883: The CPO Agenda: Scaling AI Beyond the Pilot W/ Darshan Deshmukh

0:05Hi there, my name is Philip Leidson. I want to welcome you to the Art of Procurement podcast, the podcast that helps you, a forward-thinking procurement professional, position yourself and your team to proactively take advantage of the revolution that's taking place in procurement today. By interviewing industry trailblazers and sharing insights from our own experiences, my team and I pull back the curtain and shine a light on the strategies, tactics, and tools that procurement teams are using to elevate their impact. And today on the show, we're exploring what it takes for procurement to move beyond basic generative AI use cases and start applying AI to more complex, higher value decisions.

0:44My partner, Kelly Barna, hosted this AOP live session featuring Darshan Deskmoek, the president of Procurability, to talk about the maturity curve for AI in procurement and the role of data, analytics, and intelligent automation in scaling its impact. They start the conversation with Kelly asking Darshan whether AI has got to a place where it's changed procurement's perception of risk and supply chain uncertainty. Yeah, it's a great question. I think last time when we spent a lot of time talking about risk, it was all about real versus perceived. What are we doing about that? So within the last six to 12 months or so, I would say generative AI has matured our ability to connect those dots and identify those risks faster.

1:28And if we think about risk in general, the first step is, can we identify that risk as fast as possible, even before it happens? That's the key part of identification of that risk. In the current environment, information availability is not a problem. It actually is a problem that too much of it is available, and how much of that is real, how much of that is noise, and being able to decipher that as part of your information gathering phase is very critical during this risk management. And I think AI is helping us kind of drive that efficiency and effectiveness in that process. The key part of this process is really how do we use AI to identify noise from real signals?

2:14That's number one. That helps us identifying the real triggers and noise in the information. And then the next step in the process is we found those triggers at the right time or ahead of time. And what do you do with it? do we have a way to synthesize that information fast enough so that we can actually take action on it? I think a year and a half ago, we talked about human-centered AI. How does human play or how does human judgment play into this decision-making so that when we can pair that data with human judgment at the right place at the right time, instead of just reacting to barrage of information that is available on whatever sources that we can look at, right?

2:58So information is never a problem. It's all about how do you use it? Yeah. And then the benefit of this process is really context of the information and the action that is needed to be taken. A lot of people, when they talk about AI, they talk about, they talk as if we will eliminate the risk by using AI, by predicting the future or eliminate the uncertainty. Yeah. The prediction doesn't come in. I think that uncertainty doesn't go away. But the most important benefit of AI is the speed with which we are able to identify and then couple that with our decision making and the synthesis of the information, we can react to those risks as fast as we can.

3:38Traditionally, like when I think about this, right, we would gather a bunch of information, spend hours to synthesize it, analyze it, create reports, scenarios, and then report out those scenarios to the decision makers so that we can figure out what actions to take when something triggers right with ai's ability and even accelerate that portion of the scenario building it helps us with spending time on action versus analysis um and what are the risks here like are we focusing on the problem resolution or are we getting stuck with sea of information or kind of analysis paralysis could be a challenge right and especially when you are analytical and there is so much information available, it's very easy to kind of get stuck into that analysis paralysis.

4:26Oh, absolutely. And then the second piece I see it is like, what is the most effective so that we are focused on the raw data in an effective ways instead of amplifying the echo chamber of information, right? I think that also is in the decision-making, it's important to have all these variables kind of included in the decision-making versus like a internal vision about a specific part of the information. So when we look at data, when we look at the scenarios, when we look at decision making, are we focusing on the right triggers or are we focusing on unactionable triggers? All of that became a lot more efficient and effective if we do it in the right from a kind of people process technology angle to it and being able to drive that in managing the risk has become a lot better in the last six to 12 months or so.

5:16Yeah. No, and I appreciate your point. I mean, obviously we would love to be able to foretell the future, right? You figure that out. You let us know. Everybody will be thrilled. But in the meantime, this sort of compression of time that we're able to get. So a piece of information comes in. I love your point about noise versus signal, being able to separate that out very quickly. But I mean, in a way, it is like starting to step into the future. The sooner we can alter our strategies, the sooner we can go to a backup plan. It does change, if nothing else, the future course that we end up on because we were able to change our path.

5:53Now, before we totally leave this idea of perceived versus actual, one thing I'm curious to hear your point of view on is, you know, we are all doing our best to learn as much as we can about how AI works, where to apply it. There's generative, there's agentic. And certainly there's all of the data and the processes and the prompts and the things that feed into it. how mature do you think most procurement organizations are in terms of how they're thinking about all this versus where they probably actually are on that maturity curve? That's a great question. Again, I think we talked about the readiness of AI or what do you think AI's future looks like?

6:38We talked about this, you and I, maybe 18 months ago. and I see this as like a natural progression of the question is from a maturity standpoint, where are the organizations? So when I'm talking to the clients or at a conference, talking to my colleagues, there is always a question. So can you tell me how many are actually deploying the use cases that I'm interested in? And that kind of becomes like a pretty standard conversation. Yes. And when we talk about AI readiness or mindset for AI readiness. This is a very fast changing landscape. If we're going back to our like normal distribution conversation like 18 months ago, where we thought that one fourth of the organizations thought they are waiting, one half thought they are aware of AI but kind of figuring it out, and one fourth thought they were all in, right, just from a mindset perspective.

7:28If you ask that question today, I think that left-hand portion where like the waiting is kind of gone, I would say most of the organizations are aware of what AI can bring in from an efficiency effectiveness. And that's where most of the conversations start. And then at least half are testing something or piloting something to make sure that they are continuing to make the progress just from adoption, just from being able to drive the value, at least half of the organizations are driving that. And there is a very small portion that is taking very intentional approach to digital transformation.

8:06And that's where I think it's kind of really looking, using AI as an impetus to challenge how they are transforming the work that they are doing. I think that is kind of a very small portion of the organization from a maturity standpoint. if you kind of keep like this middle portion I would say a lot of people are doing pilots and if you kind of hear some organization if you go to conference I think there is a term that is kind of becoming very popular called death by thousand pilots it's almost like there are so many pilots going on at the same time because you started with a use case that you felt is important for your organization, or you fell in love with a technology platform that kind of offered you some sort of AI solution, right?

8:59And you start with that. So instead of starting with a problem statement, you start with like a use case or a favorite platform, and you pilot it out, right? And then you just keep on trying to prove the value in a vacuum. And that sounds very basic, right? It's almost like, if you think about like 20 years ago, So when ERP and e-sourcing era was kind of in the boom, we would talk about the same thing. Don't start with the technology. Start with the process and the problem statement. And we keep on doing that mistake again and again, where we lead with the technology instead of focusing on the outcome.

9:32And all of a sudden, we are kind of unhappy about the outcome, right? So I would say from a maturity standpoint, if I want to kind of club the organizations, small portion are kind of, most of the organizations are aware of what AI can drive for them from efficiency and effectiveness. There is half of them are kind of trying something out, just like different use cases, being able to do the pilots, trying to drive the value. And then there are maybe a small percentage, maybe around 10%, that are truly taking an intentional approach to digital transformation when it comes to driving the value. Yeah.

10:08And part of, I think, what has to plug into that division is, in what sense are we talking about AI? So it's one thing to use generative AI to help us pull data points or summarize notes from a meeting or potentially even look across contracts. It's one series of use cases. But when we start looking at what agentic AI can do and thinking about this idea of building an agent or a series of agents that can actually go off and do work and have responsibility or be checking things, that feels like an entirely different threshold in terms of what we have to be prepared to set up from a structural and a guideline standpoint.

10:48And then, of course, how we roll it out to the business and have checks and balances to make sure it's working. are you seeing that most people are still on the productivity focused end of things? Or are you seeing an increasing number of procurement organizations move on? And whether they're piloting it or they're one of the small percentage that are actually being purposeful and rolling it out, allowing these agents to start playing an active role in what the team does. Yeah. So I think maybe it would be beneficial for us to kind of simplify the generative AI and agentic AI. as part of the conversation.

11:25So a lot of times people think about, okay, what does a generative AI do versus what does a genetic do? And then I think people throw names and then kind of get all kind of confused about what is really real, right? So for somebody that is like, I like frameworks, so that keeps me kind of targeted in thinking. The easy way to think about generative AI is a productivity tool, right? So making sure that we build something that accelerates our work. So when we prompt generative AI, it gives you options, it gives you answer. This is more of a reactive approach to productivity. Absolutely an important step in accelerating our work.

12:06This needs us to prompt an AI, keep on getting the options and answers. And we call it, we have to train the model, get the data, keep on feeding it, what's accurate, and then keep on learning. right while with agentic we should talk about it as an autonomous digital worker right when it comes to what does it really mean what does that really mean is really we give an overarching goal to ai and ai uses independent reasoning uses various sources of data and when it fails it's self-correct right so that's like it's almost like you have a digital friend next to you kind of working with you driving uh driving the work now how do we move from productivity to strategic capabilities There are four conditions that AI needs to do to go to that agentic approach.

12:52First of all, and I'm going to make this into more of a framework thing, is tool calling. Is AI autonomously deciding how to use it, external tools, calculators, and to complete the task without human instruction? memory and adoption, like the system,

13:17the system kind of the system remembers the context around multiple such sessions and learning from the failures and then refine the execution. And then multi-step planning, it no longer relies on like a responses, like instead of that, it kind of orchestrates the loop of information. And then the way I think about it is it perceives, it reasons, it acts on it, it observes and repeats. That's kind of the approach to think about from a strategic capability perspective. And then finally, the most important part is does that AI has decision making ability under the guardrails that humans have given them?

14:00And then it drives the full operational workflow end to end so that humans only need to be in certain steps to drive an oversight role versus a hands-on execution role. So I think when I look at the conversations in the industry, I think we're still kind of in that initial efficiency effectiveness play. I think people are starting to test out the agentic portion of it. A lot of people would say that we are going all in on the agentic side, but I don't think there is that full spectrum of deployment when it comes to driving the value. I think it's kind of in that initial few months when it comes to driving that change.

14:43But this change is coming so fast that six months out when we have this conversation, I'm sure it will be a very different conversation than what we had six months ago. Yeah. And whether it's sort of making me as a human more efficient or whether it's actually taking work off my plate, off my team's plate, it does still seem like the phase that we're in is continuing to do the same things, potentially in different ways, more efficiently, maybe more autonomously. but we're not necessarily yet at a point where procurement is delivering something different or delivering something other than what we were delivering in the past.

15:18We're finding better ways to work in doing the things that we've traditionally managed. If a CPO comes to you and you're having this conversation, and what they truly want is transformation, I mean, I guess there's different ways to look at transformation. If you can really streamline and become very accurate and very fast, There's probably a transformational benefit to that, but has it actually transformed procurement? Where do you think we need to challenge ourselves to think if we wanted to be in that, I think you said 10%, in that very small percentage of companies that is truly transforming the outcomes procurement drives?

15:59How do we get ourselves to think beyond just doing the things that we've always done in a different or more automated way? Yeah, so I think this is a great question. This has been the conversations that I'm having with a lot of procurement leaders. This is a very common discussion. We do a lot of, like, Procurability as a services advisory firm, we do a lot of organizational assessments. And one of the dimensions generally is AI readiness by processes and also by organization. And when we do an assessment, I see a few themes in that. First of all, it's always focusing on efficiency. It's about cheaper, faster, better, and driving the cost efficiency in how we do things.

16:47That's kind of given as part of the action plan. That's always the case. But if you lead with that, then it's always difficult to kind of transition into more of a transformation conversation. that's kind of generally is one of the big pitfalls when you're trying to go from efficiency gains to a transformation conversation. It's the organization struggle with that, right? Second aspect that we see pretty commonly is, is this really AI, right? Like many organizations use AI as a term and a buzzword, but underneath it all, it's really just automation, right? Are you truly using AI to drive that efficiency effectiveness or are you just using automation and then masking it under AI, right?

17:30So that's like the second element. Like if you kind of stay within that automation zone, it's going to stay as the efficiency. It's not going to go, it is transformation to some extent, but it's not necessarily traditional transformation. I think another thing, and we just talked about it at the start of the conversation, are you starting with a problem statement or are you starting with technology or a use case that is a shiny object for you, right? And if you're not starting with a problem statement, more than likely, you're going to stay on the efficiency train versus the transformation train.

18:07So I think it's important that when the CPOs or the procurement leaders want to move from efficiency gains to a transformational value it's important to start with the mindset of problem statement what you're trying to solve and what transformation you want to drive yeah what's starting with individual use cases uh you want to be enabled or you you want yourself to enable challenging the current processes current organization construct uh so leading with technology could be one of the issues in those cases so it's it's important to start with the problem statement and then use technology as an And more than likely, these organizations are not using that old framework of people, process technology, triangulated approach.

18:56Right. It's always like you start with something that you have a hypothesis and you try to prove that hypothesis as correct. And then that becomes a problem statement and just becoming an efficiency game versus versus truly transformational model when it comes to driving that change. Yeah. And it is interesting too, because I know there's a great deal of concern around, are we just going to streamline all the jobs away? Are we going to automate all the jobs away? The huge opportunity is something that AI definitely can't do because we have to teach it to do everything that it does. It's almost like we need to sit back as procurement and not think about spend management or risk or contracts or suppliers or any of those things.

19:37What is a thing we've literally never done before. And how do we insert ourselves into that? It's almost like taking an entrepreneurial approach to, okay, there's this new technology that's available. How can we create a new product? How can we create a new outcome? It's probably going to be a little bit outside the traditional lines of where procurement usually operates. But I have a feeling that based on some of the conversations that we're having, those sorts of sparks and examples of creativity are just around the corner from being made public. But I think it's not until we hear someone say, oh yeah, we figured out that if we, I don't know, bring these different data sources together and put these other triggers in place, we can start to do part of what marketing was doing or operations or pull ourselves closer to sales in a way that it, that's when you're really getting transformational.

20:29And it's starting to actually change, not just how the company sees procurement, but how the company operates because of that interconnectivity that we're building out. Absolutely. And I think when it comes to procurement's value, right, if you think about the traditional cost, risk, resilience, those are kind of still right metrics for us to continue to drive, but those are table stakes, right? I mean, that's just the way we, like, we kind of used to think about us as this is be all and all right type of metrics. But now I think if you think about procurement and the chief supply chain officer's role, I think innovation plays a key role because none of the innovation is just driven by one single organization.

21:16It's mostly driven by partners. So having that ecosystem to be able to drive that innovation is very critical. Time to market is very critical depending on the industry, like the first more advantage being able to move fast, be able to bring the product and services to the market as fast as you can so that you can capture that market landscape. I think that also becomes very critical from a value generation perspective. And that kind of translates into revenue enablement aspect. So more and more CPUs and chief supply chain officers are focused more on can you enable the revenue as fast as you can can you increase the quality of the revenue can you increase the blended mix of the revenue so that you your eb doc continues to improve in the longer run right i'll use an example uh our parent company jabil our chief supply chain officer directly reports to ceo and the chief supply chain officer has not only the responsibility of all of your cost of goods sold as a contract manufacturer, but then also responsible for supply chain services, like which we are a part of, where we are delivering advisory, managed services, digital solutions to our customers.

22:34Now, what that does is really enables a different revenue stream within the organization that improves the profitability mix. That's like number one. Second piece is it creates a multi-threading opportunity with existing customer base. So you're not just only contract manufacturer anymore you are also a solution supply chain solutions provider that is adding value uh your customer portfolio and then that creates stickiness in the longer run right so it's not just you're just producing a product but you are also be part of the innovation for your customer base and driving that value in the longer run right so the way i kind of see this is really the conversations at the ceo level and the chief supply chain officer are all starting to become more of value generation for the organization that includes revenue enablement, which was kind of not necessarily a conversation till like last decade.

23:30I think before that, it was all about cost and risk for most people. Well, and I think it's interesting because you mentioned before metrics and KPIs. It's almost like if we try to think about the contribution of the AI specifically, One way to look at it is, okay, in the past, procurement would have generated X dollars in savings or had X dollars under management or going through oversight. Or maybe there are ways that you're managing the frequency of risk incidents or how long something is taking. One way to look at AI is to say, okay, in the past, procurement's performance across these metrics seemed to be capped at roughly this point.

24:09Gee, with the addition of AI, we can do twice as well or one and a half times as well versus what you're saying with revenue enablement. If you get yourself to the point where you actually have to add a KPI or a metric to the list to capture what's being delivered, now you know you've jumped beyond productivity and you've actually started to broaden the forms of value that procurement is creating. So it's interesting that looking at it through that perspective of KPIs and metrics, I think is a good way to sort of get a self-assessment of whether you're just doing better at the things that you've done or whether you're truly starting to branch out.

24:43Yeah. And I think it comes down to really having a conversation around your true value, right? I think traditionally, procurement is not a good marketing organization. We never did a good self-promotion, right? We always saw ourselves as somebody that improves EBITDA and somebody that made sure that the supply continuity stays so that we don't lose revenue. I remember my first job in manufacturing, I'm procurement also the production control and inventory management role. And one of the leaders who was kind of my mentor, this was my first leadership role within my first job. And the advice that he gave me is you have a very difficult balance of having parts on the line, but then also manage inventory.

25:44And those are conflicting, if you think about it. And when they are conflicting, you're not going to get fired for having too much inventory, but you will get fired if you run out of parts on the line. And the line stops. And then now the question becomes, okay, I'm always going to have as many parts as I can, like as your condition to like incentives, right? So you're basically focused more around optimizing like parts at the right place at the right time and improving the cost construct, but not necessarily focused on how do you optimize supply demand planning? How do you optimize or partner with sales to make sure that the demand is given to us in a most effective way so that we don't produce too much?

26:31Like those types of conversations. But now I think they are starting to become and the industry has matured and it became a lot more of a science than it used to be art. But now I think we're going to that next phase where how can we enable the revenue in a more creative way so that procurement can add value to not only bottom line, but also to the top line? Yeah. No, and I think your point about, like you said, nobody's going to get fired for having too much inventory. But at the end of the day, working capital is going to suffer. there are still downsides to taking that approach in the big picture that might not be felt or that an individual might not be incentivized by in the moment.

27:11If we sort of stay in this theme of looking at specific examples, I want to make sure we dig a little bit deeper into sort of building out the risk management capability. You know, certainly I can envision that, okay, maybe now you can have a daily process that runs that takes sales forecasts plus on-hand inventory plus, you know, demand and when you know you're shipping things out and be rerunning that analysis independently and a lot more frequently than it would have been feasible to do in the past. Where are some other places that you're seeing companies effectively apply AI to help them improve their risk management capabilities so so i think let's let's kind of decipher the risk management into various spaces and where ai can help right just from uh effectiveness and like the efficiency aspect of the uh of the play so the first one is uh data monitoring right so to manage the risk better we need to have a continuous monitoring and early warning system uh to have multiple sources of data and be on the lookout for the risk.

28:22So AI helps us to analyze thousands of sources of data to be able to come up with supply risk, market risk, geopolitical risk, financial risk, operational risk, so that we can get all the signals into our system in real time to identify any possibility of future risk before they become disruptions. So example, now sometimes you can do it, sometimes you cannot. The geopolitical is one of the most difficult ones right now. Depending on where we are, just from a pressure perspective, how can you manage that and create the what-if scenario to be able to manage that? The next piece that is very critical is once you have that data, what are you going to do with it?

29:05Can you prioritize the risk? Can you come up with scenarios and an action plan against it? That's number one. And then some scenarios cannot be mitigated. What is the impact and what's your mitigation of that impact over the period of time? That's the second piece. And then the third piece is how much of that is real signals versus noise, right? So AI can help us manage that kind of prioritization of risk so that we can focus on where the true exposure is and build the scenarios based on that. And then the last portion is once you have all that information, once you come up with possible actions.

29:47How do you execute those proactive actions and orchestrate those across the extended supply chain? So AI would help us kind of drive that automated workflows, give us the trigger alerts, create the supplier outreach, create customer outreach, manage all the response times, because everything is interconnected now as a globally integrated enterprises. It's important how fast you gather the information, synthesize the information, come up with an action plan and drive that proactively across the multiple parties. And that needs an orchestration across people process technology. So that AI is helping us driving that as part of all these various dimensions on the risk side.

30:31So it does kind of help us on the visibility side, it helps us on the synthesis side, and it also helps us more on actioning it once you integrate the human element into it. Yeah. And to your point about that interconnectivity. We actually have a great comment from Jirgana, I'm hopefully pronouncing your name right, that's come in. And I think she's talking about really connecting all these different systems and data sources, and then the processes that go with them. But I think the key phrase that she's included in here is that they can anticipate changes earlier so that they can make better decisions.

31:08It sort of goes back to that, what we were talking about before, where, no, we can't yet foretell the future, but we can create a better future for ourselves by, I mean, you don't have to buy a ton of time to set yourself up for much better decision making. And I definitely think that the tools and some of the intersection between systems and data sources are at the point where they can buy us that time to pause and think, get another point of view in there as opposed to literally firefighting, reacting to some of the things that are going on. Yeah, I think 20 years ago when I started the career, my first job was in the, like the I2 supply demand modeling, right?

31:49So we're looking at a contract manufacturer, supply demand planning, where like where a company would send, OEM would send a demand forecast to the short term and long term. And then you triangulate all of that information with all the game theory, right? Making sure that you have right parts at the right place so that you can optimize your cost. inventory costs and the carrying cost in general. With AI, I would say the art of that data analysis has reduced quite a bit because it's that information as long as the information is available and used in the right way. So that's number one. I think number two, what is important in this setup is having the common data models that the different organizations are talking to each other, right?

32:41So there is less noise and less coordination needed from sharing the data across extended enterprise. I think that's the second piece. So that the AI can anticipate some of the risks, some of the ups and downs within the process, and being able to recommend the actions that are needed to take, right? So humans are still going to take those actions, but being able to anticipate, recommending those actions and being able to give a choice to take what actions you need to take, AI is definitely going to kind of help us optimize all of that and definitely making it a lot cleaner than what it used to be versus 20 years ago it was an art.

33:25Yeah. Now, because you talked about your background in manufacturing, it makes me think for a minute about sort of the difference in ROI and rollout and risk between direct spend and indirect spend? I mean, certainly you gave the example of like, we do not want to stop the line. That is never the objective to stop the line. Are you seeing companies look and say, okay, the biggest potential ROI, maybe even a competitive advantage becomes accessible to us if we can bring these tools and improve processes and enable people around our direct spend? Or is there a mindset from a risk appetite perspective where companies are saying, you know what, maybe let's leave the line alone for right now.

34:11Let's test something on the indirect side of the house, even if it's something critical. I mean, to say that it's indirect doesn't suggest that it's not material to the operation. Are you seeing any divide or what are some of the things that might motivate a company to consider doing this on the direct side versus doing it on the indirect? Because I'm sure there's a slightly different business priorities and objectives that would drive the decision on either side. So it's a great question. So it's interesting and it has a push and pull there. So if you think about what makes AI successful in driving the results is really having good data hygiene and being able to drive that hygiene through the process, right?

34:57having good information and being able to use that information in the longer run. In most organizations, direct organizations tend to have a lot better data than indirect. And it just because it's, first of all, it has been streamlined for a lot longer than indirect side of the business. That's like just the physics of it. And the second piece is the variability tends to be a lot more on the indirect side because it includes a lot more services than traditional parts slash product, right? So if you think about that, direct generally is a good starting point to start testing out some of these AI-related efficiency and effectiveness play because the data is a lot better, right?

35:48Now, the benefit that you can drive from efficiency and effectiveness to kind of get that initial taste of success. Actually, indirect is better suited for it because it tends to have a lot more variability. And by using AI, you can reduce that variability and have better decision support system than what it could show up on direct side. Because direct side, okay, it's already a lot more streamlined than on the indirect side. So I don't see that big of a delta as compared to where I am versus where the efficiency and effectiveness is. Versus on the indirect side, you will see a bigger data. But what we're seeing is, I think it's more on the, depending on where the problem statement is for a lot of our customers.

36:36I would say it's 50-50, 50 % of the organization are using the use cases on the indirect side, which is not necessarily a core business. So they're testing it out on the non-core side of the business. But then there are 50 % that are saying I have very clean data so I can test out the use cases on the direct side a lot better and in a very kind of streamlined way so that I can kind of measure, monitor, and making sure that we are benefiting from it or not. So it's like 50-50. So some of them are using the data, lens data aspect of the direct side, and some of them are using the opportunities better on the indirect side, and it's less risky as compared to direct.

37:15Yeah. And of course, part of all of this is that we have to get people to buy in that the changes we're making to process and systems are worth the risk. There is experimentation, there's refinement that goes into all this. You had joked earlier about death by a million pilots. What has the trust building exercise looked like? And again, this may come back to a split in terms of what's the company exposure? or is it direct, is it indirect? But how has procurement worked with, you know, whatever their starting point on perception of their data and processes and systems is to try to convince people to test some of these additional capabilities that AI makes available?

37:59Is the general bias positive towards the capabilities of AI and helping that, or is it a little bit of an uphill battle as things become increasingly autonomous? this? So I think it's a great question. So any new technology or new change, it gets tested on the results. A lot of times you're going to have the skeptics around, is this actually driving the value that we expect it to drive, right? So you always need to, from an adoption perspective, to keep on showing the progress. And I think AI is kind of going through that phase right now where people are testing out the hypothesis, testing out the use cases, and showing the actual results against it.

38:53And so that's kind of number one. And then the trust is really focused more, the issue with trust is more around the core aspect of whether you have good data that is being used to test out AI. And a lot of times, it's almost like you put AI on bad data. It's just going to amplify it. It's not going to help you drive better results. And I think it's a simplified version of it. But I think where we're seeing that trust issue coming up on the adoption side is more on the data, less about the AI. So if you don't fix the basics of the data as part of your transformation, you're going to continue to increase the trust issues within the process.

39:43So that's number one. And the second piece is, are you keeping the humans at the center in this transformation? Are you using the human-centered AI approach to driving this transformation so that you can use the authentic AI to augment judgment? You can automate the routine tasks so that people are actually spending time on analyzing the recommendations and not necessarily just focus on analysis, right? Replace our decision-making abilities with better ways to actually look at various options and being able to drive that, right? All of those are going to be scaled through targeted use cases. and the organizations are going to continue to build that trust as they start seeing the value through these targeted use cases and then be intentional about that end-to-end transformation over the period of time.

40:35But it's all about building that momentum, creating the value, testing the hypothesis, and then continue to challenge us to take it to that next level when it comes to abilities that AI can offer us from a decision-making perspective. and maybe telling somebody, come on procurement. We are, as you said, horrible at self-promotion, getting to that point where we're not necessarily sticking our necks out and saying, hey guys, we've got it all figured out. Don't worry about it. We got this. But what have we learned? What did we change? You can almost approach it from an educational perspective and sharing that internally.

41:14It's still a way to communicate the fact that something is going on, progress is being made, new things are being tried. I do think it tends to be outside of our character. It's never really been a part of what procurement has done, but it does go back to that. If we actually want all of this to be transformational, then we have to start doing some different things. And then one more thing I would add to that is really be comfortable with transparency, right? So once we have the data, once we have the assumptions, be transparent about those assumptions, be transparent about the rationale, what you're trying to test out, recommendations, so that it's not secret.

41:52It's all about sharing what you're trying out, prove that accuracy with whether it's low risk or medium risk use cases. First, build that trust, communicate that out, communicate the results, and it's not the self-promotion aspect of it, but really building that kind of overarching trust about the function itself as you kind of integrate AI into the process. Yeah. And some of that transparency, I think, has to start by existing within procurement. You know, I was thinking about, okay, so the big picture strategy is going to sit with the CPO, right? And that's where it belongs from a procurement standpoint.

42:31But at the same time, the CPO is not going to be the person refining the prompts or going in and testing the logic and checking to make sure the systems are working. You know, as much as we have conversations about how AI is allowing procurement to adjust the operating model or forcing us to change the operating model to stay relevant, it's almost like you need a little bit of an inner operating model change where you've got something of an AI task force so that you've got all the different tiers of understanding, strategy, execution coming together? Because it's an active cycle, right? It's an iterative process and making sure that as people are learning more at the prompt level or the workflow level, that that's all still getting regularly filtered back up to the strategy level so that it all continues to evolve together.

43:21Maybe as a final question to get your thoughts on, are there some examples of where you've seen teams really set this up well? Is it regular checkpoints? Is it a line of connectivity straight from CPO down through a level or two to somebody that is working actively with the prompts? How do we set up within procurement to make sure that these iterative processes don't stall out because things get busy or other things take priorities, but to ensure that it's like top to bottom, the information, to your point about transparency, continues to flow? So there are, it's a cultural, I feel it's a cultural transformation, right?

44:01So it's two elements to it. One is the operational element that is, that needs to be enabled by taking risks, right? So people that create an environment where people are willing to try out new things, right? So that's more of the operational aspect of it. And you want to drive that culture within in the organization versus fear of compliance, right? What happens in the bigger organizations is you have these AI councils that are deciding what platform to use. I think that's where most of the focus is. This is not an approved tool to be able to drive our AI strategy, right? But what does that really drive from a strategy perspective?

44:43Are you starting with a tool or are you starting with actual use cases, right? and and i think we we get kind of stuck up in the whole oh from a strategy perspective we need to have a certain tool suite to be able to be successful what's it i think what we need to drive is that inherent risk taking slash kind of the testing culture so that people can we can test it out at a prompt level translate that into the success and the failures of those into overall strategy as part of your overall organizational construct. But then don't create a culture of fear. I think have a culture of testing with the organization to be able to drive that at a prompt level.

45:29But having said that, at the same time, you need to have a governance making sure that you're not just, it's not Wild West, it's really driving the right behavior within the organization that is consistent with your overall strategy. But I think coming to final topic, it's like most of the CPOs are coming to a conclusion that this is here to stick, right? So we have to be part of it as part of the transformation. We need to be able to figure out what do we need to drive organizational capability-wise, training-wise, being able to take that organization to that next level so they can build a strategy around it and create an execution engine that helps you drive that execution without fear, right?

46:09I think that's like the second from a culture perspective need to drive that as transformation as well. Yeah. And I think too, you know, very early on you mentioned the importance of starting with a problem statement. I actually think there's something empowering about that that maybe brings a little bit of forgiveness with it. Because if you've identified a good problem and done a good job of articulating it, it does create, hopefully, some space for experimentation and failure. Because if people are going to agree, yes, we still have this problem. It almost creates the obligation to try different things if you're going to solve it, because otherwise it's acceptance.

46:44And that's not something that anybody is going to celebrate versus if you're doing something a little bit more, oh, we want to do our procurement faster. We want to do our procurement with less people. Okay. Now you're starting with literally just a straight efficiency play and there's going to be a lot less forgiveness, a lot less tolerance of experimentation that doesn't work. But I think that piece of advice you gave is key around starting with that problem statement, rally people around it. It creates a target then that everybody is working towards solving. Yeah. Excellent. Oh, go ahead, Darshan.

47:19Did you want to, I don't want to cut you off. And then one more is I think being able to drive a culture where you don't have fear of failure. Yep. It's pretty standard for a lot of organizations to, and individuals that are competitive. we are afraid of your failure, right? And you don't want to penalize people for trying out, especially in this environment, considering how fast the transformation is happening within this space. So having that culture from tops down where you drive that, like you're not going to get penalized for trying out something new and if you fail, right? I think keep on driving that innovative culture and innovation culture within the organization.

48:01That's very critical. I completely agree. Darshan, fantastic conversation as always. Thank you so much for being here, for sharing your own past experiences, as well as what you're hearing in conversations with clients now. I also want to thank everybody who was here with us live, as well as you, if you're listening to this or watching later on demand. We always appreciate everybody being here. Thank you for the comments and the questions that came in in the chat. We appreciate everybody's time. Glad you spent some of your day with us and wishing everybody a productive rest of the afternoon.

48:40If this episode struck a chord with you, please do send it to somebody. We grow here at Art of Procurement through word of mouth and that would be really appreciated. You can also support us by giving us a thumbs up, a star rating or a review wherever you listen to your podcasts. Since 2015, we've built the world's largest free resource for procurement professionals looking to elevate their impact. Our resources span podcasts like this, videos, blog posts, papers, and events. To join us on the inside and to ensure you never miss an episode, a webinar, an event, or a post, please do subscribe to our weekly newsletter This Week in Procurement.

49:17You can do that at artofprocurement.com slash subscribe. That's artofprocurement.com slash subscribe. Thank you for joining us today and I look forward to seeing you again soon.