Martin Rand: 0:03

They cannot do short term contracts today simply because they don't have enough procurement people, but a machine can. So now when the containers are still on the sea, our, bot will contact the truckers and will agree, the move, one week or two weeks ahead. So now the trucker has some time to plan capacity, and the volume is already there and they know that Maersk is not doing this in, fire fighting mode at the last moment. And this all reflects, in prices

Tom Raftery: 0:34

Good morning, good afternoon, or good evening, wherever you are in the world. This is the Digital Supply Chain podcast. The number one podcast, focusing on the digitization of supply chain. And I'm your host global vice-president at SAP Tom Raftery. Hi everyone. Welcome to the digital supply chain podcast. My name is Tom Raftery with SAP and with me on the show today, I have my special guest Martin. Martin, welcome to the podcast. Would you like to introduce yourself?

Martin Rand: 1:01

Good to be here, Tom. Thanks for inviting me. Yeah, so I am, my name is Martin Rand. I am the, uh, CEO and co-founder of Pactum, which is an autonomous negotiations commpany.

Tom Raftery: 1:14

Wow that's that's a lot right there. Autonomous negotiations company. I mean, we've, we've heard of autonomous cars, quite a lot. And, uh, autonomous robots that kind of roam around by themselves in warehouses and stuff. Autonomous negotiations. You're gonna have to explain that one, Martin.

Martin Rand: 1:34

Yeah. So an analogy of an autonomous car is, uh, quite apt. So basically, every car is automated, but, a car can also be autonomous. And the difference is that an autonomous car will figure out how to get you from a to b.

Tom Raftery: 1:51

Yep.

Martin Rand: 1:52

And so that the same with, our negotiations. There could be an automatic negotiation, but our negotiation is autonomous, which means it'll based on data, it'll figure out who to contact, how to negotiate, it'll conduct the negotiations. It'll generate the new agreement, get it signed and upload the data in corporate systems. So it's basically, a system on autopilot.

Tom Raftery: 2:20

Okay. And what kind of negotiations are we talking about?

Martin Rand: 2:26

So today, negotiations are mainly done by people, but the business environment is becoming ever more complicated. So in most deals today, actually there is a component where there is a lot of data. Or the data is very complex or the data is very fast moving or there are just tens of thousands of suppliers, which people cannot handle. And so in all of those use cases, a machine, does a better job because a machine can think in a multidimensional space and take all of the internal and external data into consideration and also can run through a negotiation simulation 100,000 times to practice before the real negotiation, which is not possible with people of course.

Tom Raftery: 3:23

Wow, but I mean, is it just negotiations for purchasing from suppliers? Is it negotiations for, sale of goods? Is it negotiation for what are the use cases we're talking about?

Martin Rand: 3:39

Yeah. So we have, three main use cases right now. The first is, logistics. We're negotiating with, truckers. so it's basically we can negotiate up to an individual container move. And of course there are a lot of these moves, so a machine can handle the scale well. So this is the first use case. The second main use case is, um, working capital negotiations. So this is reaching out to thousands of suppliers, renegotiating price and payment terms and other, uh, negotiable terms that are, that are relevant there. And this use case is horizontal. Meaning it's applicable for any enterprise, of the world. And the third use case is merchandising. So this is negotiating the items that are sold in stores and the benefit of a machine is that it can handle a lot of scale, which means it can go to item level and negotiate SKU by SKU. And, and of course there, within that lie, the efficiencies and the efficiency gains.

Tom Raftery: 4:48

Okay. So how does. The recipient of this deal with it. I mean, if I am negotiating with your autonomous negotiator, am I aware that I'm negotiating with a, for all intent purposes bot or do I think I'm negotiating with a person at the other end? Or how does that work?

Martin Rand: 5:13

Uh, absolutely, yeah, you know, that it is a bot. In fact, in some states in the US, it is, illegal not to state that it's a bot, if it's a bot. So though everyone, everyone knows that they are talking to a bot, But, interestingly, sometimes, the bot can be more human than a human, because if we look at very large enterprises which have tens and tens of thousands of vendors, then at the end of the day, people can focus on strategic deals mainly. And the, the tail, the smaller and medium vendors, those deals are left unmanaged. And so the bot is now reaching out to these vendors. It knows everything about their business, and it is brainstorming with a vendor on how to increase their business. And it is finding out and is interested about the other side. What do they value and is then asking what the bot would like and would then do the trades that create value, for both sides.

Tom Raftery: 6:19

Okay. Isn't there some kind of, I don't know, trust factor that's required to allow a bot to do negotiations on behalf of your company. Isn't that, is that a challenge you're getting from customers?

Martin Rand: 6:33

So we haven't had that issue because our customers are the largest enterprises in the world, such as, uh, Walmart and Myers. So we represent them. The vendors don't see a lot of Pactum in this equation. They see that Walmart sent them an email and is inviting them to a discussion which is happening online. They click on a link, a chat will open and they will go through a negotiate the negotiation, in a chat interface. So this is all Walmart and, and this it's evident from the chat that it knows all the relevant details about, their deal. So we were afraid that trust issues would become, uh, an issue. But, uh, no, in reality, we haven't seen that a lot.

Tom Raftery: 7:19

So I was actually thinking it from the other perspective, the, the challenge of getting the likes of the Walmarts and the Myers

Martin Rand: 7:27

Oh, right, right.

Tom Raftery: 7:28

to allow a bot to do negotiations on their behalf.

Martin Rand: 7:31

No, no, not at all. So that side, that side is good. We're working with the largest enterprises in the world, and we're getting more and more of them on board all the time. And now as, the world is entering into so much, change and, and turmoil sometimes, our order books are becoming, fuller and fuller. And they trust us partly because it's such a clear value proposition. Every procurement, professional knows that there is a lot of value left on the table, uh, within non strategic spend. And and then you have a couple of, alternatives. So you could use E auctions for instance, or you could use negotiations. However, negotiations create much more value, than any kind of bargaining or auctions, because the premise of negotiations is let's brainstorm all the things we can talk about, bring all the negotiable terms on the table and then start trading them. So every deal is custom and the vendor always gets something that they value, which is not happening so much in an E auctions use case, which is more of a zero sum let's divide, the pie, kind of game. So, so yeah, trust the simply the, the need is so large that trust hasn't been an issue. Oh. And so many people ask if people need to, kind of keep an eye on the negotiations, if the bot, how, how do they trust? What is the outcome of the negotiations? Well, the bot works on what we call the value function. It is a, a mathematical model that represents every possible state of this negotiation. And once this value function is agreed with our customer, there is really no need to oversee it because it's a machine. It won't go outside the boundaries. The boundaries have to be agreed, correctly, and, and sometimes our customers, they track the outcomes of negotiations in the beginning, but they get tired pretty quickly because they see that all of these are within bounds.

Tom Raftery: 9:42

Okay, fair enough. Fair enough. And why, why did you decide to tackle the negotiation space?

Martin Rand: 9:51

Yeah. So, I'll maybe speak about a bit about my background. I used to be a product manager at Skype. Uh, I then set up my first, uh, startup, scaled it internationally, and ended up selling it to Monsanto and I became commercial lead for Monsanto's, climate corporation. Basically part of my task was to, uh, negotiate in Europe and negotiating in Europe is, is a lot of fun as you might know, uh, because you know, you might begin the week by negotiating with the Romanians, end the week by negotiating with the French, see a new culture, a new way of working, uh, new cognitive biases and new cultural differences every day. At the end of the day, I just, I saw the money left on the table. I saw that not, not all negotiations are optimal, for people. And what I also saw is that, oh, most of the negotiation in negotiations in a corporate environment are not being had at all. So 80, 80% of vendors are unmanaged. And so, so that really pushed me, towards negotiations. I also went through, Harvard's, uh, advanced negotiation studies and so all our negotiations are based on, on Harvard's program, on negotiations principles.

Tom Raftery: 11:13

And do you use Pactum yourself to, uh, negotiate with your own suppliers? Or customers?

Martin Rand: 11:23

No, because we don't have enough scale for that. So we work with the large enterprises. We, we have just a handful of customers, but in order for our bot to become efficient, we need hundreds, preferably thousands, uh, of, of deals.

Tom Raftery: 11:40

sure, sure. A bit of a facetious question. Um, so. when you're getting into the hundreds and thousands of negotiations, as you say, you've kind of got an 80 20 rule there. This means that you are making the negotiation far more efficient for organizations because the vast majority of those negotiations are then handled by the bot and people can focus exclusively on the, the 20% and put more effort into those. That's basically the premise, right?

Martin Rand: 12:09

That's correct. Although our main value proposition is basically value. It's. getting a better deal unlocking, uh, working capital and getting a better price, from that deal.

Tom Raftery: 12:20

And you said as well that the, bot for one of a better word, pactum can go through practice negotiations for hundreds or thousands of times before actually entering into the negotiation. How does that work?

Martin Rand: 12:34

Yeah. So it actually, before every negotiation, the system simulates, a negotiation thousands and thousands of times to, to understand kind of where, where it can go. And it while doing that, it'll take external, data, internal data and information from similar negotiations, into account. So every negotiation makes every other negotiation smarter. And the external data it can take into account is, for instance, commodity price indices is a good example, or, or energy prices or, labor market indices, whatever is relevant for that, uh, use case.

Tom Raftery: 13:18

Okay, fantastic. And, you mentioned the prep call that there's a tendency now to move away from the likes of annual contracts. And this helps that. Can you speak a bit about that?

Martin Rand: 13:32

So this is happening a lot in the logistics industry. So, a month ago we did a, a public presentation and a public demo, uh, with Maersk, on the Procurement and Supply Chain Live conference, uh, in London. And, what we were talking about there was that the whole logistics industry has kind of been changing very rapidly. So it used to be built on annual contracts, but now with inflation and with energy price fluctuations, it is hard to maintain those contracts. So now the pendulum has shifted to spot pricing. As what pricing is basically agreeing every next leg of the move, ad hoc just before it happens. But, there is a problem with that. It's hugely expensive because if now, Maersk contacts, the trucker and says, I have the container here. I need to move it immediately. Well, the trucker hasn't planned capacity. Of course the prices will, they have to keep spare capacity. Of course the prices will be higher. So, we went to Maersk and asked, okay, what, if you had unlimited capacity, what would you do then? and, we came to a conclusion that short term contracts, is the answer there. So, they cannot do short term contracts today simply because they don't have enough procurement people, but a machine can. So now when the containers are still on the sea, our, bot will contact the truckers and will agree, the move, one week or two weeks ahead. So now the trucker has some time to plan capacity, and the volume is already there and they know that Maersk is not doing this in, fire fighting mode at the last moment. And this all reflects, in prices.

Tom Raftery: 15:19

Nice. Nice. It makes a lot of sense. And where to from here, I mean, you you've, gotten these big customers now. You've gotten Pactum to where it is now. What's next for Pactum?

Martin Rand: 15:30

Are we speaking short term or long term?

Tom Raftery: 15:32

Yes.

Martin Rand: 15:34

Okay. great. Okay. So, If we look at the long term, then we're dealing with buying. But there is actually in the long term, there is no reason why we shouldn't be dealing with selling as well. and we're looking at some of those use cases right now. Well, once we are dealing with buying and selling, then this is basically the majority of what companies do they buy and sell. and then, the system can go on another level. And let me explain this. So, today we are all already kind of on a more advanced level than typical enterprise software, because we don't facilitate the business process. We conduct the business process. But now once we have insight into buying and selling, we can come up with buying and selling strategies. We can come up with category strategies, that are based on a lot of such data. And so, in the future, the, human machine interaction will become more and more abstract. It is expected from people to deliver, the strategy and the machine will, do the implementation. And actually now that, the world is changing a lot we are speaking to our customers about this a lot that, Hey, let's think strategically, what is the next step? Because the world has seen now COVID, it has seen supply chain disruptions, it has seen the highest inflation in 40 years, it has seen huge commodity price fluctuations and who knows what's coming next. So now those big enterprises need to reconfigure themselves fast and this reconfiguring means that all partner deals have to be negotiated. A machine can do that, but people are expected to deliver the strategy, which way will we start, uh, running from here.

Tom Raftery: 17:39

Okay. We are recording this for, for people listening. We're recording this on Monday the 13th of June, and a story broke overnight about how a Google engineer, uh, was suspended for reporting that he, he felt his AI, that he was working on had become sentient. How would you respond to that Martin? Is Pactum ever gonna become sentient?

Martin Rand: 18:06

Uh, uh, well,

Tom Raftery: 18:09

There's a left field question for you.

Martin Rand: 18:10

Yes, uh, it's an interesting one. Uh, and I guess I, I will give it the appropriate answer then. So one of our, uh, investors is, Jon Baylin and, and he's one of the smartest, uh, AI thinkers in the world. He has, backed Open AI. he has backed the most, very prominent, AI success stories and, uh, and something that he's very worried about is singularity. So singularity is when the machines take over, singularity is when the machine will make itself smarter in such a rate where people cannot, keep up and at some point this, you know, advanced being is yeah. And why Jon funded Pactum and this is way, way, kind of looking at the very far future of course, but, it's the, to solve the problem of alignment. So to make sure that that AI doesn't take over we need to be aligned with something that is vastly, superior in terms of, uh, intelligence. And we are solving that on a kind of an rudimentary scale. We are solving that alignment problem every day because the smaller vendors need to be aligned. And the large enterprises need to be aligned with these small vendors. Both sides need to be aligned with each other, but one is the largest enterprise in the world, Walmart, by revenue and the other might be the smallest enterprise in the world. So with, developing that, hopefully we will come up with some tools to be aligned with a, vastly superior, uh, intelligence.

Tom Raftery: 20:02

I don't know if you read the reports. I, I read it this morning on Twitter and on the BBC. The output from that AI was scary in term. Not, not scary and frightening, but scary in terms of how realistic it was. And I could understand why the guy felt it had become sentient, but I obviously don't believe it has become sentient, but oh my God, it was. Wow. Anyhow, we're we're, we're getting off topic. Martin we're coming towards the end of the podcast now is there any question I haven't asked that you wish I had, or any aspect of this, we've not touched on that you think it's important for people to be aware of?

Martin Rand: 20:41

I think. Who are living, in a bit of a new technological age, and there are capabilities now available, which haven't been available before. And if we sometimes talk about, what we do, with procurement people, they didn't even realize that it was a problem until they see a solution. Well, there are systems like that, which are a solution. So yeah, I would simply say, keep an open mind and and enterprises can become a lot more flexible, thanks to that technology and to weather, the storms that are coming better, combat the possible economic downturn and as a result, create less disruption in the world. So, so everyone's in a better place.

Tom Raftery: 21:28

Fantastic. Martin has been really interesting. If people want to know more about yourself, Martin, Rand, or about pactum or any of the things we discussed on the podcast today, where would you have me direct them?

Martin Rand: 21:39

It's, uh, Pactum.com. PA C T U M. It's like pact. It means agreement in Latin. And my email address is Martin@Pactum.com. So write me an email or, or contact on LinkedIn. I'll be happy to talk to you.

Tom Raftery: 21:57

Fantastic, Martin that's been really great. Thanks a million for coming on the podcast today. Thanks so much, Tom. Have a great day and thanks everyone for listening. Bye. Okay, we've come to the end of the show. Thanks everyone for listening. If you'd like to know more about digital supply chains, head on over to sap.com/digital supply chain, or, or simply drop me an email to Tom dot Raftery @sap.com. If you like the show, please, don't forget to subscribe to it in your podcast application at choice to get new episodes, as soon as they are published. Also, please don't forget to rate and review the podcast. It really does help new people to find the show. Thanks, catch you all next time.