Automation & Digitalization: Streamlining & digitizing processes to achieve efficiency and profitability

SPEAKER A

I'm Githka Arora. I work for freepoint Commodities. I am heading up their contracts group at the moment in oil, physical oil contracts in North America as well as in Europe.

SPEAKER B

Okay, brilliant.

SPEAKER C

Adina hi, my name is Adina Fillingham. I'm the BA manager for the cotton platform with Luis Dreyfus. I'm responsible for analysis, solutions and applications for the carton platform.

SPEAKER B

Blandine.

SPEAKER D

I'm Blandine. Lewine. I work for Glencore. I've been working for Glencore for very many years, actually most of them across the street. But in 2018 we moved to Manhattan. So this is like back to our roots to be here and I head it for the US.

SPEAKER B

Great. And Jonathan.

SPEAKER E

Hi everybody, I'm Jonathan Oucher. I am vp of customer success at Clear Docs and I've worked for clear Docs since its inception as a company and my background has spent numerous years as operation manager in physical commodities businesses.

SPEAKER B

Fantastic. So I guess you'll agree with me, but a fully digitalized and intelligent workflow is the holy grail for commodity trading companies. But how can you achieve that? So our discussion today is looking into how you go about streamlining and digitalizing processes to achieve efficiency profitability. So you will hear from our experts here about their approaches to digitalization, automation of processes, AI as well, the challenges they face, but also the opportunities they see around improving the business. Please be aware that we will be leaving some time at the end of the session to take some questions. So yeah, we look forward to your comments and questions later on. There is obviously a lot to discuss around this topic so let's just dive in. The first question I have, I'm going to start with Adina on this one, is what are the key challenges you face when streamlining operational processes in preparation for automation?

SPEAKER C

So from my past experience implementing front office systems and applications, is lack of master data systems lesser master data. Another challenge is when we look at, and Alex mentioned in one of the panel earlier, is manual outside the system processes. A lot of people love excel and it's so hard to get them to adopt an application. Another challenge is having SME's locally and also looking at the domestic markets in for instance Brazil, all the tax documentation in India. There's complexity around domestic markets that we don't see when we think we can streamline or standardize processes and how we overcome some of these challenges. So I'll probably take that as well, is have strong initiatives and strong business cases and both a top down approach and a bottom up with regional, local SME's supporting us and making sure we adopt these processes.

SPEAKER B

Okay, excellent. I think, Blandin, you have some things to add to that.

SPEAKER D

Yes. I mean, I completely agree that when it comes to standardization. Right. Actually, most end users don't really want to standardize anything. They just want to work the way they've always been working with their favorite spreadsheets and their favorite ways of storing documents wherever they like. So for me, one of the biggest challenges, to get the initial buy in, to demonstrate the value of standardization. By definition, standardization means you're going to have to do things differently from what you do now and not your favorite way because it's unlikely that your favorite way is the standard way. So the biggest challenge, I think, is to be able to demonstrate this value from the beginning so that you get buy in from the users. And not only is it going to make it easier for you to deploy that solution, but they're going to then help you design the standard that will work for them, as opposed to you coming in and say, hey, you can't do it this way anymore. You got to do it this standard way.

SPEAKER B

That's a very good point. And you, Jonathan, from your past years as an operations manager, but now with what you see at clients, what do you see as the main challenges?

SPEAKER E

Yeah, so I think, as it was said, getting that initial buy in and showing the value upfront is going to streamline everybody buying into the process with clear docs. When we work with a customer, our implementation is always to start small, work through any issues, go through the complete end to end process with a use case, and then add or begin to expand the size of that use case. So when you go through and you're working, let's say with a grain company and they're moving trucks from location a to location b, you want to start with the highest volume vendors that they're receiving bills of ladings from. And then once they get those highest volume vendors in and they see the value, then you can move on to the additional remaining vendors. So getting the initial buy in is really crucial. And once you have that buy in, though, you're going to see people turn their opinion about automation digitization processes. So.

SPEAKER D

Yeah.

SPEAKER B

Okay, good. So now some practical views on this. So what are some successful use cases of automation within the trade life cycle that you have seen sort of led that really made an impact. Could we hear from you, Blandin, on that, a few practical examples?
SPEAKER D

Sure. I mean, there are several small ones, but one I will use, one that without automation we would just not have been able to do the business. And that was a case where we basically had to map what was happening on the, with the truck lifting and generate the invoice and price it and get it stamped by the regulator. This is in our Mexico business, and the truck is not able to leave the terminal unless it has this stamped invoice, which has to do like the whole workflow of data. And the trucks just literally sit there unable to leave the facility unless they have this piece of paper. So if we didn't have automation, there is no way that we would have been able to generate that invoice, get it stamped by the government and so on, and actually have it printed for the truck driver. So this was our first realization that, well, this is not an option, this is not about gaining efficiency, it's if we want to do this business, it has to be automated.

SPEAKER B

Okay, good. Interesting example. Adina, could you give us some examples.

SPEAKER C

From your experience, as Blandin mentioned in our case, also invoice generation, but we could go further to document generation and intercompany transactions, automation of contracts on the other side being purchase or sell.

SPEAKER B

Geeka.

SPEAKER A

So for freepoint, when looking at contracts specifically, I'll go through physical oil contracts. We have lots of modes of transports like rail, pipeline, waterborne bars, vessels, trucks, and any crude oil refined products that you could possibly imagine. So lots of contracts, lots of different types of documents that are coming in from brokers, broker recaps, trader recaps, counterparty contracts. We needed a way to. So at the moment, my contract analyst will sit down and compare, you know, two or three pieces of document to our trade that's in the ETRM system. We needed a way to be able to one reconcile all of these documents without a person sitting there and doing this manually, every single trade. So we looked at a bunch of vendors and we ended up looking at one of the demos that Jonathan was mentioning, use case that took it from start to finish, loved it, and we've adapted that as our AI tool. So what that does is it goes through all of these documents, compares it to our deal in the ETRM system, and highlights all of the exceptions. So it reconciles 20 to 30 fields that are on a trade that's not just for us in physical oil, but propeller and gas. We do the same thing now. So we've expanded from oil, it gives us the exceptions. So if, you know, product doesn't match on a trade, but everything else does, it gives us that exception. And then my contract analysts then can spend the time that they took to go through all of the first manual matching, they can now spend the time on clauses and reading through and putting an actual clause together, reaching out to the attorneys, responding to the counterparty and negotiating a contract, rather than, you know, just kind of going through and just manually matching one document to the other. So that's really helped us out for AI.

SPEAKER B

It's really about freeing some time for some more meaningful tasks. Okay, thanks for sharing that. How do you measure the impact of automation on operational efficiency and profitability? Blandin, could you perhaps share your thoughts on that? What sort of KPI's do you see?

SPEAKER D

Yeah, I think, well, we're not very KPI driven. I think you'll tell us about your KPI's, but we still obviously see the results. Like in the example that I brought up before. Right. It gave us scalability that we would not have had otherwise. Right. So it was, in some cases, it's literally the entry into a particular field. I think in other cases we just can free up people's time to do different things. So it's. I think we might talk about that a little bit later. But there's no such thing as, oh my God, I'm gonna lose my job to automation. Right. Because if you're doing a job that a robot or any kind of automated workflow can do, well, then you need to find a different job. Right. We need to bring value more than just repetitive tasks. So to go back to this question, how do we measure the impact? Right, well, because our employees can spend time actually thinking about what they're doing and doing what we hired them for, as opposed to buried in spreadsheets and during data and doing, you know, pivot tables.

SPEAKER B

Okay, thanks. And yeah, we have a different view on the Adina side. So can you share?

SPEAKER C

I'll be looking from two different perspectives. So we're focusing on speed to value. And when I'm saying that from a technological perspective, we deliver solutions. So how quickly do we deliver a solution to our customer to see the value? So Jonathan mentioned that we're starting small and we're measuring if we deliver an mvp in three months, what's the value that we can give or we can get out of that. From an operation perspective, we're also looking to see what the customer benefits. Build a business, ask what their KPI's are. So those would be if today takes three days to generate the documents, can we generate the documents in two days? So we're measuring that on their current applications and systems, and we're asking them what the targets are and starting small and we're thinking with the pilot we can try to measure to see if we're achieving their targets after the pilot solution is implemented. And I think some of the other KPI's besides delivering quickly to the market is number of reworks accuracy. If we're thinking of Bill of ladies, Bill of lading returns, how many do we get returns and how we can target instead of having ten per month to have five per month. And so.

SPEAKER B

Okay, did you want to add something?

SPEAKER E

Yeah. So I think everybody in this room can probably attest that if it trader has a business put on and it's making money and it's profitable, they're going to want to expand that business. And to the point with the panel here is if you can automate that business and you can scale that business, that's showing that true profitability. Where before if you said well, the operations is limited because they don't have enough people, and then you automate that business, meaning that the amount of paperwork or the amount of processing that that team can do with the assistance of automation tools can be increased. That trader, that's music to a trader's ears. He's going to go out and put more volume on, make more money and he's going to increase the size of that business. And that's what everybody in this industry is trying to do.

SPEAKER B

Good point. So there is a genuine buzz around. Aih, you might have noticed probably every panel since the start of this conference has mentioned AI one way or another. And sometimes a bit of confusion with the different terminologies, the different terms. So my question is, what positive change has that brought to your work with others, with your teams and how is AI changing how you approach automation? And I think, Geethika, you can start off with an answer for us, for.

SPEAKER A

Us from the AI standpoint, it's freed up a lot of time for my contract analyst. I have very seasoned contract analysts who have worked in this field a lot longer than I've been alive, to be honest, who would rather spend their time writing out the clauses, like I said, instead of comparing one document to another to our deal in the system and then, you know, going back and forth and saying, oh well, this counterparty said the product should be mgo marine gas oil. We have marine gas oil one. Which one should it be? Those types of exception reporting is what, you know, clear docs is doing for us. Now, those reports could directly be sent to the traders and whoever's putting in the deal or a tas that's putting in the deal, they can look that up without waiting for someone to tell them. So if a report tells you that you entered something incorrectly versus a person coming up to you and telling you that, that's always positive. Right. The other thing is also we're using clear docs to process a lot more deals, which makes, I have five analysts, if they could process a lot more in a day. Obviously they're not burned out by just looking at documents on top of documents and they can still write more contracts. So it's always more reinforcement for them that they could get a lot more work done in a day without actually measuring any kind of KPI's, to be honest.

SPEAKER B

Okay, thanks, Blondin. You had something to share on that as well.

SPEAKER D

Yeah, I mean, I think in some ways everybody wants to say that they're doing AI, right? But I mean, we were taking in PDF's and grabbing, you know, deal points out of the PDF ten years ago, and you don't need AI for that. Right. So at the beginning I was like, oh, come on, everybody's mixing everything because again, digitalization and automation have a place that is separate from AI right now. Just everybody puts it all in the same umbrella. Now, AI is obviously augmenting the capability that we had before. Right? So in the early days, you had to literally say, oh, you know, I think this field is going to be in this particular location in the PDF and that's how it would find it. Now, of course, we have document intelligence and they're able to find things. So AI, I think, is motivating people to come to us. I mean it, right. So to come to us and say we need to do AI, let's, you know, do something with AI, when actually in a lot of cases, what they really need is more, I don't want to call it basic, but earlier in a journey. Right. Than AI. But it's good because they are coming back to getting, you know, user buy in. Right. So instead of, instead of being stuck in, well, this is the way I've always done it. Now everybody is interested in coming to the table and saying, I want to be doing AI, and it gives us a great opening into how they're working to help them with a solution that is probably not very fancy AI, certainly not Gen AI mostly, but that will start with more basic automation.
SPEAKER B

Yeah, it's whatever gets them started and motivated. That's a good point. So I know we touched on that very briefly earlier on, and that's a question for Adina. First, what steps can companies take to ensure their workforce is prepared for the transition to a way of working that's a lot more automated for processes.
SPEAKER D

Yes.
SPEAKER C

I think it's key to look at the historical data and see your data sources. So minimum data sources so you don't have five different data sources that will make it difficult. I mentioned master data and also strong regional stakeholders to make sure they support you in the process.
SPEAKER E

Jonathan, so what I would say is education. I've done a couple of different panels for couple of different industries, and the thing I always want to make as an emphasis point is AI is not going to, in the short term take everyone's job. What it's going to do is it's going to help people do their job faster and better so that they can handle more and more. So the real education is internally saying to people you work with is this project that we're starting, don't worry, it's not going to affect you, but it's going to make your life easier. And I think that's a big emphasis point where people hear the word AI and they get nervous or they get scared, but it's really educating and having them understand that we don't want you in the office until 08:00 at night. We don't want you doing that. We want you to leave at a reasonable town. How are we going to get there? Or we're going to introduce more technology to do that?

SPEAKER B

That's a nice segue to the next question, actually. Did you want to add something?

SPEAKER D

Yeah, I was going to add something there, which has actually came to me in some of the sessions that we had today. Right. We've talked about in various of the sessions today. Oh, well, we're going to just use Genai, right? Genai is going to be able to compare long form contracts and tell us where the discrepancy policies are. It's going to be able to then generate the amendment and send it back to the customer. And then of course it's like, oh, well, we're not going to be able to just press a button and get that done. Everything has to be validated, read by a human. Right. Somebody who's going to be able to validate what the solution has proposed. Right. But what does that actually mean? I feel like we need to develop people with the skills to be able to take a genai generated output and say, oh, this is good, I can send this out or no, I need to validate this or that because clearly no human can actually mine the millions of documents or whatever data you've trained your model on. And to kind of come out with the same result and say, oh, I came up with the same thing, so therefore this is good. This is not how you can validate it, right? So we're going to need to think about how we can do that. And there will be jobs in that, or at least, you know, brains that will need to be able to be spent on this activity.

SPEAKER B

Definitely. So going back to the sort of workforce aspect, how do you balance the need for automation with obviously the potential risk of job displacement within your organization, or the feeling of job displacement, the fear of it. Blandin, do you want to start on this?

SPEAKER D

Yeah, I think I touched upon this a little earlier, but it's natural, right, to be a little fearful of change, and especially if you've been doing the same kind of job for some time, and now all of a sudden some it person comes and presses a couple of buttons and they do the same thing. It's devaluing in a way, right? It's like, wait, this has been my whole career to produce this thing. How is this possible that this is now happening without me? So I think we have to be very transparent and honest about that and develop ways to grow skills and leverage experience that this workforce has to better activities like, okay, so now you're not going to actually sit there and do all the comparisons, but it's much more important that for you to be able to look at the results, interpret them, determine first of all if they are valid, right? And then interpret them and assess them and ask yourself, what is this telling me? Is there a trend I should be looking at? Are there people I should be talking to about what this data is telling me?
SPEAKER B

Good answer. Geethika, you wanted to, to add something on this?
SPEAKER A

Yeah. So for my team specifically, I've made it very clear that what we're doing with Cleardocs, it's not chat GPT, it's not going to write your contracts for you. Don't confuse the two. We are using cleardocs as a tool to reconcile data so that before you start the contract you have better information. But we are not using AI or artificial intelligence to write the contract for us. That's what we're doing as humans. That's what's going to stay with us in our team. Because again, chat GPT, AI of some sort can help you put together a clause, but you can't send that out to a counterparty, just as the way they've written it, because they don't know what your deal is for another reason. You know, we talked about scalability. We also talked about in terms of when we are putting together a contract there at Freepoint at least. I mean, I've worked at JP Morgan, at Barclays. We had the same kind of rule. There has to be two sets of eyes before a contract goes out the door. So there has to be an approval process, a workflow of some sort before the contract leaves the company. We are using clear docs, or AI, as the first set of eyes when you're weeding through your documents, your contracts. The second pair of eyes is also a human. Today we have two people viewing the document. We're reducing that so that people who are reviewing the document for someone else can spend the time on their document think they can do more in other pieces. The way we're using it is small counterparties that we're working on structured trades with. We are processing documents for them. They might not have enough people, they might not have enough workforce. So what we're saying is, hey, we have this technology, we can process your documents, we can help you read invoices, we can help you read contracts, we can help you read recaps and process for you so you don't have to hire more people and we don't have to let go of anyone. So that's the way it kind of, we're approaching.

SPEAKER B

It's also to enable business.

SPEAKER A

Correct.

SPEAKER B

Basically. That's interesting. I'm going to turn to Jonathan now. What are the most critical factors to consider when you're selecting an automation technology for your trading operations?

SPEAKER E

So I would first start with the use case that you're trying to solve the goal and what your desired outcome is. And then as bladed was really talking about, like really choosing a technology that's going to work for that use case. Right. So you might come in and say, oh, we started a mexican trucking business where we're buying from one location and selling to another location, and they come and say, well, we need AI. We need AI. Well, maybe you don't. You need to fully evaluate what that process is you're trying to complete and then think about the technology that's going to be a part of that. So many times when we talk to customers or we talk to people in the industry, it's a buzzword of AI, or generative AI, and they just need to have it because they feel like they're falling behind. But to really slow down and understand what is the real technology you need for that process and what can be applied and at what scale can it be applied so it's not one size.

SPEAKER B

Fits all, I guess. Adina yes.

SPEAKER C

So I would like to add. So, yes, not all solutions are scalable, but in addition to that, other things we can take into account. Is it easily integrated with other applications? Can we build APIs to receive send data? Can we customize it?

SPEAKER D

Right.

SPEAKER C

Tomorrow the trader comes and said, I would like this new process. I would like you to start invoicing this way and produce these documents. Can we customize that ourselves or do we have to go to the vendor? If we buy or if depends on what we do with the solution, can we maintain it going forward or do we rely on someone else to maintain? And do we have documentation to know how to maintain that solution? Also for, you know, sometimes we have critical emergency releases. Can we release it or do we need to go to an other party for the deployment process? So those are some of the factors that I will take into account as.

SPEAKER B

Well, very important ones. Okay, we're going to soon open up to questions from the floor, but I guess I have a sort of big question to ask, sort of crystal ball question, a three part question. So please bear with me, and I want an answer from you all. So what emerging trends in automation and digitalization do you foresee will be playing a crucial role in the future? What next gen automation technologies you believe will have a significant impact in the next, let's say, five to ten years? And then finally, the last part of the question is the role of humans in that, you know, how do you envision their role? The role of traders, for instance, involving as automation technologies continue to advance, continue to develop. So who wants to go first? Geeka sure. That's the perk of sitting next to me.

SPEAKER A

So I'll give the example. I think for contracts, specifically financial deals, we have an idea called econfirm, where there's an exchange platform, two counterparties will come on to that. They will sign up and put their documentation in there. So it's onboarded. Nobody has to onboard every single counterparty, but they can e confirm a contract there, and there is a master agreement behind it, physical oil or physical commodities. You can't do that because we have very limited amount of agreements that are, you know, standing behind it. We have General Trump's conditions behind every deal. But I could see e confirm playing a part into physical deals where economics of a deal could be confirmed on the platform. If everybody gets onto the platform, there are different technologies that are already starting this with the big firms in oil, where they're adapting and they're saying, okay, I will come onto this platform, I will load all my information there, I will connect my ETRM system to this platform so it could send the information and we could e confirm there all the fine print, all the general terms and conditions and all the clauses that still needs to be done by a human behind the scenes. So I think that's for at least my world. I could see that and not maybe five years, but ten.

SPEAKER B

Okay. Okay. That's hopeful. Adina.

SPEAKER C

So with confidence that AI will provide good output, our goal would be to focus on strategy, decision making, and leverage AI for low, time consuming tasks. So Blandin talked earlier about how do we reassure our employees that they're not going to lose their job if we shift the message and we convey that we're focusing on time savings, on reducing these low value tasks that can be done automatically by I, we can focus on analysis, decision making and strategy more.

SPEAKER B

Absolutely Blandin, I would say.

SPEAKER D

The important part is to take the first step, right? So maybe everybody has already, but everybody is in a different place on the road, right? Maybe for some of you we're like, oh, I'm going to go to this digitalization and automation session because I haven't even done AI, so I need to do that first, right? Or you've, you already have chat, GPT generating your annual report and anything in between, right? So I don't, to me, anyway, it's not going to be like, oh my God, this big revolution where everybody is going to reach the same level of maturity at the same time. I think as a general industry, certainly on the physical side. And the gentleman who was from Binco, who was speaking at the opening session this morning said that, right, many of us are still so manual and we're not all even doing electronic bouts, right? We're still like manipulating these pieces of paper. So we have a long way to go, but if everybody again gets on that road, it will get us all to a better place. As far as the emerging, whatever your second question was. Nextgen, next gen. Exactly. I mean, how could it not be AI and Gen AI in particular, right? I think it's just going to amplify, enhance everything that we were already doing now. Kirika, I don't know if ten years from now we'll still have humanity, people actually writing contracts. Maybe not exactly. I think maybe ten years from now we won't. But we have to have trust and faith that we will figure out the boundary of where AI and GenaI can take us into this world. Or do we stay in the, no, we're going to use it for reconciliation.

SPEAKER B

And validation and how we make this role evolve, like to be part of the ecosystem in a more meaningful way. Jonathan.

SPEAKER E

So I will say that the future is going to be split between people who have started investing into automation and AI projects today and people who have not. And I think you'll see in five, eight years, as you look at trading businesses and you look at the margins and the businesses that certain companies will be able to be in, and to have good margins on that business partly is going to be driven by the automation and the AI that they're investing in today. So I see a future different than what the technology would be, but firms who invest in AI and firmsen who do not, and it will be pretty evident in eight to ten years of the people who do and do not.

SPEAKER B

Okay, thanks. So I'm now turning to the audience. We have a couple of minutes for your questions or comments for the panelists, so don't be shy. I'm going to give the mic to my colleague.

SPEAKER D

There we go. So I'll just open it for the panel. I think one of the questions I have is it's a harsh reality that generative AI is going to change the way we do business, and it is significantly going to change it in the next ten years. What is one limitation you see within your organizations that would be a hurt in embracing this with open arms?

SPEAKER E

I would say commodities firms, in general, commodities trainings firms are very secretive to their trade data and to their personal company data. And so being able to trust a large language model, a Gen AI platform, that they know they'll be able to rely, that their data is not being shared. I think that's a big security concern. We're hearing from our customers, is making sure the security protocols are in place, making sure that Genai large language models, who they choose is not, is going to retain their data and data privacy.

SPEAKER D

I think I would add the regulatory environment, right. Because they tend to run a bit behind. And the more we have solutions that are AI powered, that a regulator can come in and say, well, how did this happen and who, you know, how was this produced and what algorithm did you use and so on, we're gonna have to be able to come up with some answers for that. So that is a challenge, I think.

SPEAKER B

Yeah, on the previous discussion we had in the other room, it was all about explainability of AI. So, yeah, that's a very good point. Any other questions?

SPEAKER F

Hi, I have a question for all of you. Whomever can answer, first of all, thank you for being here and all this wonderful explanation. So I agree with what you have been said here about, like, saying, like, in ten years, maybe not a person would be writing contracts, right. But what I understand is that when we are talking about AI is implementing that for known sources, like, for example, paper to paper comparison or contracts to contract comparison. Right. But for example, I work in compliance. And when I have to do a KYC procedure, I have to turn to the Internet to dig farther and farther and farther.

SPEAKER D

Right.

SPEAKER F

So I'm wondering how you see AI helping that process, being that the Internet is a wild word. And also we have a dark website that can generate fake, deep fake information. So how do you see AI help as a helping tool to automatize that screening process?

SPEAKER B

Any volunteer?

SPEAKER E

That's a great question. I think I'll try to do my best to come up with an answer, but I think one thing to consider is that if you look at your firm's technology and AI, I think there's going to be a lot of restrictions on what it'll be able to actually generate. So in the KYC process, you're going out to try to find background information, history of people understanding, you know, really what that company or that person is about. So I envision where there's going to be security protocols where honestly, in that type of scenario, I wouldn't want to have an AI generated solution. I'd want a human there validating that information.

SPEAKER B

Your job is safe. Then.

SPEAKER F

Contrast writing. You will need someone to validate that information. That is how I see it. Right.

SPEAKER B

Yeah. It's going to be collaborative as opposed to a replacement. All right. Well, I think that's all the time we had today, so that was really interesting. Thanks a lot, panel. Gitika, Adina, Blondin and Jonathan, please give them a big round of applause.

SPEAKER D

It.

SPEAKER B

Next up is a discussion on how to develop. Let me just put it up. Yes. How to develop your technology, infrastructure and applications for the era of sustainability. That sounds fascinating. Please welcome the panel. Darcy Curran, Jad Orr, Magesh Nair, and Chris Elise, thank you very much.