Episode 30 - Procurement & AI - Interview with Martin Rand from Pactum - Pt.1 - Procurement Zen

Episode 30 – Procurement & AI – Interview with Martin Rand from Pactum – Pt.1

In today’s episode, I interviewe Martin Rand; CEO and co-founder of Pactum (www.pactum.com). Martin not only has vast experience in negotiations, but he also shows how Machine learning and AI can help in negotiations.

We go over the hot topic “digitalization in procurement”. Martin, being the CEO of a company that offers AI solutions in negotiations, has a lot to share in this regard.

Phil Kowalski:
Digitalization is huge in procurement these days, but do you know any real projects that got results? Well, you're in lucky place, because today, we have a very interesting interview. My guest does not only have experience in negotiations. He'll also shows us how AI and machine learning can improve your bottom line. Without further ado, let's dive right in.

Are you looking to up your negotiation and procurement skills? You're in the right place. Welcome to ProcurementZen with your host, Phil Kowalski.

Phil Kowalski:
Okay, everybody. This is Phil Kowalski from ProcurementZen again. As I told you in the intro, we have, today, a very interesting guest. We will talk a lot about digitalization topics, digitalization and procurement, which seems to be the buzz. We will get in very much details today. Please welcome, everybody, to the show, Martin Rand, the CEO of Pactum. Martin, welcome to the show.
Martin Rand:
Hi. Hi, Phil. It's good to be here. Thank you for inviting me. I also wanted to say thank you to you for putting so much effort and energy in getting the art and science of negotiation to the next level.
Phil Kowalski:
Thank you very much for the kind words. Martin, one of the very first questions that we usually always ask our interview partners is, can you tell us a little bit more about yourself and how you entered the world of "negotiation" so to say?
Martin Rand:
Yes. Certainly. I use to be a product manager at Skype. I was more on the technology side of things. I then set up my own startup, which was VitalFields. It was a farm management platform for farmers. We scaled that to seven countries including Germany. I then sold it to the Climate Corporation. Of course, selling a startup is the biggest negotiation one can have in his life, I think. That's one experience of cost, but then in Climate Corporation, I became the commercial lead for Europe. My task was to negotiate commercial deals with food in that companies in Europe, because Climate Corporation provided digital tools for farmers and my task was to negotiate these deals.
Martin Rand:
Basically, I lived from an airplane for two years. I could begin my week negotiating with the Ukrainians on Monday and then with the Spanish on Wednesday and then with the French on Friday. I think I got to see all of the major cultural differences in negotiations in Europe. From that, the idea was born.
Phil Kowalski:
That's very interesting. You're talking about the idea. Can you tell us a little bit about the idea? I know we're talking about Pactum, but can you give our listeners an overview what Pactum, your company, basically does?
Martin Rand:
Yes. We have three co-founders. Kaspar was the founding member of e-Residency program in Estonia, which is a well-known e-Government program. Kristjan used to be the AI lead in Starship. He has a PhD in machine learning and artificial intelligence. We decided to build Pactum. The whole idea is that we will use AI to negotiate commercial deals. It sounds crazy in the beginning. It sounded crazy in the beginning to a lot of people, but then once you get into it, you really start understanding that there is a place for AI in negotiations. We focus on the long tail commercial deals. We focus on the tail spend. In the tail spend contracts, what happens always, I would say in those large companies, is that there is not enough human power to manage those deals. That's why they're called tail spend. We do that with a computer and renegotiate those deals on a massive scale.
Phil Kowalski:
That's quite interesting. That absolutely matches my experience. Usually, procurement and negotiation capacity, whatever you call it, sourcing, purchasing procurement is oftentimes limited. There's usually never enough capacity, never enough people. You usually focus on the big projects and then yeah, all the tail spend just runs through. However, my experiences, also, it sums up over time. It can have a significant impact. Really interesting because one of my ... maybe I can call it a mantra. One of my mantras usually is that we will not reach a level where all negotiation is automated. Only the machines are negotiating with each other and procurement is no longer needed.
Phil Kowalski:
Usually, my reasoning for that is, is that face-to-face negotiations, no matter if they take virtual place in these days or really, in meeting rooms, are often driven by psychology. Things like what personality type is my counterpart? What experience does he have? What market are we talking about? Like supply market, buyer's market. What is your point of view?
Martin Rand:
I would add to that, that this is exactly our view as well. We say that we can automate up to 80% of these negotiations and according to the Pareto rule, these 80% of long tail supplier deals have 20% of supplier cost in them. Human negotiated deals will always remain simply because for those large and really strategic deals, human ingenuity is needed because an AI cannot do out of the box solutions. No matter how smart it is, it will, to a degree, always think inside some limits, but human ingenuity can provide absolutely other solutions for problems. That's for those large deals, I think, humans would be needed.
Phil Kowalski:
I like this approach very much. I also like that you say it because in one of the last shows, we talked about digitalization and procurement with another interview partner. I definitely would underline that. He said that every digital project, every digital functionality, so to say, should help buyers focus on their core tasks, ad that from my understanding is what your AI approach also does because yeah, maybe I have a super fancy dashboard, but how does it help me to improve my results to get better savings compared to what you just said where you say, "Okay, this automates, frees up time and frees up capacity and at the same time also delivers results." I like that, really, a lot.
Martin Rand:
Yeah. Exactly. It does the actual negotiations. It does, not only free up time, but sometimes it delivers extra results that humans were never capable of achieving. Quite often, in large enterprises, there are tens and tens of thousands of suppliers. Most of them are basically unmanaged. That means that we don't free up time because those were unmanaged anyway. We renegotiate those deals and release a lot of value for both sides that were locked into these agreements simply because they were unmanaged. If somebody would have negotiated them, they could have found a better solution for both sides, and that's what we tried to find.
Phil Kowalski:
Interesting. You mentioned before, I just wrote that down because I found it so interesting, you talked about these Pareto efficient deals. Can you elaborate on this a little bit more?
Martin Rand:
Yes. The Pareto efficiency theory is used a lot in negotiation science. It's a bit hard to explain with audio, but if you imagine a graph where the X axis is your value in any contract and the Y axis is your counterparts value. If you imagine a dot on this, if you add up all of the values in a contract, you end up with a dot on this graph. Now, there is a line also on this graph which is called the Pareto Optimal Line, which is usually further than those dots. Every negotiation in the world should be on that Pareto Optimal Line. This means, if they are on this line, it means that one party cannot get a better deal without hurting the other party. Because there are so many of these unmanaged deals, then usually, they are not Pareto Optimal. They are in a scatter all over this graph.
Martin Rand:
Our task is to push them towards this Pareto Optimal Line, and this will usually create value for both sides.
Phil Kowalski:
That's quite interesting because you always get what you pay for, right? If you hurt someone extremely to get some short term gains, you usually have to pay for it in the long run. That's at least my experience of close to 20 years in negotiations and procurement because there's always payback time. It works both ways if a supplier or a vendor tries to optimize him or herself on your cost, usually, there comes the opportunity to pay back.
Martin Rand:
Yes.
Phil Kowalski:
That's why it's interesting. That sounds a lot like win-win, right?
Martin Rand:
Yes. Being on a Pareto Optimal Line doesn't have to be win-win. It can be, totally, one sided, but the deal is still Pareto Optimal. What we always show on this graph, and it's a bit hard to explain, it's the fact that the shortest line to that Pareto Optimal curve is through win-win by creating value for both sides. The quickest and the simplest way to create value for our customer is to create value for the other side as well.
Phil Kowalski:
That's quite interesting. I like a lot that you also talk about the negotiation theories behind your optimization tool or platform. Because I think we all can learn a lot from this and have recently done some book reviews, the classic ones about negotiations, and of course, these concepts are mentioned there as well. What would interest me and maybe the audience as well is, I read about Pactum and how it came to life, so to say, in your medium post. You wrote a very interesting post on medium.com. To the audience, I will link that in the show notes. How did you come as an experienced negotiator? How did you come to the conclusion that there is optimization potential through AI? Did you just wake up one morning and you knew it or how did this process evolve with you and your co-founders?
Martin Rand:
Well, honestly, what I was missing, I was missing a platform where negotiations could be a trial with people from different cultures. I thought that this would be my next startup and this is what I would build, but then it evolved when I met with Kaspar and Kristjan, then it evolved into actually creating a machine that can negotiate. That's because we discovered that, "Hey, the pain is not so much in the ability for people to negotiate. The much bigger pain for the world economy is that most deals are not negotiated at all." That means that they are suboptimal. Essentially, if we able to do a machine that can make those deals optimal, we create value to the world out of thin air, basically. We make existing contracts more efficient.
Phil Kowalski:
Efficiency is something, I think, everybody in the procurement space is pretty much interested in. We talked a lot, now, about the machine and how it negotiates. Can you give us an insight how a typical negotiation with the machine from your side and maybe also from the suppliers side how it looks like, how is it executed?
Martin Rand:
Firstly, there is no typical negotiation. We work with only the largest companies in the world. Our largest customer is Walmart, which is also the largest company in the world by revenue. We focus on Fortune Global 2000 Enterprises because they have the sufficient long tail in order to have this problem. Basically, our solution is only a solution when the company has scale. Firstly, how we set up the system is that we go visit the company physically with several of our analysts. We spend time with them by understanding their strategies, by understanding what they need to achieve with negotiations, what are their positive and negative levers, how do they value these. From that information, we build what we call the value function. The value function is our interdependencies between all those positive and negative levers. We design the negotiation flow. The negotiation flow is currently in a chat format. We do that with our negotiation scientist.
Martin Rand:
We have the best negotiation scientists in the world as advisors. Our lead scientist and advisor is Jeanne Brett. She is the professor of Emeritus from Kellogg School of Management. She focuses on international negotiation. We jointly build the strategies and tactics of the negotiation specifically for that enterprise. That negotiation flow then takes information from this value function. It's able to negotiate deals in a fully autonomous way. Once it is turned on in a certain category, it will reach out to the vendors by email. It will ask them to click on a link. They click on a link. They see a chat and their face where a chat talks to them as a human asking, how are you? They start talking to the bot. They get into it, understand how it's working and then the bot knows everything about them. Because we tap into all the data resources that are possible. We say, "Hey, you asked about the methods."
Martin Rand:
Every negotiation is, of course, different and different strategies and tactics are used there. What is quite typical is that, firstly, of course, we will find out as much as possible about the other side. We will try to understand what negotiation items or these positive levers they value and then depending on the negotiation, we choose tactics, which we use. To bring an example, one tactic, for instance, that can be used that yields joint value very well is multiple equivalent simultaneous offers. The point of that is that we will offer two options, Option A and Option B. Both of these options have different items within them. Both of these options are better for us than the status quo. If the other side is willing to go ahead with one of them, we'll choose one of them and then value has been created on the other side as well. This is how we move bit by bit towards that Pareto Optimal state.
Phil Kowalski:
That sounds very interesting. You read a lot about digitalization and hear a lot about digitalization these days and the past years, especially in the procurement space, but it's really interesting to hear how it's executed if it does not only live, let's say, on a power point level, so to say, but how it lives in real life. Two questions to the example that you mentioned. The first one, as you said, the system itself sends out an email and invites the vendor to a chat or chat bot. First question, what is your experience? Do buyers from the respective procurement department open up a parallel stream, so to say, where they maybe inform potential vendors about this new approach? The second question is, you said, prior to the negotiation, the system connects to all kinds of data resources. I wonder what these data resources is. Is that the only system data like spend quantity description and so on or does it also favor in like yeah, that is maybe a very dominant partner like the domain expertise? Or is that not necessarily due to the fact that it's tail spend? That's two and a half questions, actually.
Martin Rand:
Firstly, about the buyers, if the buyers have to reach out, no, they don't have to. The bot is smart enough to be able to speak to the vendors. By the way, in our projects, 74 to 82% of the people who speak to the bot prefer the bot over a human. Next time, they would like to speak to the bot as well. This is somewhat counter intuitive because today, we see some of those customer support bots out there and people don't like them. They're not very smart, but interestingly, this effect is reversed. With our bot, most of the people prefer it over a human. I think what plays in our favor here is that negotiations is a high stress situation. People inherently try to avoid these situations. Also, if they are a small, long tail supplier, then usually, those large enterprises, they don't have time on them to find this Pareto Optimal outcome. They will just impose themselves on these small suppliers.
Martin Rand:
However, the bot doesn't do that. The bot knows all the possible benefits that can give to the other side and to trade them against benefits for our customer. It spends time. It has unlimited time to negotiate, unlike a human, to find that perfect outcome that is perfect for both sides. I forgot the other question. Sorry.
Phil Kowalski:
Yeah. The second question was about the data resources that the bots or the system connects to, and if it also favors in human interactions or if that is necessary at all.
Martin Rand:
Human interaction is not necessary. We have built ... but it's possible. We have built the bot in a way that it can negotiate the deals on its own, but if it cannot reach an outcome and if our customer has enough capacity to bring humans into the negotiations, then we can do what we call human in the loop. If it's about struggling with something, we can look, then the human and the human can resolve it or if the bot couldn't find the solution, but we see that the solution is out there, then we can bring in a human. Most of our customers prefer to leave it fully on the bot because they simply don't have the manpower to be invested. In terms of data sources, we connect the system to every possible data source that we can. This means an ERP system, a CRM system, a contract management system and even external sources if they play a role. For instance, in retail, even the weather forecasting plays a role sometimes. The more information we have and the more positive and negative levers or negotiated items we have, the better outcome we can achieve.
Phil Kowalski:
That's interesting. What I also like very much is that you said it takes into consideration all the benefits it can provide to a potential vendor or to the vendors. I think that's crucial, it trades them and because from my experience, sometimes, maybe more often than not, maybe I should not say that, but sometimes you have bad days, right? You're lazy and you say, "Yeah, I should've asked maybe for a higher payment and I should've asked for a better price, but you know, I'm fine with what I have. Let's go." Having a machine in here that says, "I have two more things, but I want something in return." Maybe it's not that easy as I describe it, but that's how I understood it. That's also very interesting because it follows this approach for like, right?
Martin Rand:
Yes. The problem, so far, has been that people don't have time for that. It takes time to understand what the other side values and to understand if we have those items that can be traded. The other problem with people is that it's hard for people to think in a multidimensional contract space. If there are 30 items to negotiate about, then think of the permutations or all of the possible combinations of those 30 items. It's a lot. People aren't built with their brains. People usually think priorities are one, two, three. For the rest, they have to create the new one, two, three, but the system can take into consideration every possible combination that it can work with and that it sees that will work best for this deal. Of course, the value of humans in negotiation is that humans have human creativity, which a machine never has-
Phil Kowalski:
Right.
Martin Rand:
... and other psychological traits. Although a lot of psychological traits can be also replicated with the machine, but the benefit that the machine has is that it never has a bad day. It never has an ego that comes into play. It just abuses their strength out of laziness. It doesn't have cultural differences. It doesn't have cognitive biases. All of those things that impair negotiation, the computer doesn't have. Of course, the list goes on. When you think the advantages of a system negotiating, then a system, firstly, learns about every negotiation. Every negotiation that we do makes every other negotiation better because we learn all the time. The system learns. The system learns about people, how people negotiate them become better.
Martin Rand:
The other advantage is that the system can do 1,000 negotiations parallelly at the same time. It can understand what's going on and all of those parallel negotiations and use this data in a live environment inside negotiations. We can do dynamic pricing, for instance. Dynamic pricing use the dynamic pricing principle on any negotiation item. If there is a scarce resource of some kind that is divided between many suppliers. That scarcity can be used and divided in live negotiations and thus, a competitive environment is created. It's somewhat like a bidding platform, but a bidding platform has one very big drawback. It mainly focuses on the price, but never is always in a commercial setting. The value of the deal comes from all items, not just price. What the system can do is create the same competitive situation as a bidding platform, but focus on all of these important items simultaneously.
Phil Kowalski:
That's a disruptive approach to dealing with negotiations at scale, huh? Next week, you'll hear the second part. Martin shares more insight. He has a very special offer to you, procurements and listeners as well. Make sure to subscribe and review to not miss out on this offer. Until next time and always happy negotiations, yours truly, Phil. Bye.

Thanks so much for listening to this episode of ProcurementZen with Phil Kowalski. For more great content and to stay up to date, visit procurementzen.com. If you enjoyed today's episode, please review and subscribe and we'll catch you next time on ProcurementZen.

Episode Highlights

  • Phil introduces Martin Rand
  • Martin shares with us his experience in the field of negotiation
  • Martin tells us about his company “Pactum” and what it does
  • Understanding what a pareto efficient deal is, why is it important to be on the pareto optimal line?
  • How was the idea of AI deal optimization born?
  • What is the process of AI negotiation and how does it work?

Key Points

  1. Human negotiated deals will always remain because human ingenuity is needed in large and strategic deals.
  2. AI negotiating is mostly useful for large enterprises where many deals are un-managed.
  3. Every negotiation should aim for a Pareto optimal outcome.

Resources Mentioned

Phil Kowalski

I am passionate about helping others to achieve more in their negotiations. I love to prep, design and execute successful negotiations. With more than 15 yrs procurement experience I sometimes feel like the Obiwan Kenobi behind this blog and my podcast.

>