure:

1. “One is the risk around regulations and what constraints are going to be there, Do you really want to go through that process of setting up this technology and then find out six months or a year down the road that you can’t use it?”
2. The second risk centers on user experience and adoption. “Adoption of the technology is critical. And to adopt it, you have to have good experiences with the people who are using it. I don’t think it’s AI to AI that we’re talking about. It’s still a human interacting with the AI.”
3. Finally, Patel highlights the critical need to address algorithmic bias: “The third risk that you have to think about is whether you have addressed biases. Do you have enough of a confidence level that all biases are being evaluated as a part of that use?”

Recent research from Deloitte indicates that organizations that proactively address these three risk areas in their AI implementations are 2.5 times more likely to achieve their intended ROI targets than those that neglect them.

Breaking Free from “Fake Deliverables” by 2030

Looking ahead to 2030, Patel offers a prediction that extends beyond AI to address a persistent challenge in procurement practice: the misalignment between payment structures and actual value creation.

“Hopefully by 2030, all of our programs across the world will have really addressed the risk associated with ambiguity, the risk associated with paying for capacity structured as a quote ‘deliverable-based engagement, Because that creates tremendous waste for us as well.”

This vision represents a fundamental shift in how procurement approaches supplier relationships, moving away from artificial constructs that create administrative overhead and toward authentic value-based partnerships.

Leading with Experience, Embracing Innovation

After 23 years in procurement, Patel views his role as balancing innovation with hard-earned wisdom. The acceleration of AI doesn’t diminish this responsibility; it amplifies it.

“If there’s a perspective that I can share that inspires people, then I’ve done my job,” 

This perspective offers a blueprint for procurement professionals navigating the AI revolution: embrace the technology’s strengths, acknowledge its limitations, and recognize that the most valuable procurement asset remains the human judgment built from experience and relationship.

As we approach 2030, the most successful procurement organizations won’t be those that eliminated human decision-making, but those that seamlessly integrated AI capabilities with human insight to create unprecedented strategic value.

Applying Lean Thinking to Real-World Procurement

If you’re grappling with procurement complexity, Leanergize helps cut through the noise with pragmatic, experience-driven guidance. Founded by Pratik Patel, the Leanergize focuses on reducing waste, improving supplier outcomes, and applying lean principles to deliver faster, smarter, and more cost-effective procurement results. Reach Pratik Patel at his firm, Leanergize: info@leanergize.com

Interview Transcript

Joseph Cole (00:00.91)
Hey, Pratik, welcome. We’re here at ProcureCon 2025. Thanks for joining us. Why don’t you introduce yourself? Tell us about who you are, what you do, who you work for, anything else that you want to share.

Pratik (01:44.935)
Yeah, absolutely, Joseph. Thanks so much for having me. I’m all about inspiring people, so I hope our conversation will inspire people. I’m Pratik Patel from MasterCard. I’ve been there for seven years. I have five categories that I manage. I’m responsible for tech services, our contingent workforce, contact center, data AI, optimize the suppliers, optimize the spend, and be a consulting arm to the business for the right supplier for the right need. And then I also have seven divisions in technology in North America.

anything over half a million dollars. So I’m kind of uniquely in a place where I understand our technology objectives, our strategy, and able to really drive influence with the business in terms of ensuring that we’re able to truly gain what they want and minimize the waste in the process as well.

Joseph Cole (02:35.842)
Lean principles.

Pratik (02:44.815)
Absolutely.

Joseph Cole (02:35.842)
Well, Pratik, is there something that you want to share that someone might not know about you?

Pratik (02:44.815)
Yeah, so maybe somebody probably doesn’t know that I’ve been to 18 countries that I’ve negotiated in as well. So this is outside of MasterCard, mostly when I was at Energizer. And I had many areas of indirect, direct areas, and probably at least three times a year I’m doing around the world trips with a meeting with various different suppliers. So it’s given me a unique perspective around different cultures and really understanding how they go about to do business in those different markets. So I’m fortunate that I’ve had that experience and I try to leverage that as we grow globally in growing our program.

Joseph Cole (03:25.87)
Sweet. AI is the big conversation of the century, I think, at this point, or the new century. It’s ubiquitous, right? So the question here, the topic that we’re going to talk about today is, will humans still be in charge of contingent hiring in 2030, which is only five years away? So let’s start there. The first question for you is, AI-driven hiring is already making decisions. How far will this go in the next five years?

Pratik (03:52.313)
Yeah, so think we got to be a little careful because we’re only as good as the test data that the AI has. So when you talk about it From a Gen AI perspective, it’s natural language processing, it’s interpretation of the English language, and that interpretation may not necessarily represent what the intent is for someone that is trying to get an input or a task or a workflow completed by that AI. So the AI really needs to have a good holistic understanding, have that tacit knowledge as well of the people that are doing those jobs. We’re the sum of our life experiences. AI has not been to 18 countries, for example, and negotiated in those countries, understands the dynamics. They can understand whatever is in the test data. So we got to understand and appreciate that the test data is the constraint with any AI algorithm. And if you’re going to be able to quote replace someone, then that test data should be a representation of a 99 % predictability, I would say. That whatever that a human is going to be doing, that test data already exists for all the different scenarios that now you can actually enable for those tasks step by step tasks to be completed by the AI because the AI will be able to do the same things because that’s the only information that’s available for either the human or the AI. So in that situation, yeah, there’s a great level of predictability and maybe there’s opportunities. But that’s not a lot of use cases in my mind.

Joseph Cole (05:33.614)
Well at least not in the next five years, but I think that’s that’s positive… Hahaha… reassuring.
So next question, and you’re in procurement, but this is kind of broad. Will AI eventually replace recruiters? even procurement leaders, maybe staffing suppliers? Or what part do you think it might replace in the next five years?

Pratik (05:59.835)
Yes, I think the AI can definitely help with executing RFPs, for example, requests for proposal. Because requests for proposal, even people that are doing those, is going to leverage previous work that has been done. So if you have a library of RFPs that you can go back and look at and say, hey, I need to do a competitive assessment in understanding what our supplier capabilities are around being able to drive growth for us in these different markets, knowing that the constraints that I have is I need to make sure that we are, that they have a good understanding of the compliance, that they have a good understanding of how to go about to drive efficiency and waste elimination in the process that needs to be executed. Now go and build me an RFP that can do that. And I think the AI can do that, but it’s going to be constrained by the test data again. But every single day that data gets better and better. So I think something like that could be a huge enabler in the procurement field in terms of leveraging.
But if you’re going to use the AI to actually select the supplier. Now you’re getting into risk elements that I don’t know the degree to what the AI would truly understand the dynamics to be able to, have emotional intelligence as well, because you have to think about the risk element of the adoption of that supplier by the business too. And who knows the business better than the people that interact with them on a daily basis? And the AI doesn’t do that.

Joseph Cole (07:41.65)
Right, right. Yeah, it’s scary though how fast AI has evolved, right? And I think at this point, it’s almost like AI augmentation with the work. And maybe, and I guess the next question, the follow up question would be, so there’s a new value prop for procurement leaders based on what you said. What would you say that is?

Pratik (07:48.679)
The value prop, think, with AI is going to be better cost-effective. It’s the quality element because it has, it’s… garbage in, garbage out. So you got to make sure that the test data is good data. And if it’s good data, that’s good quality data that you can now rely on to be able to make decisions on. The speed, it’s going to be able to do complex computations, different scenarios, much faster than a human can do it. And then the cheaper element is really coming in in terms of holistically understanding the marketplace and the dynamics of what you need to be able to look at. For example, you can tell the AI and prompt it that, hey, I get the quality, get the speed, I understand that. But now, let me understand, what are all the different elements I need to be looking at from a cost perspective in making my decision. And the AI should be able to holistically help you understand that based on what it knows. And yes, we can do that ourselves, but nowhere near the speed and the quality that the AI can do it with.

Joseph Cole (08:59.102)
Right. In some ways, then it makes the role of procurement leaders like yourself more strategic, I think, and also a bigger priority for the enterprise.

Pratik (09:15.643)
Yep.
Absolutely.

Joseph Cole (09:21.518)
So what would you say is the biggest risk of automating contingent hiring with AI? And who do you think would be affected most? I know you answered some of that, but yeah.

Pratik (09:31.387)
So I think there’s three components to this, right? So one is the risk around regulations and what constraints are going to be there because law, you cannot break laws. So you got to make sure that you clearly understand what regulations are out there, what is also being proposed out there, because do you really want to go through that process of setting up this technology and then find out six months or a year down the road that you can’t use it? So that’s number one, right, is the regulations. That’s one element of risk. Number two is how fast you go to go and try to adopt this technology without understanding the customer experience.
Because adoption of the technology is critical. And in order to adopt it, you have to have good experiences by the people that are using it. So you’ve got to make sure that you’re at least evaluating that element and not just kind of pushing a technology, pushing a solution to the people that are interacting with it. Because I don’t think it’s AI to AI that we’re talking about. It’s still a human interacting with the AI. So that’s the second piece.

Joseph Cole (10:21.88)
Yeah.
Very much.

Pratik (10:44.85)
And I think the third risk that you have to think about is that have you addressed biases? Is there? Do you have enough of a confidence level that all biases are being evaluated as a part of that use? And there’s great technology out there today, which I love, which really kind of lets you at least have an awareness around what that technology is that you’re using, that you’re going to be using AI around, and whether it is holistic enough to think through the potential biases. And that’s wonderful. I think that’s a great evolution that we have really kind of taken over the last year really to be honest is like understanding that okay you’re using this new technology you’re going to use it to make recommendations use it to maybe execute tasks but at the same time you have kind of an audit that’s taking place in the back end that okay is this technology really providing us a response without any inherent biases.

Joseph Cole (11:51.758)
Yeah, that, makes 100 % sense. So last question for you is 2025 and five years away it’s 2030 and it seems like such a big, you know, milestone for us to reach. What predictions do you have for our space? Doesn’t have to be just AI, it could be anything really.

Pratik (12:01.937)
Yeah, so think one thing is fake deliverables. Hopefully, I’m hopeful that by 2030, all of our programs across the world, that we’ve really kind of addressed the risk associated with ambiguity, the risk associated with you’re paying for capacity, you have it structured as quote a deliverable based engagement, because that creates tremendous waste for us as well. Because when you’re putting it into this fixed price milestone deliverable based engagement, then what’s happening is you’re also kind of putting this constraint of what the execution parameters are and you have a change request that will take place too from that. And so hopefully the fake deliverables are going to be gone by 2030 because we’re learning about the waste that it creates and it’s not healthy for us. And you can make every engagement time and material. You can’t make every engagement a fixed price deliverable.

Joseph Cole (13:18.35)
Well, thanks so much Pratik, a world of knowledge. But anything that you’re excited about seeing here at ProcureCon or learning about.

Pratik (13:34.501)
Yeah, absolutely. It’s the interactions with people, it’s understanding what are they doing out there that maybe I can take away as well. And it’s also paying it forward. I mean, I’m 23 years in procurement now. And if there’s a perspective that I can share that inspires people, then I’ve done my job, right? I mean, at this point in my career, it’s leaving a legacy towards really driving impactful optimization from all of the experiences that we have.

Joseph Cole (14:06.798)
Amazing, well Pratik, thank you so much for joining us.