AI’s Impact on Energy Trading: Insights from Istvan Czilik, CEO at Revenue. AI on Future Disruptions, Challenges and Opportunities

Ben Hillary – Hello and good morning, good afternoon or evening to all of our viewers, wherever you're listening in from today. My name is Ben Hillary, Managing Director of Commodities People, and I'm really, really delighted to be joined here by Istvan Czilik, CEO of Revenue.AI, a leading developer of AI solutions for the commodity trading sector. This is actually quite a special interview as we've personally worked very closely with Revenue.AI in recent months to develop an AI copilot and enhance searchability across our own data and our many, many years of recorded content.

We've been really, really delighted with the results. Do please check for yourself by going to our Commodity and Energy Trading Insider content hubs. Over the next 15 minutes or so we'll be taking a deep dive into all things AI.

We'll be talking current adoption, the low hanging fruits, hidden opportunities, roadblocks, use cases, and finally a glimpse into the future of what the future holds for Revenue.AI and the wider world of AI in commodities. So Istvan, welcome and thank you very much for being here. 

Istvan Czilik – Yes, absolutely. Thank you very much for inviting me and I am very happy that actually I'm going to be able to talk with you. We got to know your team very well during this small project, that you are mentioning. So I can also recommend it to everyone to please have a look. I think it's a very, very useful tool altogether for the commodity lovers to learn anything around the market. What you can learn out of it is like outstanding. 

Ben Hillary – Yeah, absolutely. Thank you. Well, let's get the ball rolling. If you could start by just telling us a little about what it is you do and yeah, by giving us an overview from where you sit, current status in AI adoption for energy and commodity trading.

Istvan Czilik – Absolutely. So, not surprisingly probably that if you want, the listeners are aware of who is Revenue.AI. So let me start there for a little bit. So we are probably one of the few companies who are born agile in AI based on the work that we started to do. So we started around 2020 with purely the mission of making sure that actual pricing is going to be AI-enabled and making sure that actually we are helping in the AI adoption for anyone in the market. And what is a better area for this than commodity trading. Surprise, surprise. 

So this is how we are starting to work with commodity trading as one of our major markets where we are providing solutions. And inside there, the special niche is energy, which is even more special than the rest of the commodity trading, especially in the AI part, I would say, because it's more ahead.

So it's an interesting topic what you were asking, because we have been working with multiple big commodity houses, small commodity houses across various areas of what they are trying to achieve, especially around pricing or AI strategy. And what we have seen is that energy trading is ahead of the curve inside the commodity space. So we have a lot of requests. I can really easily categorise the various incoming interests from the different companies. Now that, there is this AI hype, everyone wants to implement AI. What I would say is that lots of the companies are still in the experimental rule. So what this means is lots of companies want to implement a ChatGPT-like solution to help them have some quick wins, but there is no real strategy.

Then there are some beginners who are actually actively looking into enabling their capability growth and they already have some success stories. So there are some very interesting business cases coming our way, where we were able to already successfully deliver results in this space. For example, even funnily, some stuff like daily PNR reporting is enabled and augmented by AI. It actually can become a very fluid exercise and not so resource-heavy as it was previously in the market space. And then when we talk about leaders, this is where actually there is some very smart energy trading companies who did already win because they strategically place bets in some areas and they actually have a very well-deserved AI vision, which AI vision is enabling them to land in some classical areas. Again, but now with AI enablement and AI empowerment, like trading and others, which is helping them to be ahead of the curve and ahead of the game. So that's pretty much it. This is where I see a lot of the game players. 

Ben Hillary – Yeah. Excellent. Which business areas do you see as receiving the most attention when it comes to applying AI, essentially, which are providing the lowest-hanging fruits? 

Istvan Czilik – Yes. So I would say that lots of the middle office and trading enablement areas are capable of AI enablement. So my favorite areas are actually trade execution and forecasting. For example, in this space, immediate wins can come from automated trading or algorithmic trading, naturally, or forecasting and similar. But also another very interesting area which I see coming up, for example, is market analytics. So one of the things that can come up with AI enablement is that actually you are able to understand your market much better. And because of that, you are able to implant these market analysis into multiple areas of your actions. 

Coming up with immediately the next area, which I really see as an interesting and exciting element, which is risk. So risk and compliance, I see that here, like a lot of classical, from an AI perspective, classical best practices can be re-implemented simply by using the currently available risk databases and all the market analysis information. So that actually risk can be automated by AI and it can be much smoother, which is naturally let’s not to forget it is one of the trickiest elements in trading altogether. So it's very important to have as secure a company as you can get from this perspective. So this is where pro protection or memory detection and similar areas can be really quickly enabled with the use of AI technologies. 

And last but not least, if you don't mind, this is the last point, right? Is a real trade capture and operations that actually some years ago we have already been talking about it, but I really see that the combination of robotic automation together with trade capture is a wonderful area for AI adoption altogether, to make sure that actually every part here becomes as automated and as little resource-having as it can get. 

Ben Hillary – Yeah, excellent. Conversely, which areas do you see getting less attention presently than they might deserve, but you can see providing potentially great benefit? 

Istvan Czilik – So I would say that one of the things where it's a bit more complex and because of that, it's naturally like coming as a one of the not in the lower hanging fruit area, is optimisation of trade strategies.

So here what I see is that because the market analysis, which I was already mentioning and am catching up, needs to get more and more data, and that needs to be adapted to the trade strategies. This is where I see that there will still be lots of pickups, but it's still in its infancies when we look into this area. And I would say that especially connected to energy trading, I would say, and renewables, I would say anything connected to smart grids and the related usage is also something where I see a lot of places where improvement can be made. And it's not necessarily because people don't want to make this AI adapting steps. But actually, like because of this, the whole smart grid is so intensive, it's literally so in nature. So that I would believe that in a couple of years, it's going to catch up big and go ahead this way.

Other areas, which I would like to mention, which applies literally to every commodity trading, no matter if it's energy, are more on the back end, right? So like lots of things, people look into the fancy stuff, right? How can I do automated trading? How can I actually make sure that I have my ChatGPT-like solution, as I mentioned, but AI can really help in things like data, data quality, data gathering, integration in between existing infrastructure elements and commodity companies literally have a history of not investing here. So I would say this is a very big area for implementation to the point that actually this is one of the areas where we invested ourselves as well, to make sure that we can help commodities to grow faster in this space. Because I really, really find that like in lots of cases, whenever I go into commodity trading houses, I start to talk with someone who is playing global and I am talking with someone who has like ten different ETRM systems or even more across the global, which means that you can dream about AI adoption. But ultimately, as long as your basics are not done well, there is a bit of a challenge as to how you can grow from that area. 

Ben Hillary – Yeah, excellent. What do you see as the key roadblocks to the development of AI within energy and commodity trading and how do you think we can resolve these challenges? 

Istvan Czilik – Very nice question. Yes. So there are some patterns. There are some patterns, definitely. So I would say let's start where I left off then, right? So like this disconnectedness of the systems. So this is one of the roadblocks that I see for enterprise wide adoption, especially for bigger companies.

So there are multiple CTRM systems or other systems running around. And because they are not connected at all, like this means that the data they are speaking is not the same language. And if you don't have the same data language being spoken, then this means that there is going to be a very difficult way you can connect a Singaporean trader to a London trader, to a Geneva trader, to a Chicago trader, which is a problem. And this is where these decisions must be made very, very well in every area of the market. 

Then I would say that there is going to be a guide to lots of quick wins, there are some very exciting quick wins, to be honest. Actually we also built a couple of steps to enable these because these are key for the win adoption. But like ultimately then you start to scale. Like the quick wins need to turn into an organised approach. And this kind of AI governance is going to be an important element for the adoption for multiple reasons. On one hand, because any company who adopts into this space will have to have some kind of leverage, like what kind of AI solutions you will have. This is one.

But on the other hand, beyond having instant benefits, what is going to come up that any kind of AI solution will need to go through a compliance stats as well. So every global government will implement AI compliance requirements. And if you are not taking it seriously, then you should, because all of these things are going to be resulting is that actually some AI solutions you are investing today with a complete disregard of these directions, will end up obsolete or worse in a file, because simply the solutions are not compliant with the regulatory requirements as well. 

And I would say, let's park it here. I could go with a couple of other examples, but let's have a part here. Let's put it this way. When we built our own best practices, we really created a five step process to guide through any adapter in a direction. It's like, how can the team be trained? How can they actually start to build their quick wins? And how can these quick wins be turned into immediate business value and after that into an AI strategy. Because without this, literally, there is no scalable AI in any trading or any company whatsoever.

Ben Hillary – Yeah, excellent. Well, shifting this conversation, then, looking at Revenue.AI itself, what developments can we expect from your side from Revenue.AI in the coming months? 

IC – Yes, thank you very much for asking. This is also very exciting area, naturally. As I was saying, we were born with AI vision, and we really believe when everyone was saying is chatbots are not relevant, we were already focusing on, it's not about having a chatbot, it's about talking to your own data and your own system and making sure that you can chat with that. So this is where we started to build various copilot personas to be able to talk to your software. And in this way, I would say is that what we are doing right now is that we are going to be consistently refreshing the core offerings what are successfully on the market at the current moment.

So, for example, this five step process, which I was explaining on AI adoption, now, we have plenty of relevant, low hanging food business cases, so anyone is coming our way, we are putting there more and more. So there is more ideas for anyone who is just knocking on your door, or our door, saying it like, “Hey, Istvan, can you help us with PNL? Can you help us with IPV? Can you help us with market-to-market?” We can go and help with direct ideas like how that can be turned into not resource-intensive and very much AI power solution. So this is one part.

Also, we are continuously refreshing and releasing our nice copilot there, Zeta. So, which is actually like one of our flagships for making sure that our communication and reporting works out there for commodity trading companies.

So we invented this very ingenious process where anyone can feed down any kind of data coming from their ETRM, CTRM pricing, or whatever databases and play it back. So like Zeta can become your AI advisor now. Zeta is now getting trained on more and more areas. So the stuff that I was mentioning about making sure that the VAR can be clearly done by Zeta or there can be a PNL report and the PNL report, you are getting an absolute refresh on that one. Or the work automation for a trader can be provided by Zeta. This is where we are providing a lot of updates as we move on with time, so the functionality gets richer.

And last but not least, like when we go into data, another element where AI is going to be a disruptive technology, we are actually working together with a couple of the data providers, global ones and local ones as well, to be able to provide a uniform data strategy for AI transformation. So this is also a very big destructive elements. 

Back to the trend. What you were asking, right, is like, because market analysis is crucial, making informed decisions on trading is crucial and making sure that there is a great decision when, let's say we are lending a trade, how the margin is going to end up in the end, is crucial. This is where data analysis and data unification, we make sure that this becomes not heavy on resources because what's happening today – and then let me finish over here, right? – Potentially big companies have lots of data crunchers who are delivering the data into their databases. Data comes in with a couple of months delay or sometimes if the data organisation is bigger than its couple of weeks delays. And then basically then you can try to make decisions. By the time you get to the data, the decision point is obsolete.

So what happens is like it's going to be a drawback in your delivery. So this is where our implemented process and our invented process makes it literally real time. And that part is the part where we are coming up with newer and newer data integrations and releases thanks to our partners and our own engine as well.

Ben Hillary – Excellent. Final question from my side would be we're looking into the crystal ball. We're having this interview five years from now. What are we discussing? 

IC – Okay, if you don't mind, I will start with the challenging part, right? And the challenging part is like the rule of the trader. So I will be really like, if anyone asks about five years and what's going to come in five years or even ten years or similar, coming with AI background, I can say it is very difficult and very challenging to predict. So I would say that there is going to be, I'm not sure if you are familiar with the term agile, artificial general intelligence, so I think that's going to land a huge disruption in the trading industry as well. And this is why I'm mentioning it. Like we will be talking about on one hand what is the actual future role of a trader and how to leverage it best when we are looking into trading, because this area is going to grow incrementally and exponentially in this space. 

And the other one is going to be like how to avoid the mistakes others made. Because the other area, which I am believing is going to happen quite a lot, is that now there is a lot of adoption hype. The adoption hype is going to result in several issues and I'm pretty sure that there will be some global risks, frauds, mistakes, right? We can come up with a business case and we can talk about that like how to avoid those kinds of issues as well.

Again, back to the point. What I am seeing is that it's okay to build an internal with less of a strategy, but if your AI system makes a decision which is ultimately going to become impactful for the trading community, that is going to be a completely different goal game. So I would say that in five years, AI is going to be everywhere. It's going to be commonplace. So we will be talking fact-based, like how to avoid some of the mistakes and what is the future role of the trader as the big things. 

Ben Hillary – Excellent. Good. Well, that pretty much brings us to time. So well, I'd say thanks. Thanks very much, Istvan, for a fascinating conversation. Also to our listeners for joining and until next time, wishing you all the best and thank you. 

IC – Great. Thank you very much.