PRODUCT

PRODUCT HIGHLIGHTS

SOLUTIONS

The value of a technologist's mindset in procurement transformation

How to approach procuretech and artificial intelligence in procurement digital transformation

Episode highlights


When choosing a solution, start with the process
Play
Comparing the maturity of the procuretech with other disciplines
Play
Procurement's access to internal and external data
Play
Hard vs soft savings in spend management and delivering ROI
Play
AI in procurement
Play

Posted by: ZivioReading time: 96 minutes

With Vincent Teyssier, Partner for the Digital Transformation and ProcureTech practices, Devolut

00:00:00 - The core skills of building trust and relationships in procurement
00:07:05 - Procurement's access to internal and external data
00:13:35 - Communication skills and personality types in procurement teams
00:16:45 - Comparing the maturity of the procuretech with other disciplines
00:26:15 - When choosing a solution, start with the process
00:29:20 - User centric design in procurement technology
00:37:30 - Hard vs soft savings in spend management and delivering ROI
00:46:20 - Comparing in-house procurement and consultancy experiences
00:53:10 - Addressing what works as well as what doesn't in driving change
00:55:00 - Collaboration with different functions in value creation
00:58:00 - AI in procurement

Transcript (Auto-generated. Please excuse any minor mistakes)

00;00;00;00 - 00;00;14;07
Jonny Dunning
Brilliant. So I'd like to introduce my guest to today's podcast, Vincent Fischer, who is partner for Digital Transformation and also to procure tech practices at Devolut. Vincent, thank you very much for joining me. How are you?

00;00;14;09 - 00;00;21;08
Vincent Teyssier
I'm great. It's a great set on a Friday afternoon in Singapore. What is a good time to record some podcasts with you?

00;00;21;11 - 00;00;28;06
Jonny Dunning
Yeah, and looking forward to some relaxing Friday evening entertainment and socializing afterwards. I see.

00;00;28;08 - 00;00;31;06
Vincent Teyssier
Oh, definitely opening a bottle of wine after that.

00;00;31;08 - 00;00;53;04
Jonny Dunning
Excellent stuff. Good to hear it. Cool. So we're we're here to talk today about some topics around digital transformation and procure tech. I've seen I've seen you on podcasts before. I've seen some of your content. I really enjoy the stuff you put out there. You've got to quite strong points of view and some a real kind of unique approach in the way that you talk about some of this stuff.

00;00;53;04 - 00;01;14;28
Jonny Dunning
So I'm really looking forward to dig into some of these topics with you today. And just in terms of a bit of background, you're clearly very passionate about procurement. And I know you've kind of talked to me about how it's such an incredible function for learning business. How did it all start for you and how did you end up doing what you're doing now?

00;01;15;00 - 00;01;42;12
Vincent Teyssier
Yeah, first of all, I'm absolutely not a procurement person. I'm a technology person. I code since I'm ten years old. I was working in marketing technologies before moving to procurement, and initially I got an offer to move to procurement in Singapore. I was very excited by the move. Much less buy is a function. When you're in marketing, you look at it as a non sexy function as boring, to be honest.

00;01;42;14 - 00;02;21;11
Vincent Teyssier
But then I arrived in Singapore. Yes, it's full of processes, full of compliance and people are not especially joyful, but ultimately it's a mindset. It's up to you to make it fun for yourself and to to dig into it. And the more I was digging into it, the more understood to that it's a function that all deals are departments in your company, you work with I.T, you work with finance, you work with marketing, sales, all the different functions, whether it's through digitalization programs to help the users to order whatever they need to order, but also some on some high projects.

00;02;21;13 - 00;02;48;15
Vincent Teyssier
And I was in Telenor, which is a telecommunication company and, and then we were sourcing a multi billion contracts which definitely touch a big part not only of the organization but also of the top management. So very quickly became extremely interested to be exposed to so many different processes within the company, but also to be exposed to top layer, top management.

00;02;48;18 - 00;03;24;23
Vincent Teyssier
You don't get that anywhere else. This is really the only one which tells everything. And for me, what makes something very interesting is a diversity of things you're doing. Let's say that today I'm looking at compliance through actual change settings. Tomorrow I'm going to look maybe at improving the efficiency of of the auction model, or I'm going to look at creating adoption falls of pursuing requests, or I'm going to look at benchmarking performance management.

00;03;24;29 - 00;03;56;25
Vincent Teyssier
It's extremely diverse what you can do in it. Also, you're not centered internally procurement is how internal you demand management in stakeholder management and project planning, but it's also how does external supplier relationship management, when you tendered to the market, when you do market analysis and gather benchmark data from external sources, it's really, really not as it sounds from the external.

00;03;56;25 - 00;04;22;03
Vincent Teyssier
It's a very sexy function actually. It's a lot of process re-engineering. It's a lot of digitalization efforts and it's a lot of common sense. I think probably 80% of procurement is common sense. Of course, you need the experience, you need a bit of knowledge, but common sense will get you very far away immediately. Yeah.

00;04;22;05 - 00;04;38;09
Jonny Dunning
And as you as you kind of came through the different things that you've done, how important has it been for you in terms of kind of building your network out and expanding your reach in in, in the way that you have done coming through to what you're doing now?

00;04;38;11 - 00;05;01;17
Vincent Teyssier
Yeah. So I think that it forces you definitely to go outside of social in procurement. You do not spend all your day crunching numbers or writing emails. You need to meet people, you need to create a relationship. People need to trust you to understand the values that you provide. And so definitely going out, meeting people, establishing relationship is a big part of the work.

00;05;01;17 - 00;05;38;08
Vincent Teyssier
I would say. I think we'll be going to talk about transition to consulting a bit later, So that's a big part of it. But I think that also one thing which is very important and I see the major vendors in the procurement space insisting on the community aspect of it because when you are too focused internally, you tend to miss actually a lot of improvement opportunities and suddenly you're being at the conference from a vendor or you're going to meet all the customers from a vendor that you're using.

00;05;38;11 - 00;06;06;16
Vincent Teyssier
And these guys are the same challenges as you, but maybe they solve them very differently. And exchanging is a huge part of bringing new ideas to the business. And I was talking about performance management recently and benchmarking. And when you benchmark, you look at the leader score one of the traditional pitfall events to just try to copy the leaders and copying instead of innovating is not going to bring you very far.

00;06;06;19 - 00;06;42;19
Vincent Teyssier
The reasons of leaders as a leaders is because they were centering their views on their problem and they sold them by being creative. If you just copy a recipe that works, yes, it's going to be being movements, of course, but you're going to plateau. What's on it? At some point? It's really about understanding the big picture, the specifics of your industry, of your organization and of your landscape by looking at it and being creative that you can really aspire to to become a leader in a very trying to achieve more.

00;06;42;21 - 00;06;53;21
Vincent Teyssier
So yeah, this community aspect for me is essential in procurement and it's really, really developed for me even more than in most of the users functions.

00;06;53;23 - 00;07;12;27
Jonny Dunning
Yeah, I see that. That's kind of the reason why we're having this conversation, isn't it, To, to, to get the dialog going to put stuff out there. A couple of really interesting points you, you brought up that I just want to have a look at. Procurement is in a quite unique position in the way that it touches so many different functions within an organization.

00;07;13;00 - 00;07;51;15
Jonny Dunning
But I think it's also in in a very interesting position in the sense that it has access to and is is a kind of a fulcrum for a lot of different data sets and data points. And one of those is the internal versus external balance that you talked about. I find that particularly fascinating. As you said, we can talk in a minute about your cross to the consulting world from kind of practice in house, but that's quite unique viewpoint really, isn't it, to be able to see what's happening within the organization and how the organization views their external kind of suppliers, which if it's delivering goods, that's fundamental to their business.

00;07;51;18 - 00;08;10;25
Jonny Dunning
If they're delivering services, they're almost like extended capacity of the organization. But to get a view of what's happening internally and what the internal stakeholders think of the external suppliers supply chain and the other way round, I think that's a really unique viewpoint, just being able to see well, you know, we don't want to operate in a vacuum as a business.

00;08;10;25 - 00;08;23;06
Jonny Dunning
What will the other organizations that are doing clever, cool stuff, what are they? What are they think of what we're doing? I like you say that kind of innovation and creativity, that's where that's the sexy part of it really, isn't it?

00;08;23;08 - 00;09;02;16
Vincent Teyssier
Yeah. So whether you're a procurement category manager or category director or your process Center of Excellence, your goal is to be data driven, you need to have all the data. So you start to build a performance management framework. You start with basic KPIs and you see the allegations from these KPIs. The same when you gather data from external benchmark and from internal sources and you compare them, you start with the basics and then you start to find opportunities of improvement or opportunities in the market.

00;09;02;18 - 00;09;28;28
Vincent Teyssier
But very quickly you will reach a limit because you don't necessarily know what is the business process behind the data. So you need to dig constantly to try to explain what is happening behind it. For, for example, I remember we're implementing some processes and we look at the ledgers and we see some reversal transactions and we're absolutely not understanding what was happening that was skewing one KPI.

00;09;29;00 - 00;09;52;07
Vincent Teyssier
So we have to get to the bottom of it and we understood that it was substitution, that you receive a product without a purpose for it, and then six months later when you sell it in telco, you have prepaid and postpaid, you suddenly assign. It's the purpose and this purpose as a very different financial meaning, whether it's prepaid and postpaid.

00;09;52;10 - 00;10;22;23
Vincent Teyssier
So you had that kind of reversal then rebooking of the same transaction, and that's an example of how deep you can dig into the data and really understand what is business process behind. I discovered processes. I had no idea where existing consignments, substitution items, exchange rate variants. When when you look at accounting in Europe, exchange rate variance, yes, most of the people don't truly know what it is.

00;10;22;26 - 00;10;43;22
Vincent Teyssier
But in Southeast Asia, where you have a lot of variance in the currency, the impact on panel is actually enormous. So they book it and they're very, very rigorous in calculating it. So, yeah, it's just you look at the data you need to dig, you need to find what is the root cause, what is the root process behind it.

00;10;43;24 - 00;11;12;19
Vincent Teyssier
And I find this really captivating evidence I'm quoting since I'm ten years old, and that's what I always loved when I was coding, is that you need to find what's wrong in your code, right? You need to dig, you need to grind the tangles. And that's exactly what what procurement gives you. It gives you the opportunity to understand so many processes in very deep details and to improve efficiency one process at a time.

00;11;12;21 - 00;11;33;27
Jonny Dunning
Yeah, that is quite like coding really, isn't it, In terms of the creativity of seeing an overall architecture, a developer being able to see an overall architecture in their head of where they want to get to, what fits, where, how you can add new processes in without destroying old processes, without making the whole thing totally inefficient. Yeah, that's an interesting, interesting kind of angle to view it.

00;11;33;29 - 00;12;02;20
Vincent Teyssier
It's a process, you know. And to put it, I've talked a lot about performance management, but to put it more in that, you know, category management perspective, when you go in front of the top leaders of your organization, I'm going to push for a spend of, let's say, $200 million. This is enormous. You know, there's enormous pressure to find the right vendor to make sure that you have the scope right, that every all the terms of the contract are right.

00;12;02;20 - 00;12;21;16
Vincent Teyssier
There's so many different elements in that. Right. And that you can only do it if you have an efficient data processing going because you need to have the right data. You have no idea what the management will ask you, but they will ask you what is this? Some vendor do we have existing spend with them? How much do we spend from the world group?

00;12;21;18 - 00;12;55;10
Vincent Teyssier
And suddenly everyone is panic? We don't know, right? You cannot know everything, of course, But that's where procurement digitalization takes place. It's in facilitating the access to all kind of data that would answer all kind of questions. When top management sees that you have answer and data driven answers to these kind of very specific questions, the trust is much stronger and therefore you get more autonomy in driving your earlier projects.

00;12;55;13 - 00;13;23;00
Jonny Dunning
Yeah, and I always find that an interesting contrast in the sense that you need to be very data driven. And when you're talking about these overarching processes and the architecture of bringing all of these processes together, obviously you can have differentiation and function within procurement, but that's quite analytical, very structured, logical thinking. But then you've got to manage different people, you've got to manage different stakeholders, you've got to manage the C-suite, you've got to manage, you know, buying stakeholders within the organization.

00;13;23;07 - 00;13;35;00
Jonny Dunning
You're interacting a lot with finance, etc., etc.. So I think it's it's I always see that as quite an interesting contrast in the skills required to be able to deliver that kind of high performance and procurement.

00;13;35;02 - 00;13;54;26
Vincent Teyssier
So one thing that I done with my team, we have team meetings every second week and I was always bringing topics which are outside of procurement to try to upskill everyone. And for a while we looked into details about the desk, you know, what is this desk? Political tests.

00;13;55;03 - 00;13;56;19
Jonny Dunning
Are okay.

00;13;56;21 - 00;14;23;01
Vincent Teyssier
A bit like Myers-Briggs, right? Know people who were red were blue, you know, so analytical, emotional data and goal oriented. So we went into that and we did role play. I took some what was empathy and I tell, okay, you're going to play the role of the really angry manager who's red and outcome driven, write the exact opposite.

00;14;23;01 - 00;14;52;04
Vincent Teyssier
And then I took someone who was outcome driven. I'm going to ask them to be a soft, emotional person. And we do a role play and we analyze how different characters are going against each other. And all the team members say, Okay, what do you see in this discussion and this role played? I think the ability to recognize what type of character you have in front of you and to adapt your speech to this kind of character is very, very powerful in driving positive outcomes.

00;14;52;11 - 00;14;54;17
Vincent Teyssier
Otherwise you just drive conflict.

00;14;54;20 - 00;15;15;05
Jonny Dunning
I mean, it's a super high level communication skill, but you know, a lot of people at most I'd say most people are capable of it because it's kind of an inherent human quality, but you have to kind of recognize it first, don't you? Might be ignoring it or putting it to the side, or it might be not so not be something you kind of necessarily doing naturally.

00;15;15;07 - 00;15;20;23
Jonny Dunning
But I do think that for a lot of people, that's something that they can they can do it. They just need to recognize it.

00;15;21;17 - 00;16;03;12
Vincent Teyssier
I think that once you play this kind of role, role game, you start to see it everywhere. That was the training that I went through long time ago, maybe ten years ago, and I still have it in mind. I remember being in front of my CFO one day and I'm explaining him the importance of the project. Very important, not emotional, but very engaged in the discussion and then at some point I cool down and I think he's a CFO, He wants figures, I need to throw all the figures I need and that really helped the discussion actually.

00;16;03;14 - 00;16;33;23
Vincent Teyssier
You have some people you need instead of going straight to the point, you need to really explain all the detail very carefully, very attentively, and being careful about the emotions. When you do that, you just need to do it once or twice. Just for fun. We do it with your wife or whoever, just for fun. And that's going to become a reflex, actually, as something that the discussion you're going to cool down and you got to think again Who do I have in front of me?

00;16;33;26 - 00;16;38;22
Vincent Teyssier
How should I adapt my my speech, my, my style?

00;16;38;24 - 00;16;54;10
Jonny Dunning
It's a it's a great skill for everyone, basically, isn't it? Yeah, but just just another thing I wanted to touch on. It's a bit of a kind of a personal point of interest. You, you said that you came, you came from did you come from actually, did you do some work in marketing?

00;16;54;13 - 00;16;58;17
Vincent Teyssier
Yeah. Well, you think I was responsible for the system sending email, basically.

00;16;58;24 - 00;17;23;04
Jonny Dunning
Right. Okay. So, so when you look at the marketing technology sector, industry, whatever you want to call it, and you look at procurement, technology industry, I always I always find that an interesting comparison because I feel like the procurement technology stock is growing massively. There's huge innovation and, you know, lots of companies coming in to procurement technology, lots of investment going into it.

00;17;23;07 - 00;17;45;12
Jonny Dunning
But I feel like it's really early if you compare it to something like the maturity of marking technology in terms of just the value of the market tech marketing technology sector, the investments got into the massive unicorn giant companies that exist in MarTech marketing technology compared to procure tech. It feels like procure tech are right at the start of the process compared to marketing technology.

00;17;45;15 - 00;17;47;17
Jonny Dunning
Would you agree with that?

00;17;47;20 - 00;18;22;21
Vincent Teyssier
Hmm. Well, the ecosystem in marketing technology is gigantic. Yeah, but also because every niche application actually can unlock really high value. Ultimately, you capture market shares with marketing in procurement technologies. Yes, we are early, but I also think that most of the solutions are way too niche today. If you look at the spider map from Dr. Luis Epstein, you see hundreds of technologies.

00;18;22;24 - 00;18;53;16
Vincent Teyssier
If you implement five, it's already going to cost you millions, so shoot them in ten to maintain the processes, to maintain the knowledge. It's insane, right? So there's a need, I think, of consolidation in the in the market. Um, we saw initially the best of suite model that was working very well ten years ago. Then at the same time it most of them and sees develops integration platforms.

00;18;53;16 - 00;19;29;08
Vincent Teyssier
So when integration is not any more a hurdle, the best of breed bundle started to be pushed and and today the recommendation everywhere is go for best of breed right but we start to see a reversal not back to best of suites but but just in the middle something which is in between having synergies of suites a more integrated model so that you don't have a plethora of very niche solutions, but something that's also have some some type of expertise.

00;19;29;11 - 00;20;05;12
Vincent Teyssier
I think where it's becoming the most advanced today is in the cloud market. I think in this market you get a lot of solutions that provide a lot of niche functionalities, contract mining obligations, management, OCR, close management, you've even redlining collaboration. I think these are very well integrated for one specific purpose of the procurement process in the P2P area.

00;20;05;14 - 00;20;34;28
Vincent Teyssier
It's pretty much integrated as well, but I think all the rest, which is very, very, very niche, if you look at whether it's vendor discovery, risk management for each and every risks, you have a plus platform, basically supply chain like your digital twin. There's so many solutions. What do you choose when you make an impact ultimately boils down to how you you drive your strategy internally.

00;20;35;01 - 00;20;39;17
Vincent Teyssier
But I think there's a lot of room for consolidation in all these niche solutions.

00;20;39;17 - 00;21;01;27
Jonny Dunning
So yeah, that's it's a bit of a conundrum for the solution providers because you need to you need to deliver value, but that value needs to be distinct enough to fit into a gap that you have in your existing infrastructure, which, which naturally drives people towards more niche solutions.

00;21;01;29 - 00;21;21;01
Jonny Dunning
Because most organizations are going to have, you know, a POC process to procure to pay platform in place. And they're probably going down the route of, if not already in place where the source to pay solution. And a lot of organizations are struggling to make those SDP solutions work. And maybe they kind of thought it was going to do everything.

00;21;21;08 - 00;21;48;16
Jonny Dunning
And that was the pitch. I'm maybe a bit further down the line. They're realizing that it can't necessarily be all things to all people. It can be very effective. But there are, as the industry develops and as people want more granularity of data in particular areas, there's certain things that we need more specialized solution. So so the the kind of best of breed solution providers are trying to create something that is almost like the smallest solution that can deliver the maximum value in their particular area.

00;21;48;18 - 00;22;07;06
Jonny Dunning
Because as soon as you start going broader than that, then you're into all sorts of potential conflicts with like, well, this part of this system crosses over with this part of that system and so it's an interesting conundrum. And I think you're right. I think there will be market consolidation and that will make it easier, but it's almost probably quite a similar process.

00;22;07;06 - 00;22;42;18
Jonny Dunning
You mapped it out. It probably happens in most sectors where you're getting this development of the technology in a particular area, because if you I don't think there's much uniformity or standardization in the way procurement teams are structured. Operating models is less there's less overall difference in the basic operating models. But in terms of how the tech stocks are set up, for me, a lot of times it's there's so much variation that I don't think it's that easy to just say everybody thinks of the world in these five segments or these two categories.

00;22;42;20 - 00;22;45;20
Jonny Dunning
It's it's it's quite a lot of differentiation.

00;22;45;20 - 00;23;03;05
Vincent Teyssier
Yeah. So the reason I'm insisting on cognitive load and train my team on that is because when you design a process, this is exactly the same thing. And that's what my team was doing. They were redesigning processes. Basically you need to have in mind the end user.

00;23;03;07 - 00;23;35;23
Vincent Teyssier
You need to reduce as much as you can the cognitive load when you're designing this process for the end user. Why? Because then you're not relevant only for experts, you're relevant for the user lambda and ultimately satisfaction and adoption comes from easiness of use comes from minimizing the cognitive load. When you do that, when you deploy P2P system or contract systemic cetera, what is the number one critique that you hear?

00;23;35;26 - 00;24;05;14
Vincent Teyssier
I don't know how to use the system. I don't know how to do this, how to do that. That's a major issue because if cognitive load is too high, because for the sake of process optimization, a lot of complexity has been added and it's not necessarily easy to remember. I remember we made some analytics on on piles and invoices and we figured out that 80% of the invoice is were handled by 20% of the users.

00;24;05;16 - 00;24;22;20
Vincent Teyssier
That means that the remaining 80% of the users were having a very little amount of viewing invoice in the system. And the average was that they had to invoice to approve that. Yeah, that means they log into the system every six months. They don't remember anything.

00;24;22;22 - 00;24;24;12
Jonny Dunning
Of course, they're not changing.

00;24;24;14 - 00;24;48;26
Vincent Teyssier
For the first time to the right. Maybe sometimes it is right to, sometimes it's difficult and that's normal. That's not the job that it's our job to make it easy for them. So really, I was training my team on that because you always need to have in mind the end user. It's customer centricity. It's exactly the same as in marketing.

00;24;48;29 - 00;25;20;06
Jonny Dunning
I think it applies. So it applies to coding, it applies to process, it applies to decision making very clearly so that that has applications in how tech providers best of breed tech providers like ourselves and, you know, procure tech providers when your when you're proposing a solution to an organization, to a procurement buyer, the cognitive load of what you're putting across has to be absolutely minimized.

00;25;20;06 - 00;25;49;20
Jonny Dunning
So it has sales implications as well. It kind of reminds me a little bit of that idea about the confusion of too much choice. You know, if you have a restaurant menu that's got too many items on it, people people just don't know what to choose and have a bad experience. If you've got three items that I can make a choice, but I'm, you know, which is another thing that drives it into these quite specific solutions, because if you've got a solution that's too broad, then people are going to look at it and say, I feel like this is going to run into other things.

00;25;49;20 - 00;26;09;02
Jonny Dunning
It's going outside my remit to say to this stakeholder and this stakeholder over there is going to call in these different functions and it's going to potentially cross over with something we do in this piece of technology here, in that piece of technology there. So I think it's critical for solution providers in the industry to be able to conquer that as well.

00;26;09;04 - 00;26;32;21
Jonny Dunning
But ultimately, it kind of comes down to what you're talking about, which is the the process. But understanding the process and I know that's something that you advocate very strongly for in terms of when you're looking at your technology stack and you're looking at that best of suite versus best of breed or a hybrid solution, it's about addressing the big problems and starting with the process.

00;26;32;24 - 00;26;47;14
Jonny Dunning
And how do you know, do you find much resistance to that within teams of people? Kind of like jumping at solutions where they see something shiny. You have to kind of bring people back to the process? Or do you think it's something that people get straightaway?

00;26;47;16 - 00;27;14;22
Vincent Teyssier
Well, that depends on the stakeholders. Ultimately, there's one thing they're seen working very, very well that I learned in 90. You know, it's user centric design. So when you want to implement a system. So the last one I did like that demand management and portfolio management system. I start with the person who is going to use it all.

00;27;14;24 - 00;27;41;09
Vincent Teyssier
It's a procurement manager, it's the CEOs, its top management, and it's maybe the end user or the functional stakeholder. And for each of them I start to write a user flow because they will not do the same thing in the system. So we write the in for I log in to the system for what? For consulting reports, for inputting new initiative, for updating initiative, for XYZ for that.

00;27;41;12 - 00;28;09;01
Vincent Teyssier
And then thanks to that, you can build first roles and responsibilities that you configure in the system, a kind of permission table to make sure that they have the experience of they're supposed to have as they see what they're supposed to see. A lot of frustration is, Oh, I cannot see my contract, for example, I cannot see that when you do user centric design, that you minimize this type of problem to start with.

00;28;09;04 - 00;28;42;16
Vincent Teyssier
And you really took granular in granular to the person who's supposed to use the system and then remembers that on the launch, this is a system which had the most positive feedback because finally it was designed for specific people and not for specific functions. What we see in the Big four implementation of large systems, they are very strong structures, have very good methodology, but it's not user centric, it's process centric.

00;28;42;18 - 00;28;53;22
Vincent Teyssier
We need to force the process on the users. And that's why you have problems with adoptions. Yes, beautiful processes, but you need to think of the users as well.

00;28;53;22 - 00;29;15;13
Jonny Dunning
Yeah, that's a really great point because a lot of the time when a problem is addressed, is addressed with with a function in mind, rather than going maybe into a bit more depth to look at the the stakeholders within that particular function. But again, that can be quite company specific. So doing that sort of analysis internally is extremely useful.

00;29;15;15 - 00;29;36;25
Jonny Dunning
So I mean, how well do you think most organizations utilize the technology they've already got? What you know, bearing in mind you look out there, there's this extremely complex ecosystem of providers in the kind of best of breed category, lots of new solutions coming onto the market, new approaches to the solution to the to the workflows of procurement.

00;29;36;27 - 00;29;59;12
Jonny Dunning
You know, where does intake happen, how to utilize AI, all these different factors, the importance of sustainability, reporting and things like that. How effectively, what sort of start point are most organizations coming from in the sense that, you know, do they really have an accurate understanding of what their technology capabilities are first before they look outside?

00;29;59;14 - 00;30;25;05
Vincent Teyssier
It really depends. Of course, you have top performance. I've seen a few. They are absolutely amazing how they perform in terms of metrics in terms of engagement, and they don't necessarily have the simplest technology landscape. It can be very convoluted and very complex. But this one thing I've seen, which is constant across the top performance is a metrics mindset.

00;30;25;08 - 00;31;00;10
Vincent Teyssier
It's a performance management culture. Everything is measured. And I remember the first time I came across such organization, we were actually exploring the same solution. So so we discussed a bit and my counterpart said, okay, did you make a list of, of criteria for evaluation? I said, Yes, yes. We have a few questions that we're going to test the the solution for testing to a few users, and then we have a few questions for the users is like, yes, sure, but how do you evaluate it?

00;31;00;12 - 00;31;20;29
Vincent Teyssier
And it well, we have the questions and will compile the feedback, right? Yeah, but this is commentary. What do you do with that? And that's the first time it really struck me is that, yeah, it's compulsive, it's very subjective. Actually. What I should do is get the comments or it's very important to have the score is give a score 0 to 5.

00;31;20;29 - 00;31;49;14
Vincent Teyssier
Just a feeling from the person. How user friendly was it. 0 to 5. How's it. And then you compile this score, you can adjust the weightage and everything, but this is having a metrics mindset is putting figures everywhere. You don't need to put targets everywhere, but you need to measure everything because then you understand what are the real bottlenecks, what are the real problems that you need to address, and even what is the size of the problem.

00;31;49;16 - 00;32;14;00
Vincent Teyssier
And also thing that is a bit controversial is typically when you assess technologies, you have your metrics which is cost versus impact. And one day we're brainstorming about that and someone came and told me, hey, we move the costs. We don't need the costs. Of course we need the costs. If we don't have the budgets, there's no need to explore it.

00;32;14;03 - 00;32;28;07
Vincent Teyssier
I said no, if the problem is big enough, you'll very be likely to put the money for solving it. And he's right. It's all about the problem, all about the size and the impact of the problem. It's not about the cost. The cost is secondary.

00;32;28;10 - 00;32;45;10
Jonny Dunning
Yeah, like you say, it's secondary. It still it's still an important consideration. But in terms of putting forward a business case, if if the if the need is great enough, the problem is great enough, then inherently there's going to be a large amount of value associated with it, whether it's in an efficiency saving or cost saving or whatever it might be.

00;32;45;12 - 00;33;14;28
Vincent Teyssier
And I think on this topic, I would like to dig in on the one specific thing. You talked a bit about procurement in tech a bit earlier. The procurement intake is great, but it's all by user experience. It's all about the end user, the land, the user, not a procurement user, what I would call the amateur purchaser. This is only valid for indirect procurement.

00;33;15;01 - 00;33;21;11
Vincent Teyssier
They do not direct material. All the direct material is purchased by professional purchasers.

00;33;21;13 - 00;33;22;07
Jonny Dunning
Interesting.

00;33;22;07 - 00;33;53;02
Vincent Teyssier
And how do you achieve efficiency through user experience with professional purchasers knows the system because they use it every day. You don't. And certainly procurement intake is reduced to only indirect and a lot of the P2P solutions are actually reduce to indirect use cases. And when you put the size of indirect versus the cost of this solution, suddenly the ROI is much lower.

00;33;53;04 - 00;34;12;26
Vincent Teyssier
So so that's what I was telling earlier, that if you start to have so many solutions, that's going to cost you a lot. But each solution is very niche in addressing specific problems. So how do you make it more generic or how do you consolidate that? Yeah, it's.

00;34;12;28 - 00;34;42;05
Jonny Dunning
Again, it's a conundrum, isn't it? Because in terms of if you're buying direct materials versus buying indirect, let's say, services versus buying goods is very different. You know, if you're buying, if you're buying goods, it's a complex supply chain, but it's a it's a simple thing that you're buying. It's fairly binary. It either is that thing. And that way, that measurement, that number, that quality or it isn't services is completely the opposite in the sense that it's generally a simple supply chain, but it's an extremely complex thing that you're buying in a lot of cases.

00;34;42;07 - 00;35;03;22
Jonny Dunning
Hard, hard to define, hard to tie down. I think the other thing you were saying about insight kind of crosses over into the use of I tools, for example, where I know you're coming through in a bit. I know you're doing some interesting stuff with Gbtc as well, but we've done some stuff around scoping for services requirements, statements of work type, scope creation, that sort of thing.

00;35;03;25 - 00;35;20;02
Jonny Dunning
Very effective to do that. But you also have to consider that when you're going through that process, some people know what they're going to buy and some people don't necessarily need to be led through a process and guided quite so much. They know what they want, but that's easy to take into account as long as you do take it into account.

00;35;20;02 - 00;35;39;04
Jonny Dunning
And like you say, some of the kind of intake intake approaches are based on the approach of assuming that the person doesn't really know anything about what it is they want to buy from the start of it. So I think that will evolve over time. But there has to be slightly different pathways through the process depending on the level of expertise where you can kind of go, I need to skip this.

00;35;39;07 - 00;36;06;11
Jonny Dunning
I need to just get to where I need to get to. And I think that's something that we've seen in practice where we're dealing with the procurement of services, complex services, typically in our use of our system with live customers, we've had to really build that into it in all parts of the process in the sense that an educated buyer who's done this lots of times, like you say, the kind of the the 20% of people that do 80% of the post, they just need to go, go through it quickly.

00;36;06;11 - 00;36;24;01
Jonny Dunning
They know what they need to do, they know how they need to do it. So you need to have the right pathways and again it comes back to user experience, user centric design. So I think that's a really good point and it's something that I don't think a lot of people have picked up on yet. So I think that's quite an interesting, interesting angle on it.

00;36;24;04 - 00;36;48;01
Vincent Teyssier
And most of the vendors that try to sell you as many licenses possible. So they will tell you this is for everyone's procurement should be for everyone. All the organic organization, you have ten cells in a place, 10,000 license. Easy, right? But that's wrong. And there was the odds are extreme. You only buy three and a license, which are only your professional processors.

00;36;48;03 - 00;37;11;24
Vincent Teyssier
Then they will become a bottleneck to organization. So there's the model in between, which is a semi professionalization of procurement, where you have procurement champions in each and every department. And these guys are doing the push for their department. They have more familiarity with the system, with the process, they have better contact with the procurement to transactional team or whatever or tactical team.

00;37;11;26 - 00;37;20;26
Vincent Teyssier
But at the same time that helps the core procurement team by giving additional capacity in terms of manpower.

00;37;20;29 - 00;37;46;16
Jonny Dunning
So one of the things you just touched on there was licenses kind of going on for a bit of a tangent here, but I'm just interested to see what your opinion is. Do you think that the use the license model is still I'd say that's still a prevalent model in procurement technology. But if you look at, for example, the use a license model versus spend based models, how do you see that that what's what's most relevant and how do you see that potentially changing moving forward?

00;37;46;18 - 00;38;15;25
Vincent Teyssier
I think today's is an equal amount of revenue based model versus versus license based. What's important is to have flexibility. You always have a base license, access to the platform, right? And then after that you can have a revenue share or whatever. Some organization are ready to to put a lot of money upfront and are very risk averse when it comes to revenue sharing.

00;38;15;28 - 00;38;38;18
Vincent Teyssier
Some of the companies, they don't have high budget, but they want efficiency. So if they can have it cheaper as a barrier of entry and pay a little bit more in revenue sharing. So some companies would do that. So I think you can have whatever model, if you have a bit of flexibility around that, that's going to broaden your customer base.

00;38;38;20 - 00;39;04;00
Jonny Dunning
Yeah, I agree. I think I think flexibility is required and that's where when you get best of breed platforms coming through and new entrants into the market that were more disruptive, they're typically going to be more agile and more flexible than the big incumbent platforms that have been around for a long period of time. And by the very nature of their success and by the very fact that they've done a great job, they end up being quite inflexible as platforms because they've got so many big customers, so much legacy.

00;39;04;07 - 00;39;24;02
Jonny Dunning
The tech's been around for a while, so I think that's going to be interesting to see how that kind of predictions for transitions and trends in that area comes across. But but one of the so firstly, when you're looking at those costing models, there's the element internally of what I need to know, I need to plan this, I need to plan out those costs.

00;39;24;04 - 00;39;51;14
Jonny Dunning
So if you're leveraging a revenue share, for example, then you know that it's going to be you're going to be happy with the outcome in theory. But there's also just that kind of budget planning. What are we going to be spending on tax? What are we going to spending on? Why? So there's a bit of a kind of contrast there, which is a little bit like with the benefits of using procurement technology and going through a digital transformation, I would say is you've got these kind of hard savings versus versus soft savings.

00;39;51;16 - 00;40;23;09
Jonny Dunning
And if you look at an area of complexity, like if you take a complex services category lot in a consultancy, you might be able to see the hard cost saving of how much money you've saved against budget of what's contracted with the supplier for a for a consulting project, for example. But in terms of the so they've you've got you're kind of hard saving of we've saved X against budget in the procure of the service in the delivery of that service all sorts of things might happen, it might overrun, things might change.

00;40;23;16 - 00;41;00;06
Jonny Dunning
And it's just the inherent nature of for example consulting. So, so measuring the actual return on investment, the value you got from that project is so much harder that I think I feel like to a certain extent that can be beyond some organizations capabilities at the moment that they're so far, far further back down the track that if procurement are likely to be targeted on the hard savings, in a lot of cases, the soft savings and these other things, these more intangible savings around efficiency and around using the best suppliers in the future and all those sort of things, that's surely where it's got to move to.

00;41;00;12 - 00;41;07;20
Jonny Dunning
But I still wonder how many companies right now can buy based on those softer, more intangible savings.

00;41;07;23 - 00;41;38;19
Vincent Teyssier
Yeah, I think we should really stop to measure procurement on hard savings only. Yeah, of course it's important. But let's say that the average saving in the procurement function is about 12%. When you're in an inflationary context of 7%, then suddenly you're just saving actually 5%. That's it. And if you're the CFO of put the target to 10%, how do you achieve that?

00;41;38;22 - 00;42;27;07
Vincent Teyssier
You need to do 70% to achieve that. That's ridiculous. However, cost avoidance, very important obligation management, there's a lot of value that can be unlocked through obligation management, a cinematic organization. They're bringing millions and millions by by efficiently spotting contract deficiencies in terms of obligation and claiming that's money. Basically, like you said in the services as well. Services shouldn't be difficult to track, but because historically that has been a very bad visibility of the spend and of the lifecycle, lots of spend not only is a signature of the contract, but how it evolves because of this lack of visibility.

00;42;27;09 - 00;42;54;11
Vincent Teyssier
There's so much deviation and noncompliance issues. You see guys on the consultancy contract for five, seven years. These guys should have been replaced by the left. Right. It doesn't make any sense. And a lot of room for fatigue and behaviors as well, right? I wouldn't say corruption, but there's these kind of things in the services industry as well.

00;42;54;14 - 00;43;40;05
Vincent Teyssier
It's just about visibility is and analytics. But across the lifecycle of different type of spend, it can become pretty complex. I think there's a lot of value to unlock there in obligation management in general on top of hard savings and cost avoidance, it can be as simple as we have products A which costs 100 today. There's a used supply chain issue for this specific because there's only one chip manufacturer for one component there and therefore we will not buy it for hundred, will not buy for 90, we will buy it for 150 plus 50%.

00;43;40;07 - 00;44;08;17
Vincent Teyssier
But guess what's actually the worst vendor was offering us a 200. So we actually our baseline was 200 and we went down 250. It's not going to show like that on the savings, but this is a reality that the category managers are facing every day. Righty price increase and basically they need to manage the price increase and two less impacted panels go for the one which is the less.

00;44;08;19 - 00;44;28;25
Jonny Dunning
Yeah, it's the devil's in the detail, isn't it? And I sometimes feel that if you if you separate pre contract and post contract, you know, the ability to measure at the point of contract most people have you know that's that's within the realms of capability of most procurement relatively developed procurement functions as soon as you go past that is very difficult.

00;44;28;27 - 00;44;51;26
Jonny Dunning
It's just more complicated but also it falls down if you're not capturing the detail that happens post contract, But also whose responsibility is it post You know, in a lot of organizations, where does that responsibility lie? You know, is that the procurement team's fault? Is that is that the buyer in some cases it might be the supplier. It falls into that that kind of slightly more slightly grayer area.

00;44;51;29 - 00;45;03;06
Jonny Dunning
How do you kind of see that sort of thing playing out? I feel like it's I feel like it's something that still needs to be resolved in most organizations. It's a progression from where we are now.

00;45;03;09 - 00;45;35;15
Vincent Teyssier
Yeah, you know, we have extremely complex policies and manual in most of the companies enforcing the compliance across these policies is close to impossible. The more data you capture, the more advanced you have, the more you can be compliant, of course. But because you've detected noncompliance, I don't really have a solution yet. I think you're right. This is an area where there's a lot of potential to think through.

00;45;35;15 - 00;46;10;03
Vincent Teyssier
It. But I think there's so much more pressing areas right now in terms of inflation, supply chain disruption that it's probably a bit behind in terms of priorities. The basic compliance is more or less okay. Most of the companies risk based approach starts to be implemented to bring efficiency to the compliance process. Um, so yeah, I mean, I cannot deny that it's important, but is it really the top priority?

00;46;10;04 - 00;46;11;25
Vincent Teyssier
I'm not 100% sure.

00;46;11;28 - 00;46;40;04
Jonny Dunning
Yeah. As I said, I feel like it's an evolution. There are problems that need to be solved prior to that. So. So let's go back a step and just one of the things we were talking about earlier was obviously you've gone from being in-house to now working in on the consultancy side when you're going into organizations now what, what, what usually are the kind of most pressing issues that you're having to look out for those organizations that you're working with?

00;46;40;07 - 00;47;12;14
Vincent Teyssier
I think best practice everyone is looking for best practice. How do you do this process the best? Because some even exist communities. There's a lack of communication. The big four are coming, are typically. Yeah, I wouldn't say not knowing they know, but that depends what partner you get in. Right? Observation of the partner which is focused on digital transformation is probably not as savvy as one was focused on procurement.

00;47;12;16 - 00;47;35;27
Vincent Teyssier
And there you can skip implementation of best practices and you just implement what's common sense and you just implement what's recommended by the vendors, which is not necessarily the best practice. By the way. Ultimately it really boils down to what exposure do you have? Do you talk to a lot of people? Do you see what works? Do you see what doesn't work?

00;47;35;29 - 00;48;03;05
Vincent Teyssier
And I think when I made the transition, I understood one thing very, very quickly in that when you're in the industry, unless you're in top general management, you are in a very, very niche area. You're a specialist and very quickly, when you transition to consultancy, you need to broaden your scope very, very quickly. You need to read the lot.

00;48;03;06 - 00;48;28;24
Vincent Teyssier
You need to look at a lot of videos, podcasts because you're too niche, you're going to have plenty of questions that go beyond. What was your scope previously in the industry? People want not to know what your scope was and how do you handle it, but what your world procurement function was doing there and how this was implemented, how you do disease, how you do that.

00;48;29;01 - 00;48;54;10
Vincent Teyssier
Of course you don't reveal secrets because some things are confidential, but what you can see is, okay, these are the best practices, recommended practice, and you apply it, you visualize, you apply common sense, and it brings that on the table. So, yeah, I'd say the question number one is best practices. The question number two is change management. How do you actually implement these best practices?

00;48;54;10 - 00;49;06;24
Vincent Teyssier
How do you handle a different type of stakeholders, especially the resistant one? I wouldn't say the finance people, but they will recognize themselves and change.

00;49;06;28 - 00;49;09;04
Jonny Dunning
I'm sure they would agree.

00;49;09;07 - 00;49;40;20
Vincent Teyssier
Yeah, Yeah, they will now. I mean, it's a different culture, right? I'm going to go back to to something else to when I was in marketing technology, we had a huge reorganization driven by one of the big four and the very pioneer program we were implementing Agile methodology in 80. So in I.T, they are pilot teams, they have very clear structure, very clear roles and responsibilities.

00;49;40;23 - 00;50;09;01
Vincent Teyssier
They spent one year really crafting a good plan. What are the differences? Coalition Channel 62 Actually, they implemented it. It went like a breeze. People were happy, more productive. The change was very well managed and then they did the same thing in marketing. Marketing is a different kind of people. You know, people are very detailed, very process oriented because when you code, if you don't got the right thing, it doesn't matter.

00;50;09;01 - 00;50;45;04
Vincent Teyssier
The computer will not run in marketing. If you make a mistake on on one word doesn't really matter. It's okay. Right. And it's a more creative world. Yeah. So when they implemented the first Agile teams in marketing, the pilot worked very, very well. And I remember we were on this big room planning. They announced reorg for marketing and I see the structure of no, but okay, you have the teams, you have the mythology, but how do you create alignment between the teams?

00;50;45;06 - 00;51;08;06
Vincent Teyssier
What are the top level channels to create alignments? So I raised my hand and the answer was like, Oh, we have not figured out all the details yet. This will come in the coming weeks. And of course it was a disaster because it's all about alignment. But that's why, because it's a different mindset and that's not being put into account by the people implementing the methodology in it.

00;51;08;08 - 00;51;38;07
Vincent Teyssier
People are very detailed in marketing, they are very creative. Yeah. So of course you don't. That's the same outfit and I think the same in procurement versus finance procurement. There's very different goals and objectives and finance they have much less compliance issues. They have more negotiation issues, which is a more creative activity. So expecting to bring the same framework to finance people than you do in procurement when you implement P2P, for example, that's not going to work.

00;51;38;09 - 00;52;01;11
Vincent Teyssier
You need to adapt your speech once again. It's like the colors we talked about before. You need to adapt your speech. Your finance don't care about ESG. Let's be very clear about it. What they care is working capital. It's completeness of accounting. Beyond that, they want to hear about it. So then you need to have basically a dual business case.

00;52;01;13 - 00;52;29;28
Vincent Teyssier
You need to have a business case which shows an efficiency, which shows increase in working capital through, let's say, supply chain finance onboarding, which shows better visibility on your spend on on the process itself. That means better completeness of accounting that are actual reconciliation. When you build your business like that, the finance is on board, you should build it saying that sense to the system, we're going to generate 20 million savings more.

00;52;30;00 - 00;52;56;23
Vincent Teyssier
They couldn't care less. We're going to achieve our ESG targets. They don't care. That's not touching them directly. So it's really how do you adapt your speech to different type of people? That's a bit high buys in terms of change management. So yeah, Question number two, best practice. Number one. Best practice. Number two, change management. And beyond that, it's the classic how do you implement the system?

00;52;56;23 - 00;53;01;19
Vincent Teyssier
What technology technology selection?

00;53;01;22 - 00;53;40;21
Jonny Dunning
Yeah, I think when you're talking about best practice and you're talking about kind of where the industry is and where most companies are, I think one of the points that comes across in talking to you is what doesn't work needs talking about just as much as what does work. So for example, for example, your example in that scenario with the marketing function and this is where I feel like this sort of conversation with people like yourself, this is this is the type of conversation that needs to happen in the industry as much as possible.

00;53;40;23 - 00;54;01;10
Jonny Dunning
And I think procurement people are actually generally pretty good at talking about the stuff that didn't go right rather than just saying, Oh, I've got all the answers and we did this and we did that, and it was perfect because it pretty much never is. It's too it's too young as a function for it to be perfect it and it's too varied because every organization has different demands, different sets up different culture.

00;54;01;13 - 00;54;16;05
Jonny Dunning
But I think that's quite a powerful thing in terms of what's the most useful conversation in, the market to advance the procurement profession. It's understanding what the options are, what the pitfalls, what other people are finding when they're trying to do just sharing information.

00;54;16;07 - 00;54;49;01
Vincent Teyssier
Yeah, but ultimately, when you want to have this discussion, what went wrong or what is wrong, you need to have the data. Yeah, I there's one topic recently I have with with everyone I talk with. It's the other procurement process from analyzing the demands to implementing the contracts. You have different stages, right? I've not found a single organization in the past weeks which measures the cycle time between each of these steps.

00;54;49;03 - 00;54;54;20
Vincent Teyssier
They have no idea was bottleneck is they have an idea because they know the process, but they use.

00;54;54;22 - 00;54;55;19
Jonny Dunning
An extra.

00;54;55;21 - 00;55;35;29
Vincent Teyssier
Syllable insights telling them exactly their that yo, is it the red lining? Is it is it issuing the RFP? Where's impact on that? And no one knows. I think the the metrics are very well developed for transactional procurement. But when it comes to tactical procurement and strategic procurement, there's very little metrics available in organizations to date and I think a lot of new solutions are actually around that, around bringing you more data about your your arrestees around your category management.

00;55;36;01 - 00;55;41;01
Vincent Teyssier
That's where there's also a lot of potential to improve the process.

00;55;41;03 - 00;56;00;14
Jonny Dunning
Yeah. So another question for you in terms of that switch from being an in-house procurement practitioner to working on the consulting side, how do you find what are the differences in terms of how you were interacting with the different teams within an organization? Is that is that something that's changed at all or is it very much the same?

00;56;00;16 - 00;56;34;01
Vincent Teyssier
Yeah, very different, very different because here I don't of our to say the upper hand and I'm here to offer a service so I need to be customer centric. My priority is doing what the customer needs, doing what's best for the customer that leads to value creation and that means that them mostly in very, very positive interactions. The only negative is me with myself as when I'm pushing myself to do better for them.

00;56;34;03 - 00;57;00;15
Vincent Teyssier
But it's mostly positive interaction. How can I help? How can I support you? Okay, this is a plan. This is how we will unlock value and hopefully this is what we deliver. Right? When you in the industry, this is slightly the opposite. You're measured, you have targets and very often too ambitious targets. So you're going to fail or you're going to have problems here, problem there.

00;57;00;17 - 00;57;24;21
Vincent Teyssier
Talk to any top managers. They will tell you that their job is not really to be a procurement spiritual finance expert or whatever. It's to put down fires. You constantly receive an escalation as you manage escalations. These is mostly negative interactions which are not about unlocking value, about creating value, but about avoiding impacts, avoiding damage and preserving volume.

00;57;24;24 - 00;57;56;12
Vincent Teyssier
So it's a very different type of interaction. Also, you have to handle operations. When you're in consultancy, you typically don't send those operations. But when when you do, that's where all the negative interaction comes because it's never satisfactory. And of course there's always it would be no room for improvement. But I think that's what's changed a lot. I went to very, very, very positive discussions.

00;57;56;14 - 00;58;30;06
Jonny Dunning
Interesting. Okay. So just a kind of a potentially quite a big topic, but let's let's just kind of touch on it fairly briefly and kind of wrap things up around this. I am procurement. So so you and I both done some work in this area. What is it that actually got you interested in it? I've got an inkling because obviously you come from a kind of coding background, but what is it you find interesting about it and how do you think that procurement can most effectively take advantage of the opportunities that are now starting to crop up and be available?

00;58;30;09 - 00;58;58;07
Vincent Teyssier
Yeah. So since I mean programing since long time, I've always followed the kind of progress in I, I implemented a solution 310 years ago. It was a clustering algorithm for core decisions and at that time there was no search engine for code decisions that was very valuable for lawyers. And then different solution came when I moved to move to marketing technology.

00;58;58;07 - 00;59;17;26
Vincent Teyssier
I remember there was a vendor which had the solution to send smarter emails to to your prospects. And when I asked what was under the hood, because I'm very interested in the technology, it was extremely basic. It's an algorithm from 20 years ago, really. That's what they sell.

00;59;18;01 - 00;59;30;14
Jonny Dunning
And it's not really AI, is it? And I think that that's where it must have been. It must have been an interesting conversation where you were kind of asking the pointed questions, must have been quite awkward, the people on the other end.

00;59;30;16 - 00;59;53;06
Vincent Teyssier
So it needed them. No, they have no idea. When you ask too much details of product managers, they them. But forget it. I think there's a lot of hype. There's a lot of diligence that that that's a typical curve. I say go from gov right. The reason this disillusionment is that people don't truly understand what it is about.

00;59;53;08 - 01;00;33;22
Vincent Teyssier
Even today, people don't understand why they think the tragic defense reason it doesn't. What's LGBT is it's a completion prompt. It basically texts your question as a sentence and predicts what is the next word to put as an answer. That's what he does. Nothing more is not statistical. It's smarter than that. But that's what it does. And when you do prompt engineering, you keep that in mind because ultimately you want to write a question in the form of a text that feels natural for LGBT to then compliment after your prompts.

01;00;33;24 - 01;00;48;23
Jonny Dunning
Do you think do you think that a lot of people I feel like this when I've when I've seen people playing around with like the Chat geek playground, I feel like a lot of people almost try to talk to it in the same way that they talk to Google the moment they.

01;00;48;25 - 01;01;17;09
Vincent Teyssier
Yeah, that's what they try. And the limitation in the case over the console was September 21 after that, it doesn't know anything anymore. So right now there's a lot of solutions, including solutions that I'm developing myself, which are about augmenting subject with actual data from today wizards data, which is internal, external or a mix of it. But without trying to pitch my project, I'm going to talk about being checked.

01;01;17;11 - 01;01;46;26
Vincent Teyssier
When you go to Bing chat and you take a question, what does it do? It doesn't query Chargeability first. It's first doing a search through Bing. The top three articles are returned. This article then said as a context in a prompt to charge it and change it, you will basically use this concept context to write the answer and that's it.

01;01;46;28 - 01;02;01;11
Vincent Teyssier
That's how they augment it with actual data from today. But then when you think like that, the use cases are enormous because that means that it's all about the context that you feed inside the prompt behind it.

01;02;01;13 - 01;02;02;29
Jonny Dunning
Yeah, I totally agree.

01;02;03;02 - 01;02;31;06
Vincent Teyssier
And the possibilities are endless. In this context, you can retrieve data from a database, you can retrieve internal documentation. What we see today also, the more the models advance and compute becomes cheap, the more the context window can be big. So and until last week could put 4000 tokens as a context suggested, it is 3.5. Today there is the new version.

01;02;31;06 - 01;03;02;16
Vincent Teyssier
You can go up to 8000. That means instead of putting 40 pages of documents, you can put 80 pages of documents as a context and such a pretty basic sensor on this amount of facts. And that is going to really augment the capability to to drive information, relevance of information given by tangibility, just if it is good at two things, one, which is completions and what we talked about, if you enrich it with context, you have a very good use case.

01;03;02;19 - 01;03;28;19
Vincent Teyssier
The second one is summarization, yeah, when you give a lot of information and especially with a context windows going bigger and bigger, you can give more and more information and summarize it in the more efficient way where it can be supply and procurement reporting. You can get a lot of feedback, which is conversational from many stakeholders and summarize it in just one click.

01;03;28;22 - 01;03;57;28
Vincent Teyssier
And we know how reporting is a big problem In procurement. Typically you have one resource, a team dedicated to reporting so that this one use case planning. When you're doing your supply chain planning, it's Also a lot of data that you need to crunch and summarize with a combination of rule rules and prompt engineering. You can make the planning process more efficient.

01;03;58;00 - 01;04;25;06
Vincent Teyssier
I remember when GitHub Copilot was released, do you think it was a year or a year in the house to go Very quickly? People started to imagine what are the use cases with? The programmers become useless, right? Two days have not lost their jobs. They're just more efficient. Instead you have a transition, a transition from doing the job to reviewing the job today.

01;04;25;06 - 01;04;47;19
Vincent Teyssier
If you're a programmer, you're going to get a piece of code from copilot or from serendipity, paste it, review it to make sure that it didn't. Are listening in procurements, you can do the same. You can fill it with information, ask it to do a kind of market research, and then you just need to reduce the output to make sure this is proper.

01;04;47;21 - 01;05;08;28
Vincent Teyssier
So we're going to transform our work into a lot of review. I'd say that basically is a balance between and review is going to be reversed, but that doesn't mean that we're going to be replaced or whatever. We're going to be more efficient and we're going to be the judges of the quality of the work done by the machine.

01;05;09;00 - 01;05;32;14
Jonny Dunning
Yeah, yeah, absolutely. Elevates the the cognitive process required. It's a more strategic approach rather than doing the kind of basic work. So one of the things you talked about was the fact of the type of data the these large language models are actually using. So for example, Bing Chat is using up to date information, but it's using that specific set of information.

01;05;32;22 - 01;06;08;17
Jonny Dunning
You can obviously provide a lot more data on. I know the tool that you have is looking at internal documents and data, isn't it? One of the things I find quite interesting about the the kind of changes in how open AI and chat GPT work is, the fact that they now have this relationship with Microsoft in terms of like data security is quite interesting in terms of the policies about they separate themselves in the sense that, you know, the data that's provided is walled off from being used to develop the model effectively so it can used to provide it can be used as context for activity to provide an answer.

01;06;08;17 - 01;06;17;29
Jonny Dunning
so taking that into context, looking at the data that GPT is using is quite interesting.

01;06;17;29 - 01;06;34;24
Jonny Dunning
So for example, Big Chap is using up to date data that the search engine providing, it's doing an analysis on that and doing a very effective summarization of the information that's being provided to you. You don't have to search through all the different pages. It's just giving you the answer that you're looking for. So that's really interesting use case.

01;06;34;27 - 01;07;15;04
Jonny Dunning
What you also talked about was the fact of it being able to use external data and and much larger amounts of external data. Now, as I understand it, what you're doing with your product help is actually looking at internal data and documentation. And one of the things I just wanted to briefly touch on was the increase in data security that has happened with with open AI and chat since the partnership with Microsoft, where it's effectively, as I understand it now, you're really at a point where you could, you know, if people are worried about trusting, using chat, then they should be worried about using outlook and teams and all these other things that are all

01;07;15;04 - 01;07;35;17
Jonny Dunning
part of the same Microsoft set up. And I think that's quite game changer in the way that that's been moved. I mean, ultimately, as I understand it, the information that's provided is not going to be used to train the model. It's just used for that specific response. Do you have any particular opinions on that? I think it's a bit of a game changer.

01;07;35;19 - 01;08;02;05
Vincent Teyssier
Yeah. I think people have been very quick at trying justify their fear of adopting a quickly a technology by the data security issue. And I was actually at the opening. I tour in Singapore where some man was there and was repeating that they do not use the data for fine tuning, they do not use it for retraining the model not at all.

01;08;02;05 - 01;08;37;24
Vincent Teyssier
And he said we were actually pretty bad at explaining that, actually. But then beyond that, you know, Microsoft, which is capitalizing on their investment. So they opened the Azure Open Air Service in their Azure Cloud services and they give even additional guarantees. Now, the models are deployed within Azure ecosystem and infrastructure. They have guarantees by contract that there is no use for retraining, no use for fine tuning, that the only use of the data is for debugging process and you can actually ask to opt out from it.

01;08;37;26 - 01;09;06;11
Vincent Teyssier
So they give a set of get up to me. Good, good luck. Like as you say. Otherwise you would stop using outlook, you would stop using the analytics from office three, six, five, etc. And however, even today you have some industries which are extremely regulated banking, gambling exit, racism and these companies typically have their own data center. They do not have a cloud policy.

01;09;06;11 - 01;09;37;24
Vincent Teyssier
They want to have everything on premise For the use case of internal data, what you use is only the ability to rephrase the context. Okay? That means that you don't need for that would be you need 64 just to the completion with a very strong context you don't need very advanced model. You can actually do with 82 and these models can be deployed on premise.

01;09;37;26 - 01;10;02;16
Vincent Teyssier
You have much more advanced activity to that. For example, Dolly from Databricks, you have plenty of models now with commercial licenses that can be deployed on premise is something the fear is more because people didn't do their homework, didn't have the time or or whatever. But but I think the data security and data privacy aspect is fully covered today.

01;10;02;18 - 01;10;22;23
Jonny Dunning
Yeah, it's really interesting to hear you explain that. You explain that very well. Our CTOs are very good at explaining it. He's a technical person. I'm not quite so good at explaining it, but but I definitely think this is one of the first questions that last large organizations ask with you're showing them something you've done, like the scoping tool that we've done, we built using, using chat.

01;10;22;23 - 01;10;42;02
Jonny Dunning
GB Tay So let's, let's wrap things up there. I really enjoyed that conversation. It's great to get your views on this stuff and to hear also the way that you apply it to real experiences. I think that's that's really useful for me and hopefully useful for other people as well. Do you have any kind of what are your what are your kind of predictions?

01;10;42;04 - 01;10;58;29
Jonny Dunning
Obviously, we're in it. We're in an interesting time where there's lots of supply chain disruption, inflationary pressures. What are your predictions for how the kind of big challenges that procurement are going to have to deal with over the next 12 to 18 months?

01;10;59;01 - 01;11;21;14
Vincent Teyssier
Well, who would have predicted COVID? Yeah, we would have predicted that the supply chain disruption and the war in Ukraine. No. One. Right. And and, you know, people like prediction, people like to have predictive models for their supply chain. And they're saying, but as soon as a real disruption is coming that no one's coming, then these models, that's totally useless.

01;11;21;16 - 01;11;51;05
Vincent Teyssier
So I think it's not about predictions, it's about capabilities. If you have the capabilities to give you the data when you need it, if you have a flexible process, if you have the right structure, you can be very agile in handling disruptions. Ultimately, what you're trying to increase in any organization is the velocity. The velocity at which you deliver innovation, at which you get insights for everything.

01;11;51;08 - 01;12;06;21
Vincent Teyssier
So the more data you get, the more you have built system, which allows you to answer any question. This is what's really going to make you futureproof to any disruption. And I'm not in the record. I'm not going to predict anything,

01;12;06;21 - 01;12;28;27
Jonny Dunning
But just to go back to what you were saying, so what you're effectively saying is the most important thing that procurement organizations can do right now is build the capability to to be able to answer any questions that might come up in the future and to be able to deal with the unpredictable nature of the market conditions that we deal with.

01;12;28;27 - 01;12;33;27
Jonny Dunning
Now and and ultimately to have flexibility.

01;12;33;29 - 01;13;09;15
Vincent Teyssier
Yeah. So I'm going to give an example. When the inflation started to kick in, a lot of organizations were unable to track this inflation. I've seen one organization coming out with a report of 80 pages category by category, with the curve showing the impact of inflation and adapting their category strategy based on this findings. And this was done very, very quickly, the ability to output this kind of reports is really going to give you an edge in adapting your strategy.

01;13;09;18 - 01;13;15;24
Vincent Teyssier
So the faster you are at giving insights to management, basically the faster you can pivot in your strategy.

01;13;15;26 - 01;13;34;29
Jonny Dunning
Excellent stuff. When I said thank you so much for taking the time to champions, I really enjoyed it. I think there's a lot more we could talk about and maybe we'll have to have a follow up session at some time, but I really appreciate you taking the time, some of your thoughts across, giving some of the benefit of some of the experiences that you had sharing that information.

01;13;35;01 - 01;13;39;04
Jonny Dunning
It's really useful. I very much appreciated. So thanks very much.

01;13;39;07 - 01;13;41;10
Vincent Teyssier
Me a revision for me as well.

01;13;41;13 - 01;13;42;19
Jonny Dunning
Excellent.

LATEST

Taking procurement value beyond supply chain savings

When it comes to SoW, who should be involved and how does procurement ensure everyone is adding value?

Taking procurement value beyond supply chain savings

FEATURED

Using the right mix of technology to deliver an effective extended workforce strategy

Software Strategies - Decoding Value in the Extended Workforce

When it comes to SoW, who should be involved and how does procurement ensure everyone is adding value?

The challenges of delivering value under SoW engagements