Full Transcript for Victoria Folbigg, CEO of Folbigg Consulting & AI Procurement Academy Co-Founder - Why AI Won't Replace the Best Procurement Professionals

00:04:35.920 You know, you said you enjoyed talking with the procurement folks the most at the job fair. 00:04:36.079 --> 00:04:38.000 What do you think it was that really clicked there? 00:04:38.079 --> 00:04:43.120 I mean, if you I mean, I know I'm I'm asking you to go back in ancient history, but 28 years. 00:04:43.680 --> 00:04:44.560 28 years. 00:04:44.879 --> 00:04:54.959 What was so interesting about Procter and Gamel, I still believe it's one of the best companies in terms of procurement organization, is that they had a real European roaster of people. 00:04:55.120 --> 00:05:06.240 They had men and women, they had young and older experienced people, but you you go in there and you could identify with a team, with a subset. 00:05:06.399 --> 00:05:12.399 And I really wanted to not go into, you know, 70% male, if I may say so. 00:05:12.480 --> 00:05:15.439 I I wanted to find my my tribe that I could deal with. 00:05:15.600 --> 00:05:17.680 There's a lot of young people as well. 00:05:17.920 --> 00:05:29.600 So the people that I'm still in touch with, and the lady who I actually met up with in DPW was uh a lady who was there in Barclays 28 years ago, but she was a couple of years above me. 00:05:29.680 --> 00:05:30.959 She was one of the managers. 00:05:31.199 --> 00:05:34.000 So we're still keeping in touch, and she's an Irish lady. 00:05:34.240 --> 00:05:47.439 So people who were, yeah, I mean, the people who were invited to, and I was also speaking to on a lot of graduate events, the people who they bring out to the graduate events are not necessarily directors of procurement. 00:05:47.519 --> 00:05:52.959 They're people who are a couple of years older than the grads, and you can you you can kind of click with them. 00:05:53.120 --> 00:06:02.720 And they were dynamic, young, sharp, you know, people that uh are from different various backgrounds, and you say, Yeah, I want to be part of that kind of energy. 00:06:03.040 --> 00:06:06.800 But they curated it consciously or subconsciously really, really well. 00:06:07.120 --> 00:06:07.920 That's amazing. 00:06:08.079 --> 00:06:13.759 Well, I'll tell you, I don't know if we talked about this, but my mom actually spent her career at Procter and Gamble as well. 00:06:15.199 --> 00:06:18.480 But I think she was a little bit before your time, to be totally honest. 00:06:18.639 --> 00:06:23.839 But she was uh she always talked, she was in the marketing side, so kind of the opposite side of the house. 00:06:24.319 --> 00:06:35.920 And uh, and her, you know, and I remember when I was growing up, her sort of claim to fame, if there was if there was one, was that she was working on Olay, which is the sunblock. 00:06:36.079 --> 00:06:41.040 And at before she was involved, it was just about SPF, which was the level of protection. 00:06:41.279 --> 00:06:50.079 And then she was the person who came up with the concept of like the four-hour versus all day, the sort of duration element of the sun block too. 00:06:50.160 --> 00:06:52.800 And that was like a big breakthrough moment for the for the brand. 00:06:52.879 --> 00:06:57.439 So it seems small, but these small things can make a huge difference when the messaging really lands, you know? 00:06:57.680 --> 00:07:25.759 It's huge, and marketing in PNG is where everybody, like, I didn't even know that marketing existed, but part marketing within PNG is where everybody aspired to be because the creativity of these people, so you come from an amazing bedroom because I remember how they would talk about like I had to in my last gig for Procter and Gamble buy seaweed for creams and so developmental chemicals. 00:07:25.839 --> 00:07:35.040 So seaweed, typically at that point, again like 25 years ago or something, was only something that they bought for food grade, but not for creams. 00:07:35.199 --> 00:07:45.680 So, what they've asked me then to do is scaling the production of something with actually a cheaper ingredient because it didn't have to be food grade. 00:07:45.759 --> 00:07:49.920 And at the scale that we bought it, we needed it to be more cost effective. 00:07:50.160 --> 00:08:02.000 So I had to like do all the research when Google and AI was non-existent to find, you know, production European and global production facilities, yeah, for for suit. 00:08:02.160 --> 00:08:05.920 And so all these things were dreamed up by marketing and RD. 00:08:06.079 --> 00:08:16.319 And it was amazing to be part of like the creativity and the creation of new products that PG brought out, especially in that my that beauty care division that she was also part of. 00:08:16.639 --> 00:08:22.800 How did you I'm I'm you know, I've I've basically been with the internet as as as long as I can remember. 00:08:22.959 --> 00:08:26.959 How do you even go about finding suppliers in a in a pre-internet world? 00:08:27.040 --> 00:08:29.199 I mean, is it all you know, word of mouth? 00:08:29.279 --> 00:08:30.000 Who do you know? 00:08:30.079 --> 00:08:31.839 I mean, how did how did you go about doing that? 00:08:32.000 --> 00:08:34.159 That's an it's it's amazing. 00:08:34.639 --> 00:08:36.480 A lot of it was word of mouth. 00:08:36.559 --> 00:08:44.960 Um, but also you had these massive yellow books and you'd go through and you know, expos, fairs, you know. 00:08:45.120 --> 00:08:45.519 Right. 00:08:45.759 --> 00:08:47.840 We had these massive business cards. 00:08:48.000 --> 00:08:54.879 So when I started, there was like a handover plus a stack of business cards that were given to me. 00:08:54.960 --> 00:08:57.279 You know, these are the people you might want to talk to. 00:08:57.440 --> 00:09:00.000 Yeah, and expos, a lot of expos. 00:09:00.080 --> 00:09:18.240 I tell you also, like just to geek out on uh PhD and marketing, because I was in the health and beauty care division and I was then buying the packaging, so the carpool packaging, we were allowed to go to Monaco to the expo that was about the the new packaging

development. 00:09:18.559 --> 00:09:22.960 You know, you'd come by helicopter from Mise to Monaco. 00:09:23.200 --> 00:09:23.759 Oh my gosh. 00:09:24.080 --> 00:09:36.240 And it was a most amazing, lavish, like palace come conference center where on every store, like we talk about DPW, and yeah, it has like the vibe. 00:09:36.480 --> 00:09:45.200 This vibe was glitter, you know, like champagne on every kind of area of a supplier. 00:09:45.279 --> 00:09:54.320 Because of course, what they were selling, they were selling premium packaging for the for the you know amazing creams and perfumes. 00:09:54.399 --> 00:10:01.200 And I remember as as an aside, as procurement, we were saying, well, what cost saving strategies could we bring to the table to procurement? 00:10:01.360 --> 00:10:07.440 One of them was, well, we take away the the paper packaging of cream, and people were like, ha ha ha ha ha. 00:10:07.679 --> 00:10:11.840 Never like they buy it for the packaging, they buy it to unpack. 00:10:12.000 --> 00:10:15.519 And then we said, Well, this is all virgin paper, you know, it's all white. 00:10:15.600 --> 00:10:17.600 So when you rip it, it's white inside. 00:10:18.000 --> 00:10:23.200 Can we not like put it white, but on the inside have like recyclables cheaper? 00:10:23.519 --> 00:10:24.399 You crazy? 00:10:24.559 --> 00:10:28.320 They want to feel the big white virgin paper. 00:10:28.639 --> 00:10:32.320 So it was all about experience even there. 00:10:32.799 --> 00:10:45.679 And maybe it wasn't as finesse as behavioral science, you know, but it was still like pretty much live about what experience can you give in these luxury high-end cosmetics and perfumes. 00:10:45.759 --> 00:10:47.120 Yeah, it's really amazing. 00:10:47.200 --> 00:10:51.919 I feel like I feel like these sometimes these these lessons come back again and again. 00:10:52.000 --> 00:10:57.679 And actually, you know what I have on my desk is a uh a case for an Apple uh AirPods. 00:10:58.000 --> 00:11:05.279 And so I think it's sort of funny because everyone says, like, okay, they did such an amazing job with their packaging, and that's part of the experience. 00:11:05.360 --> 00:11:09.519 But look, this is you know, this far predates the the uh the iPod. 00:11:09.600 --> 00:11:16.320 And and so, you know, people sometimes learn, you know, reinvent these, these, these age-old lessons again and again, and they they can work, you know. 00:11:16.639 --> 00:11:26.080 Or I think they borrow them because how do you unpack a uh you know, a luxurious perfume or an amazing champagne bottle? 00:11:26.240 --> 00:11:36.799 You know, you they when they did this whole thing in Apple, I think they looked at the most amazing unpackaging experience and then they went a little bit further of saying, How do I make sure it doesn't close? 00:11:36.879 --> 00:11:39.840 But it gently, you know, kind of closes. 00:11:40.159 --> 00:11:40.960 Put the lid on. 00:11:41.120 --> 00:11:41.440 Yeah. 00:11:41.600 --> 00:11:43.039 And it and it, oh. 00:11:43.360 --> 00:11:44.080 And it doesn't. 00:11:44.159 --> 00:11:47.759 And it doesn't, it gently comes down, but it doesn't flop, right? 00:11:48.000 --> 00:11:49.919 Yeah, it's got like an air pocket or something. 00:11:50.000 --> 00:11:50.240 Yeah. 00:11:50.559 --> 00:11:51.279 That that's it. 00:11:51.440 --> 00:11:52.000 That's it. 00:11:52.240 --> 00:11:53.360 Yeah, it's amazing. 00:11:53.519 --> 00:11:56.639 Well, let's let's zoom forward 20-something years. 00:11:56.879 --> 00:12:01.600 You know, I I uh I saw the LinkedIn update that you're now working as the co-founder. 00:12:01.679 --> 00:12:04.159 I love co-founders, I empathize with them all. 00:12:04.720 --> 00:12:05.519 I am one. 00:12:05.840 --> 00:12:07.200 The AI Procurement Academy. 00:12:07.360 --> 00:12:08.639 I'd love to learn a bit about that. 00:12:08.720 --> 00:12:12.960 And you know, in the description, it says we keep hearing the same thing from procurement professionals. 00:12:13.120 --> 00:12:16.080 AI sounds exciting, but I don't know how to use it safely in my work. 00:12:16.240 --> 00:12:19.120 I'd love to learn about what brought you to working on that. 00:12:19.279 --> 00:12:21.440 What is the the AI Procurement Academy? 00:12:21.600 --> 00:12:24.240 Where do you see the big blockers and opportunities are? 00:12:24.559 --> 00:12:25.440 Hey, thanks. 00:12:25.679 --> 00:12:40.720 Yeah, we're we're actually now in the process of testing our three courses with a cohort that we've selected, and uh we're growing this thick skin about like, oh, they like this because uh, they didn't like this because oh, well, they like this cause. 00:12:41.120 --> 00:12:41.519 Yeah. 00:12:41.759 --> 00:12:46.240 But the genesis of this, and actually we found each other. 00:12:46.480 --> 00:12:52.799 The guy who I've co-founded it with, uh, Richard Reynolds and me, we actually never met face to face. 00:12:53.039 --> 00:12:56.480 But we've been on a lot, yeah, that's another very interesting thing. 00:12:56.559 --> 00:13:02.960 So I meet a lot of co-founders who've grown up through uni together or been working together 12 years. 00:13:03.200 --> 00:13:15.600 This is the amazing life of the internet now, and really profiling people is so difficult now because uh even oldies like me can can behave in a very erratic way. 00:13:15.759 --> 00:13:25.279 But from starting the idea in June, July to now of having an MVP really is what, 20 weeks, whatever. 00:13:25.440 --> 00:13:25.679 Yeah. 00:13:25.919 --> 00:13:35.279 And and what we've done is we came together because both of us were geeking out on AI, both of us as uh former CPOs, but also people leaders. 00:13:35.440 --> 00:13:48.960 And I do a lot of uh curation of executive forums and roundtables for category managers, so pretty much very, very into still procurement as a subject and educating people who are doing procurement. 00:13:49.039 --> 00:13:59.519 Well, what I'm hearing is a lot

of noise from consulting companies about the art of the possible and that you can actually drive efficiencies 50, 60%. 00:13:59.919 --> 00:14:07.600 And then you see MIT, or actually 95%, don't get the ROI, and then so a lot of confusion in the middle. 00:14:07.919 --> 00:14:20.559 And then we started seeing, or I started seeing a lot of training sessions that were either software-led, which is nothing wrong with that, but it gives you a specific outcome and vision, right? 00:14:20.720 --> 00:14:35.919 And also a couple of them that were very high touch in terms of you want to get a cohort together that's live, and then they have to do certain sessions and they are on webinars together for four hours or something. 00:14:36.159 --> 00:14:39.279 And this is just not the reality of like most procurement. 00:14:39.440 --> 00:14:45.440 So, what our vision was is to create bite-sized procurement courses that are 100% online. 00:14:45.919 --> 00:14:48.639 Sure, if there's a community, we curate them. 00:14:48.720 --> 00:14:51.440 If they want to have a community, everybody wants to have a community now. 00:14:51.600 --> 00:14:54.480 But quite frankly, not every community is going really well. 00:14:54.639 --> 00:15:01.120 You know, I think people are a little bit also like tired of pseudo communities. 00:15:01.200 --> 00:15:01.919 Do you know what I mean? 00:15:02.159 --> 00:15:05.759 So, what we wanted to do, light touch, very digital. 00:15:05.919 --> 00:15:09.759 We created our own avatars because we also were drinking the champagne. 00:15:09.840 --> 00:15:17.519 We were also looking and investigating and trialing and working on AI marketing and AI educational products. 00:15:17.679 --> 00:15:24.240 And so we created a couple of courses that are based on our knowledge, but are really easily accessible. 00:15:24.320 --> 00:15:28.320 So the whole point is that you're on a commute home, you can listen to it. 00:15:28.480 --> 00:15:36.639 So you don't have to actually say, Oh, I have to stop and um stop what I'm doing, and I have to take a day off, or I have to do it over the weekend. 00:15:36.720 --> 00:15:37.840 You do it when you want to do it. 00:15:38.000 --> 00:15:39.360 It's very easy, it's digestible. 00:15:39.440 --> 00:15:49.360 It has a lot of checklists, ebooks, etc., with it if you want to supplement your learning, because your learning is multi, you know, visual sensual. 00:15:49.759 --> 00:15:51.360 How do you say that in the English? 00:15:51.679 --> 00:15:53.600 Like you want to reach, you want to listen. 00:15:53.679 --> 00:15:57.039 Yeah, you know, like sensory, multimodal, sensory, okay. 00:15:57.360 --> 00:15:58.879 Yeah, multi-sensory, you know. 00:15:58.960 --> 00:16:00.320 So we've catered for all of that. 00:16:00.399 --> 00:16:04.399 But the whole point was also what we found is a lot of people are talking about it. 00:16:04.559 --> 00:16:13.840 But if you start talking to people even in big organizations, and I spent 12 years in Singapore, big organizations, but the regional hubs are often forgotten. 00:16:14.080 --> 00:16:19.200 So maybe something amazing is happening at the headquarters, but the regional guys still don't know what's happening. 00:16:19.440 --> 00:16:23.919 So, and they want to get into the conversation and they want to upskill themselves. 00:16:24.080 --> 00:16:28.720 So, this is kind of the persona that we have designed the training for. 00:16:28.879 --> 00:16:30.320 So it's not for corporations. 00:16:30.480 --> 00:16:33.200 If somebody tells me, hey, can you design something for a corporate? 00:16:33.279 --> 00:16:34.080 I'd love to do that. 00:16:34.240 --> 00:16:36.320 But the primary reason is for individuals. 00:16:36.639 --> 00:16:56.720 And is the is the use case around you know helping them to use things like the foundation models of Chat GPT in their day-to-day work, or is it to help them understand more about the underlying technologies and capabilities, or maybe other vendors that are incorporating things into more like procurement-specific workflows? 00:16:56.879 --> 00:17:04.960 You know, I I guess I'm kind of curious, like, you know, if somebody's, you know, I maybe if I think about it, what's the dream state for for the folks coming out of the out of the courses? 00:17:05.359 --> 00:17:08.240 And we are writing them, and you know, nothing is standing still. 00:17:08.319 --> 00:17:11.599 The demands are different because the technology is moving so fast. 00:17:11.839 --> 00:17:14.640 And so we started off just demystifying it. 00:17:14.720 --> 00:17:17.920 Because what I like to do is I like to actually say, but how do I do it? 00:17:18.079 --> 00:17:20.480 Like answer the question, what does that mean for me? 00:17:20.559 --> 00:17:21.200 What shall I do? 00:17:21.279 --> 00:17:22.000 What shouldn't I do? 00:17:22.640 --> 00:17:25.279 So there is a baselining of risk burst. 00:17:25.440 --> 00:17:37.039 There's also, I geeked out on prompting quite a lot about okay, you know, about saying what to do, what not to do, but also things like, hey, ChatGPT is agreeable, just understand your own bias. 00:17:37.119 --> 00:17:41.279 If you put a certain question in with a small bias, it'll reflect it back to you. 00:17:41.359 --> 00:17:45.119 So you want to know these are amazing tools, but you need to know how they're built. 00:17:45.359 --> 00:17:51.039 And so then you you go, hey, Deep Seek is really good, but it's good for these things, probably better. 00:17:51.279 --> 00:17:55.920 ChatGPT is amazing, but please don't do any searches on ChatGPT, even 5.1. 00:17:56.000 --> 00:17:57.279 Just go to perplexity. 00:17:57.920 --> 00:18:03.839 And we're doing that through actually having these tools ourselves and using them every day and then distilling down. 00:18:04.000 --> 00:18:16.720 And doing, I think, these things that procurement wants to do but can't do because they have their day jobs, is just literally taking the same prompts across

the different gen AIs and going, okay, how does it work? 00:18:16.880 --> 00:18:18.799 This is one angle of it. 00:18:18.880 --> 00:18:19.039 Yeah. 00:18:19.359 --> 00:18:24.799 Is being non-technical, but saying, hey, this is this is that that the beauty of it. 00:18:24.960 --> 00:18:32.000 So you're basically working alongside a person who might be very, very smart, but actually who can also hallucinate. 00:18:32.079 --> 00:18:34.640 And this is how you can minimize your hallucinations, right? 00:18:34.880 --> 00:18:39.359 Because it's not tech, it it's it's an augmented kind of individual. 00:18:39.599 --> 00:18:43.680 I'm not going to talk about feelings, but in terms of how it learns, right? 00:18:43.839 --> 00:18:44.640 It learns. 00:18:44.880 --> 00:18:48.240 So then you talk about what to put in there, what not to put in there. 00:18:48.319 --> 00:18:57.599 We do a lot of confidential stuff, but it's amazing how often we still need to say, please don't put supplier names in there, please don't put, you know, people's names in there. 00:18:57.839 --> 00:18:59.200 And that that's one aspect of it. 00:18:59.359 --> 00:19:08.559 The other aspect of it though is procurement as category strategists need to know certain things and they want to do certain things faster. 00:19:08.720 --> 00:19:15.599 So, how can it apply, whether it's Microsoft Copile that's already within your infrastructure or something that you use on the site? 00:19:15.759 --> 00:19:19.759 How can it apply to augment your team, uh, your performance? 00:19:20.079 --> 00:19:21.279 What are agents? 00:19:21.440 --> 00:19:23.039 Like, is that even a thing? 00:19:23.200 --> 00:19:36.319 Should you build it from scratch or should you take it from uh uh and these are opinions, but I'd like to think that these are educated opinions of people who've been testing them, trialing them, talking to a lot of you know, experts. 00:19:36.400 --> 00:19:42.720 And that's what we'd like to do, but also looking at analysis, for example, that we major in as procurement. 00:19:42.960 --> 00:19:56.640 Like, do you, you know, there is there are a couple of off-the-shelf AI augmented analysis tools, and they might be good, but then also you could create something pretty neat in generic gen AI tools, right? 00:19:56.720 --> 00:19:59.039 Uh and they're really powerful. 00:19:59.200 --> 00:20:04.960 And in fact, what I'm seeing is a lot of the stuff you can create is is beyond belief. 00:20:05.200 --> 00:20:09.039 It's like, you know, the limit is only our creativity in terms of what we can. 00:20:09.599 --> 00:20:11.440 We need to understand the guardrails, right? 00:20:11.759 --> 00:20:13.200 Yeah, I completely agree with you. 00:20:13.279 --> 00:20:18.559 And I mean, it's it of course, like at Light Source, we're using it, you know, I think about AI in a few different ways. 00:20:18.720 --> 00:20:21.759 One is, of course, we have to use AI in our product, right? 00:20:21.920 --> 00:20:24.640 The second

is how do we use AI within our business itself? 00:20:24.799 --> 00:20:28.480 And you know, we recently hired a leader of our solutions engineering. 00:20:28.799 --> 00:20:37.920 So that's the group that will work with the sales folks to do really tailored demos and uh prepare like use cases specific to the client. 00:20:38.079 --> 00:20:46.720 And uh he's built out these great sets of uh of Gemini gems within uh Notebook LLM, the uh the Google products. 00:20:46.880 --> 00:20:56.079 And uh it's not that I think Gemini is in particular better or worse than you know, we use we use mostly anthropic and a few others in our product, but uh that interface is really nice. 00:20:56.240 --> 00:21:05.839 You can drag in a bunch of PDFs and then have a uh a structured prompt and say, you know, I want this, and then it outputs uh a briefing for the account executive. 00:21:05.920 --> 00:21:15.759 So when they come into the conversation, it's as if, you know, it's not sure, it's not better than if they spend weeks researching the company on their own, but it can get them, but that's not happening, frankly. 00:21:15.839 --> 00:21:19.759 So it's like better to get them up to speed so they can at least have a really thoughtful conversation. 00:21:20.000 --> 00:21:28.319 I agree with you, and I think that there's also, I feel like there's an entire spectrum of of AI, you know, skepticism versus evangelism. 00:21:28.400 --> 00:21:43.759 You know, some people you meet them and the future is TBD, but there's some people where they're like, you know, they really don't believe in this stuff, they don't think it's interesting, or they think it's a you know a hoax or it's gonna, you know, have its uh its winter, you know, soon and the bubble's gonna pop. 00:21:43.839 --> 00:21:48.000 And then there's some folks who say, like, we haven't even seen the tip of the spear yet, you know. 00:21:48.160 --> 00:21:53.359 And I I'm certainly in the camp of I just think these technologies are absolutely stunning. 00:21:53.440 --> 00:21:54.799 And I just can't, I can't. 00:21:54.880 --> 00:21:55.920 I'm so excited actually. 00:21:56.000 --> 00:21:59.759 I've never been so excited in my whole life about where uh where technology is happening. 00:22:00.400 --> 00:22:01.519 But are you seeing the same thing? 00:22:01.599 --> 00:22:04.319 And where do most procurement folks fall? 00:22:04.400 --> 00:22:09.759 You know, and and how do you think the job is going to change as these technologies kind of come into the fore? 00:22:10.000 --> 00:22:13.519 And one of the biggest questions also is does it augment or does it replace? 00:22:13.680 --> 00:22:16.000 You know, that's one of those central questions. 00:22:16.079 --> 00:22:16.799 So I'm I'm curious. 00:22:16.880 --> 00:22:22.160 Yeah, I mean, I don't nobody knows the answers, but I'm curious to hear you pontificate a little bit on this stuff, you know? 00:22:22.480 --> 00:22:24.000 And you asked me three questions. 00:22:24.240 --> 00:22:24.480 I know. 00:22:24.720 --> 00:22:29.039 And as I was thinking about one question, I forgot the other the other two. 00:22:29.279 --> 00:22:47.039 But but in terms, in terms of like my stance on AI, I am excited, but I'm also annoyed with a certain type of either procurement or tech professionals who put AI on everything. 00:22:47.599 --> 00:22:56.880 And I am enough of a tech nerd uh to say if robotic process uh automation is an answer, just for heaven's sake, do that. 00:22:57.440 --> 00:23:04.960 You know, like RPA is really good for moving from one system to another system. 00:23:05.119 --> 00:23:08.880 Don't say everything has to have AI sprinkled on top of it. 00:23:09.039 --> 00:23:18.240 It's maybe because I'm so into this topic and I am uh very excited about the technology and scared a little bit as well. 00:23:18.480 --> 00:23:38.079 So I just want people to know what it is and make their own decisions and also not say, like we're hearing uh or I heard at DPW, some people went around and said, Hey, my CEO said or my CPO said, you know, let's just where are my AI projects? 00:23:38.160 --> 00:23:40.000 And I'm like, what's your problem? 00:23:40.240 --> 00:23:42.000 Like, what problem are you trying to solve? 00:23:42.160 --> 00:23:55.519 Like, let's solve some cool problems here that we probably didn't manage to solve because of scale, because of all the data that wasn't uh read properly, or or the insights that we couldn't produce, et cetera, et cetera. 00:23:55.680 --> 00:24:01.680 So from my perspective, I love it, but it has certain use cases, and in certain use cases, don't use it. 00:24:01.839 --> 00:24:02.240 Yeah. 00:24:02.400 --> 00:24:05.519 And then you go down to the different uh AIs. 00:24:05.599 --> 00:24:12.880 You have the generic ones, and then you have the ones that the source-to-pee systems haven't built in, like you guys and everybody else. 00:24:13.119 --> 00:24:18.880 And then the question is, is it better for you to develop yourself or is it better for you to do this? 00:24:19.119 --> 00:24:38.559 And of course, I am on the other side of all of this as well, as a procurement person saying, I want to understand what the suppliers are doing with it, how much they're investing, but what data they're collecting and how they are uh setting themselves up because I want to understand what is happening in that black box too. 00:24:38.799 --> 00:24:39.200 Yeah. 00:24:39.440 --> 00:24:42.240 So so so there is so there is that. 00:24:42.400 --> 00:24:54.400 And the I I was geeking out because I did an executive forum of European procurement leaders on Wednesday, ran a face-to-face one in Amsterdam, been back to Amsterdam this week. 00:24:54.559 --> 00:24:59.920 And so a lot of time we spoke about actually the EU AI Act. 00:25:00.559 --> 00:25:02.240 Something interesting for you as well. 00:25:02.400 --> 00:25:02.880 Yeah. 00:25:03.200 --> 00:25:09.680 Because you see that it is going to change the way Europe

consumes AI. 00:25:10.400 --> 00:25:18.160 So procurement is not only the procurement systems, but it's also how do you negotiate with a company that has AI, number one. 00:25:18.240 --> 00:25:26.400 And number two, how do you ask the right questions of the companies that have AI within their products and services? 00:25:26.720 --> 00:25:38.960 And so a lot of the AI that is a little bit hyper-personalized is actually something that could be interpreted as high risk in Europe because of the regulation. 00:25:39.119 --> 00:25:48.079 So applicant tracking systems, profiling, for example, I think, is gonna come up as high risk within Europe. 00:25:48.240 --> 00:25:57.119 So all these applicant tracking system companies will have to justify what they're doing, how they're holding data, what are what are their algorithms and so forth. 00:25:57.279 --> 00:26:06.640 So there's a lot of stuff happening that procurement doesn't have time to necessarily think about and distill we want to bring to it. 00:26:06.880 --> 00:26:14.319 So I am I am amazed by the technology, but I'm also not blindly jumping in because everybody is jumping in. 00:26:14.400 --> 00:26:18.079 I'm trying to figure out where the best use case is, where this technology is. 00:26:18.400 --> 00:26:19.920 Yeah, it makes perfect sense. 00:26:20.000 --> 00:26:22.240 And and everything you're saying, I completely echo. 00:26:22.319 --> 00:26:33.440 I mean, I was on the other side of that table where people would come up and say, you know, we're looking, we have an AI initiative and we'd like to bring AI into the company, and people are asking me, what's your AI strategy? 00:26:33.680 --> 00:26:34.960 And I'm like, great. 00:26:35.200 --> 00:26:38.400 So and I I I say, What well, what do you want to achieve? 00:26:38.559 --> 00:26:41.519 And then it's like it's like crickets, you know, blank stares. 00:26:41.599 --> 00:26:42.559 It's like, we want AI. 00:26:42.640 --> 00:26:43.920 I'm like, to do what? 00:26:44.160 --> 00:26:45.599 And they don't really know. 00:26:45.920 --> 00:26:55.440 So it's it's one of these things where you know, I I guess that's you know, it's like anything, like you know, that's a it's a it's a risk, but it's also an opportunity, right? 00:26:55.680 --> 00:27:01.519 Because, you know, when I think about it, it it gives us then an opportunity to really define what the technology is. 00:27:01.680 --> 00:27:10.799 And you know, at some level, it's like you know, that old Henry Ford quote if you ask people what they wanted, they would say a faster horse, right, instead of a car. 00:27:10.960 --> 00:27:16.720 And so, you know, people are asking, you know, we know we have something powerful, what do we do with it? 00:27:16.880 --> 00:27:21.599 So I kind of feel like in some ways it's up to the, you know, we're product people, right? 00:27:21.680 --> 00:27:26.880 And so, you know, it's our responsibility, but we have to understand the pain points that people are looking to solve. 00:27:27.039 --> 00:27:38.000 Like that's the thing, that's where the the balance is between the customer and the technology provider is that the customer, what I you know, I tell my team to do this, and I'll I'll just tell you in secret, although it's recorded. 00:27:38.160 --> 00:27:44.640 I tell them listen very intently to their problems and pain points, but ignore their ideas on the technology front. 00:27:44.880 --> 00:27:53.599 Because a lot of times folks have like a they they know like like the back of their hand exactly what they're experiencing, their pain points, what they need to get done. 00:27:53.839 --> 00:27:59.119 But then the way that they think technology will solve the problem is is not always correct. 00:27:59.200 --> 00:28:08.480 And so that's on us on the engineering side to say, hey, look, you know, maybe we can actually think a little bit bigger about the art of the possible beyond what people are are asking for. 00:28:08.640 --> 00:28:13.680 And so, because what people ask for generally is what they've already seen or variants of what they've already seen. 00:28:13.839 --> 00:28:23.519 So, but I'm curious from your perspective, you know, what do you feel like are the biggest problems within procurement that your you know your instinct is telling you, hey, AI should be able to solve these things. 00:28:23.759 --> 00:28:31.279 You know, is it is it small pieces like helping you draft emails to suppliers or responses or evaluating bids, or is it much bigger things? 00:28:31.359 --> 00:28:35.279 You know, I'm I'm trying to think about, and and I don't have the answers to these things, to be clear. 00:28:35.359 --> 00:28:38.720 You know, I think about them every single day, but I don't I still don't have the answers. 00:28:39.119 --> 00:28:43.279 So I must agree with you on what you've said. 00:28:43.519 --> 00:28:52.160 People will want something comforting that they already see, and everybody wants to solutionize, and nobody actually understands the tech better than you guys. 00:28:52.400 --> 00:29:01.039 And so smart procurement people should be, and this is like the evolution of procurement, they should be telling you, this is the outcome I want. 00:29:01.920 --> 00:29:04.480 Show me three options how to get to the outcome. 00:29:05.039 --> 00:29:13.839 You know, and that is what a good procurement person should be doing, rather than saying the outcome has to be a three-step, you know, kind of process. 00:29:14.079 --> 00:29:14.400 Yeah. 00:29:14.720 --> 00:29:27.279 I don't know how many steps you guys have and how it looks like, but this is how it should make me feel at the end, and this is how or the stakeholders, and this is how it should, you know, it should solve these problems. 00:29:27.440 --> 00:29:41.200 So this outcome-based working with the supplier is something that is happening with more mature and sophisticated uh procurement organizations in in every sphere of actually categories. 00:29:41.279 -->

00:29:43.200 That's what they should be doing, yeah. 00:29:43.440 --> 00:29:45.599 But of course, there's also human nature. 00:29:45.839 --> 00:29:47.119 So I agree with you. 00:29:47.200 --> 00:29:48.480 Like you are the experts. 00:29:48.559 --> 00:29:50.720 So let the experts do the experting. 00:29:50.960 --> 00:29:53.519 You need to know what problems they're trying to solve. 00:29:53.839 --> 00:30:05.759 In terms of the problems, again, they need to figure out on a baseline level what AI can and cannot do in order to be able to give you the problems. 00:30:05.920 --> 00:30:07.200 It can draft emails. 00:30:07.279 --> 00:30:09.839 That's a like a real easy level. 00:30:10.079 --> 00:30:12.400 Yeah, ChatGPT can write emails better than I can. 00:30:12.480 --> 00:30:12.720 Yeah. 00:30:12.960 --> 00:30:18.079 But if that is what it takes to go step one, then that's what it is. 00:30:18.400 --> 00:30:25.359 But the capability of research and analysis on the data set that you have is huge. 00:30:25.599 --> 00:30:30.480 And so that augmentation, that other insight, that other voice. 00:30:30.720 --> 00:30:47.440 So for example, you feed it contracts, you feed it your own templates, you feed it what you want, you know, the contracts to be, and then you feed it the supplier paper and say, okay, give me an understanding of how it de-risks me or exposes me. 00:30:47.680 --> 00:30:49.119 It'll have an answer. 00:30:49.200 --> 00:30:53.200 It'll have an answer faster than you will have it, or any paralegal or illegal. 00:30:53.359 --> 00:30:54.880 Is the answer 100% correct? 00:30:55.119 --> 00:30:55.440 No. 00:30:55.680 --> 00:30:58.960 Is that another opinion that you can have a look at and spar with? 00:30:59.119 --> 00:31:00.160 Totally, yes. 00:31:00.480 --> 00:31:04.559 So in CLM, I see that there is contract lifecycle management. 00:31:04.640 --> 00:31:06.240 I see that there are huge opportunities. 00:31:06.400 --> 00:31:16.160 But I also see huge opportunities in it as a layer on top of robust data sets to query it in an easy, non-techy way. 00:31:16.400 --> 00:31:33.279 So show me my biggest supplier, show me my biggest, show me my biggest risk, show me where I have the most amount of late payments, you know, something that maybe you needed to ask some analytics IT group, and they'll probably say, Oh, I need a development. 00:31:33.440 --> 00:31:40.880 This is something if you have the right, you know, baseline data, AI could give you so much easier an answer for. 00:31:41.039 --> 00:31:43.759 And the other thing is hyper-personalization. 00:31:44.000 --> 00:31:48.240 So that's something that I don't see a lot of people doing yet. 00:31:48.480 --> 00:31:53.039 But for me, that is so exciting, and I think that's so exciting for marketing as well. 00:31:53.279 --> 00:32:10.000 Is I think that the future of the procurement teams who will crack it, Spencer, is they will be using AI the way marketing and behavioral scientists use AI, but on their

own stakeholders. 00:32:10.240 --> 00:32:24.160 And so I'm kind of working through something that's called the procurement influence engine that says we as procurement should gather data like crazy, like everybody should catalog, like the world depends on it. 00:32:24.319 --> 00:32:30.480 Because the more data we have, the more we know everything about a stakeholder. 00:32:30.640 --> 00:32:31.039 Yeah. 00:32:31.279 --> 00:32:49.680 Their compliance suppliers, their non-compliance suppliers, their spend patterns by year, by department, by individual, their projects, their budgets, their KPIs, whatever is on their internet or intranet, because every big department has an intranet that says what they do. 00:32:49.920 --> 00:32:53.920 You know, but is that actually what the spend data shows that they do? 00:32:54.160 --> 00:33:15.039 And then you have that spend cube, that data of that stakeholder marketing, and then you create procurement strategies based on actually understanding your stakeholder, like, you know, like doing what behavioral scientists do by studying consumers. 00:33:15.200 --> 00:33:21.839 Because there's one thing that marketing says or finance says or logistics says, and then there's the uh actual behavior. 00:33:22.000 --> 00:33:22.799 You see what I'm saying? 00:33:23.039 --> 00:33:25.680 Like what they actually say and what they do. 00:33:25.839 --> 00:33:28.240 And what do we can now decipher? 00:33:28.720 --> 00:33:29.759 It's really interesting. 00:33:29.920 --> 00:33:32.480 I mean, this is a very fascinating topic. 00:33:32.720 --> 00:33:35.119 It's fascinating for a million reasons. 00:33:35.200 --> 00:33:38.640 And I think that the I think the question of data is an interesting. 00:33:38.720 --> 00:33:43.759 I mean, I've you've inspired like 15 different thoughts, so I'm just trying to triage which ones to focus on. 00:33:44.000 --> 00:33:45.680 The first is around data collection. 00:33:45.839 --> 00:33:49.519 It used to be that people would say garbage in, garbage out. 00:33:49.599 --> 00:33:51.599 That was like a phrase that used to exist. 00:33:51.759 --> 00:33:54.720 In five years, nobody's gonna be saying garbage in, garbage out. 00:33:54.799 --> 00:33:58.640 They're gonna say anything in gold out, right? 00:33:58.880 --> 00:34:11.199 Because it used to be that if you had poor data quality, you know, like two, like if a vendor had two different names and it wasn't in a clean vendor master, that would mess you up. 00:34:11.360 --> 00:34:15.599 But now AI is robust to being able to deal with language. 00:34:15.920 --> 00:34:21.840 That's a it's a much more uh, you know, there's a much bigger margin of error that it can accept and understand. 00:34:22.079 --> 00:34:30.320 And so, and so in some ways it doesn't really matter as much if your data is clean, because you can actually start to do things with it, aggregate it, clean it. 00:34:30.800 --> 00:34:33.440 So I agree with you to a point. 00:34:33.840 --> 00:34:38.320 And so that's where we need to become hyper-focused on the data. 00:34:38.639 --> 00:34:56.079 So this example of Deloitte and Touche, D ⁇ T, or Deloitte or whatever, you know, all the different permeations of names that now AI is smart enough to roll up for the purposes of giving you a report that is probably gonna be to 99% correct. 00:34:56.239 --> 00:34:58.079 There's one use case, and I agree with you. 00:34:58.320 --> 00:34:59.760 I give me another use case. 00:34:59.840 --> 00:35:02.639 So I used AI-augmented ESG systems. 00:35:02.800 --> 00:35:13.599 And what they did in certain countries is they would give me loads of false positives because the name of the supplier would be featured somewhere, which is not a name of the supplier, but it just got it wrong. 00:35:13.920 --> 00:35:28.320 AI improves, as you know, all the time, but there are still a lot of issues though, when you have to uh do a certain type of high-risk decision based on the data that AI gives you. 00:35:28.480 --> 00:35:30.079 So you need to know its limitations. 00:35:30.239 --> 00:35:44.559 So, in some cases, these like these issues where AI would scrape the internet and bring back certain things upon your strat upon which you can then decide your strategy. 00:35:44.719 --> 00:36:11.920 That is for me high risk, and I would not 100% go, it's fine, AI will sort it out because it still doesn't understand language, it still doesn't understand behavior to the to the level that we need it to, and there are still too many suppliers in, you know, I've been in Asia for 12 years, too many suppliers, but even the small social enterprises in the UK and the US that do not have a strong enough digital footprint. 00:36:12.159 --> 00:36:22.559 So if you use AI on to use AI and nobody else, no human in the loop for decision making that impacts stuff, you know, right? 00:36:22.800 --> 00:36:26.719 Like that, you are losing out, and a lot of people are losing out. 00:36:26.800 --> 00:36:35.440 So I'm saying still a lot of false positives in some amazing ESG uh software that was scraping using AI. 00:36:35.679 --> 00:36:44.239 And also, I'm seeing a huge disadvantage of small businesses that do not have a strong AI footprint. 00:36:44.480 --> 00:36:47.039 So depends, you see, depends. 00:36:47.199 --> 00:36:50.960 I tell you another one that I'm geeking out on cataloging. 00:36:51.199 --> 00:36:52.159 So boring. 00:36:52.559 --> 00:36:58.320 But the reason why I think we need to catalog is not because AI now can find it on the net. 00:36:58.480 --> 00:37:00.880 I don't want AI to find it on the net. 00:37:01.039 --> 00:37:05.920 I don't want that because I want to restrict my people purchasing IT. 00:37:06.079 --> 00:37:10.079 I want them to go to my internal catalog first. 00:37:10.400 --> 00:37:28.800 So having enough information about what you have in stock in order to take from that first and then go outsize, or even being able to do zero-touch invoicing because your base

data is correct, rather than doing more sourcing activities. 00:37:29.039 --> 00:37:33.760 Like so it becomes hyper hyper-focused on the use case, is what I'm saying. 00:37:33.920 --> 00:37:42.719 And that's where I'm also thinking like saying AI is good or AI is bad is so last century because we need to be really specific in what it does. 00:37:42.800 --> 00:37:45.199 It's like you're saying, what problem are you trying to solve? 00:37:45.519 --> 00:37:46.880 Yeah, it's fascinating. 00:37:47.119 --> 00:38:01.920 So I I guess here's the sort of ultimate question in all of this, which is you know, do you feel that well, I mean, I guess maybe this is another last century phrasing, but do you feel that procurement will be augmented or replaced eventually by these tools? 00:38:02.000 --> 00:38:07.760 And there are different companies at DPW, some are tools to help you do more work faster. 00:38:07.920 --> 00:38:12.079 And then there was one company I remember where they were wearing these headbands that said hire me. 00:38:12.320 --> 00:38:16.320 And I thought they were young kids trying to get internships or something. 00:38:16.480 --> 00:38:18.320 And what they actually meant, it was cheeky. 00:38:18.400 --> 00:38:27.360 They were saying, hire me, but really hire our agent, our bot uh to do to do the job that you would have otherwise hired a person for. 00:38:27.519 --> 00:38:32.480 So that might be one question, or maybe the other question is just how does the job shift? 00:38:32.559 --> 00:38:39.119 Because it could be that the procurement managers of today end up looking like data scientists tomorrow, or maybe it's the opposite. 00:38:39.280 --> 00:38:51.760 Maybe the data science is really so taken care of now that it becomes purely about the in-person interactions and purely about the relationships, which are things that internet-based LLM just can't do. 00:38:51.920 --> 00:38:54.239 I mean, there's no physical body for an LLM. 00:38:54.400 --> 00:38:58.960 It can't, it can only do things that involve reading and writing from a screen and clicking and typing. 00:38:59.039 --> 00:39:03.760 I mean, those that is the full scope of what it's it's its input-output system is. 00:39:04.000 --> 00:39:06.320 So, you know, I'm curious to get your thoughts. 00:39:06.400 --> 00:39:08.559 You know, how does that how does it change the job? 00:39:08.719 --> 00:39:18.320 Just summarizing my my compound question, replacement versus augmentation, and then how does the job change over the next you know five to ten years if if if augmented? 00:39:18.800 --> 00:39:29.039 The roles would definitely shift and it'll definitely augment it, just as like old tech, you know, as you know, I've had a stint in SAP AREBA. 00:39:29.119 --> 00:39:32.320 Before I had a stint and SAP Ariba, I implemented SRM. 00:39:32.480 --> 00:39:36.559 Was that painful over like 75 countries in a in a bank? 00:39:36.800 --> 00:39:45.840 And every time we thought that procurement will be, you

know, replaced or people will be replaced, we won't need so many people. 00:39:46.079 --> 00:39:53.280 I would say that maybe there is a shrinkage, shrinkage of the number of people, but the roles move around. 00:39:53.440 --> 00:40:03.599 So from having roles, having bigger category teams, you might have smaller category teams, but you would have a bigger data science and analytics and an insight team. 00:40:03.920 --> 00:40:14.400 But you cannot replace people hundred, well, you can replace people 100% if the role is SOP based. 00:40:14.639 --> 00:40:23.360 So if you write everything down and you do everything by a process step, then totally procurement can be replaced. 00:40:23.599 --> 00:40:34.880 So my challenge to procurement when I talk to them is actually the people of the future, they say, oh, there's going to be a small uh, you know, team of uh deep commercial experts. 00:40:35.280 --> 00:40:48.639 For me, actually, the team is gonna be deep tech and tech that understands LLMs and Gen AI because you need to still make sure it doesn't go rogue or it doesn't go crazy or it doesn't protect itself. 00:40:48.719 --> 00:40:54.719 Yeah, it does what it wants and it's exciting, but you have to have people looking after it, right? 00:40:55.199 --> 00:41:06.400 You have people who are very strong commercial experts, but actually the people who are going to evolve and thrive are psychologists and behavioral scientists. 00:41:06.639 --> 00:41:19.920 Because at the end of the day, what you want to be doing is you want to influence your internal stakeholders to buy in the way that mostly benefits the organization, not the individual people or egos or whatever. 00:41:20.159 --> 00:41:23.039 So for that, you need to understand human behavior. 00:41:23.360 --> 00:41:31.199 So actually, procurement always says, like, oh, you know, you should speak to marketing different finance, but do they actually know what that means? 00:41:31.519 --> 00:41:35.119 Is that is that an Excel versus a pretty PowerPoint? 00:41:35.280 --> 00:41:41.039 And that's the extent of maybe personalization that they go to. 00:41:41.360 --> 00:41:46.239 For me, I think procurement that survives is going to be hyper-personalized. 00:41:46.320 --> 00:41:55.039 And that's why I'm geeking out on this behavioral scientist thing, because I believe that we need to start behaving like marketing. 00:41:55.199 --> 00:42:09.519 And we need to profile, profile the organization, really understand what they want, create pathways for them to buy or curate the budgets in a way that is hyper-personalized for stakeholders. 00:42:09.679 --> 00:42:14.320 So you're already being very specific as saying direct and indirect is not the same. 00:42:14.480 --> 00:42:16.880 And I agree with you, but I'd go even down. 00:42:16.960 --> 00:42:44.320 I have that cool conversation with somebody, and I said, if I start my CPO career again, rather than just putting a rebound, I would understand what my three top categories of spend are and really find the solutions that address the pain points of these categories and then bring it all together with a shiny layer, like not intake layer, but a shiny general workflow management layer. 00:42:44.639 --> 00:42:51.199 Because the behavior of every single, you know, department is very, very different and their needs are very, very different. 00:42:51.360 --> 00:42:58.960 So if we want to stay relevant, we would need to understand them so well that we're really in step with how they think. 00:42:59.039 --> 00:43:00.800 And that's what AI can give us. 00:43:00.960 --> 00:43:02.800 So that's what I'm very excited about. 00:43:02.960 --> 00:43:09.039 So I think we're gonna have behavioral scientists sitting there, data geeks and analysts sitting there, and commercial people. 00:43:09.119 --> 00:43:12.159 But commercial and behavior people, it's like a hostage negotiator. 00:43:12.239 --> 00:43:13.440 Do you know that guy, Chris Voss? 00:43:13.679 --> 00:43:14.559 Uh, Scorpio. 00:43:14.639 --> 00:43:14.800 Yeah. 00:43:15.039 --> 00:43:15.280 Yeah. 00:43:15.440 --> 00:43:18.800 Well, he's coming to uh to for his show. 00:43:19.039 --> 00:43:24.079 He's coming to do a show in April in London that I'm gonna go have a look at. 00:43:24.159 --> 00:43:27.599 Because I I'm I've followed him quite a lot, but I want to see him face to face. 00:43:27.760 --> 00:43:33.280 But this guy is is uh steep deep and he's doing negotiation, but it's all psychology. 00:43:33.440 --> 00:43:35.039 It's everything is psychology. 00:43:35.119 --> 00:43:39.440 So if we can do that, if we are up for it as procurement, we're gonna thrive. 00:43:39.599 --> 00:43:43.039 We're gonna be a commercial function, commercial function. 00:43:43.280 --> 00:43:55.440 Otherwise, we're gonna be replaced because if we put everything on SLPs, three quotes an hour, uh, you know, and this is how you decide a supplier, yeah, that AI can do at scale and better than we do. 00:43:56.400 --> 00:43:57.599 Makes a ton of sense. 00:43:57.840 --> 00:44:00.559 Well, I this has been an incredible conversation. 00:44:00.719 --> 00:44:02.480 I think that's needless to be said. 00:44:02.639 --> 00:44:30.480 Um, we've talked touched on a range of topics from uh the mastery of marketing and procurement at PNG uh 30 years ago to retooling and retraining the procurement function for the modern era, questions of replacement versus augmentation, uh the actual uh how to stay close to the business uh and whether SOPs will lead to uh standard operating procedures will lead to job replacement over job augmentation. 00:44:30.639 --> 00:44:33.440 So this has been a fascinating, fascinating conversation. 00:44:33.519 --> 00:44:37.840 You've given me and our viewers a lot to chew on. 00:44:38.000 --> 00:44:46.079 So I think we're gonna have to do another session because I think we've just gone through part one of maybe part two uh or multiple more. 00:44:46.159 --> 00:44:48.320 So thank you so much, Victoria, for joining today. 00:44:48.639 --> 00:44:48.960 Thank you. 00:44:49.039 --> 00:44:50.559 Thank you for inviting me, Spencer. 00:44:50.639 --> 00:44:51.519 It was an absolute pleasure. 00:44:51.599 --> 00:44:53.280 It's great to see you again.