Podcast

Welcome to the Data Talk podcast. My name is Jirka Lecherek, and it's my privilege to welcome two esteemed guests from Enverus: Rich Tuehrs, SVP of Engineering. Hi, Rich.
Hello.
And Lukáš Turek, Senior Product Owner at Enverus. Hi, Lukáš.
Hello.

In this episode, we will explore how data analytics, machine learning, and Gen AI are transforming the energy sector and what Enverus is doing in this field. We will discuss Enverus products, their office in Brno, why a global company such as Enverus has a branch in Brno, and what the Brno branch is working on.

But first things first: for those listeners who haven't heard of Enverus yet, Rich, can you briefly introduce the company, its history, and what you do?

Let me tell you a little bit about Enverus. We are the leading SaaS provider of software for the energy industry, and we have been building technology for energy markets for over 25 years. Our mission is to power the global quality of life through intelligent data, and that drives everything we do. Our solutions cover the entire energy value chain, from exploration and production to trading and renewables. We are headquartered in Austin, Texas, but we are truly global, with offices across North America, Europe, and India. One of our newest locations is Brno, which opened in 2022. How big is Enverus, people-wise?

We have just under 2,000 employees.

So let's start with you. What's your story? How did you become SVP of Engineering at Enverus?

Sure. I studied computer science in college and was fortunate to land an internship that got me started in software development. That was over 30 years ago, right around the time the dot-com boom was just getting started. I spent a big part of my early career as a developer, which definitely helps me relate to the teams I work with today. After spending about 15 years as an individual contributor, I found myself drawn to leadership. Since then, I've had the opportunity to lead some incredible teams building software for the energy industry. Back in 2018, the company I was with was acquired by Enverus. At that time, I was leading software engineering. Since then, my role has grown. Today, I lead not just software engineering but also generative AI, data science, and research innovation teams. My focus is really on shaping the engineering and Gen AI strategy that drives our mission and keeps us aligned with the evolving needs of our customers.

What technologies did you use when you used to code?

When I used to code, most of my development experience was in Java. I did a lot of work in Java. On the front end, it was mostly Servlets—basic Servlets. Then it moved into more JSP development. Later, we worked with more traditional frameworks like Angular and React.

Do you still code?

I do not code anymore. I wish I did. Coding was where I always felt like I was able to provide some creative energy. I can't play any instruments or draw, so coding was my creative outlet.

[Text ends here.]

Tady je opravený text s drobnými gramatickými a stylistickými úpravami:

I always felt like development was where my talent was, where I was able to express a bit of my creative nature. So what about you, Lukáš? What's your story? How did you become a senior product owner at Envers?

I actually come from a technical background. I studied information technology and started as a QA engineer. After that, I spent some time as a developer, and then I found my way into product management when we launched an internal startup at the company I was working for. I joined Envers over a year ago as a senior product owner. Now, I work closely with our UX team on the main product, Prism.

When I was looking for a new job, I wanted something that would be both impactful and innovative. The energy sector is something really fundamental, and I was lucky enough to get hired based on my technical background and product experience, even though I didn’t know anything about the energy industry at the time. So it wasn’t a requirement for the role, which was great because it gave me room to learn as I went—and honestly, I’m still learning. There are many fascinating use cases and problems we’re trying to solve.

Yeah, that brings me to you, Lukáš. You were hired in Brno. You mentioned that Envers is a global company with almost 2,000 employees. So how did Envers come to Brno? Why Brno?

That’s an interesting story. Our CTO actually opened an office in Brno over 10 years ago and had a lot of great success there. He felt that the talent pool in Brno was rich. There are a number of universities in Brno, so he always felt really positive about the city and had great things to say about it.

We planned to return to Brno around 2020, but as everyone knows, COVID happened. So we actually waited a few more years until things started improving, and then we decided to go back in early January 2022 and started opening positions. I was lucky enough to hire the first handful of engineers in Brno around March 2022. So was 2020 the first time you heard about Brno or the Czech Republic?

To be honest with you, I didn’t hear about Brno until 2022. When our CTO was talking about opening the office here, there was another gentleman, one of my peers, who already had people in Brno and was familiar with the market. He actually came out here in January, and then around February everything fell apart. So I probably heard about Brno in late 2021 when we were planning for 2022. So when opening the new office in Brno, there are almost 200 people there now?

Right now, there are about 110 people in Brno, mainly focused on the technology sector. Of course, we have some HR staff to help recruit new positions, and we also have product management, but we’re really heavily focused on tech.

Pokud chcete, mohu vám pomoci i s překladem do češtiny nebo další úpravou.

Here is the corrected version of your text with improvements for grammar, clarity, and flow:

At the moment, we have around 110 people. We hope to grow to close to 200 by the end of the year, with even more growth as we move forward. Exciting times!

What does it mean to open a new branch with your first-hand experience coming to Brno and hiring the first group of great people there?

It’s exciting because I feel like I have a personal connection to that office. I really feel like I have a piece of it. The people I’ve met there are amazing—great, motivated, highly talented individuals. It’s just something truly exciting. Whenever I have the opportunity, I try to come out here because every time I do, there’s this vibe, this energy. People in the office are enthusiastic and enjoy working there. We have a great relationship because I was there early, got to know the team, and helped build the office up. So yeah, I really enjoy coming out to the office.

Great to hear that. I hope you will continue to enjoy our country and Brno especially. I love Brno—I used to study there.

Let’s move to the energy market, the sector Inveris is driving innovation in. Lukas, you said you had no prior experience in the energy sector and that your expertise was in technology, UX, and product management. What was the most surprising aspect of joining Inveris, a global leader in data analytics in the energy sector? What struck you the most?

As I mentioned, I didn’t know much about energy when I joined. During my first round of interviews, they gave me a walkthrough of the products I’d be working on, along with some interesting insights into geology and energy. I was really impressed. I thought, even if I didn’t get the job, I’d still have learned something new. What struck me most was the enormous amount of data we need to process and the fascinating geological facts we need to understand to help our customers gain insights from that data.

Yeah, what I love about it is that the energy sector is really important—which might sound like a meaningless statement—but everyone has recognized its importance in the past few years, especially with the war in Ukraine and the gas situation with Russia and everything.

From your perspective, Rich, can you give us an overview of how the energy sector will look in 2025? You’ve been in this field for a longer time, right? You mentioned the previous company Oil Dex.

Correct. I’ve been in the energy market for quite a while. When I first started, we primarily focused on the oil and gas market. Over the last five years, there’s been a shift from referring to just oil and gas to the broader energy sector because of the ongoing transition. When I was at Oil Dex, we primarily served the North American market. After we were acquired by Inveris, which operates globally, the company transitioned from being a core oil and gas company to serving the entire energy market. They were building up their power and renewables division at the time when I started…

If you want me to continue or edit further, please let me know!

Here is the corrected and polished version of your text:

I think when we started, power renewables wasn't even a division or business unit at Inveris. It was really part of another business unit. Then, a few years later, it started growing significantly. Over the past five years, we've seen double-digit percentage growth—sometimes 40%, 50%, even 100% growth—because that market has been expanding rapidly. This growth wasn't entirely organic; a lot of it came through acquisitions. We acquired several fantastic companies, both in the United States and Europe, including a few in Spain, which really helped us advance our power and renewables products.

If we look at Inveris’s customers, can you name some? Like producers or sellers?

We operate with all the major operators in the North American market; many of them are using many, if not all, of our products. I probably won’t name specific companies, but if you think of the big players, they’re likely our customers. Additionally, we serve thousands of operators globally, and we have a high percentage of those using various products.

That was what really amazed me—how large a part of the entire value chain your products cover. As Lukas said, from primary resources, such as determining where geographically to drill an oil well, to energy trading and the business side of things. Rich, could you go through the product portfolio or the value chain? What are the use cases? How do your clients use Inveris products?

Sure, absolutely. Happy to do that. As mentioned, oil and gas energy markets are massive. There’s upstream, midstream, and downstream — and we cover all of it.

Could you explain what upstream, midstream, and downstream mean?

In simple terms: upstream refers to getting the resource from the ground; midstream covers transportation; and downstream is the consumption of oil and gas.

Great to know—and you cover all of that?

Yes, we cover all of it. That’s quite a range—from resources to selling energy. So, how does the portfolio or product suite look? How many products do you have?

We have a number of different products across several platforms. Our flagship product platform is Prism. It’s our core intelligent platform for energy professionals, bringing together massive data sets, machine learning, and visualizations to help users model production, optimize operations, and make smarter decisions across the energy market, including power renewables. So, Prism is the centerpiece, the heavyweight of our product suite.

That’s where the bulk of our products live and where most of our revenue comes from.

If you want me to make it more formal or more casual, just let me know!

Here is the corrected version of the text with improved grammar, punctuation, and clarity:

As well. So, can I use a different product without using Prism? Yes, you can. We also have products from where I came before we got acquired. Our flagship products are called Open Invoice and Energy Link. These are deeply integrated in the AP automation platform and also support revenue and joint venture accounting. These tools handle billions of transactions and simplify financial workflows across the industry. About half a trillion dollars flow through these platforms every year.

Then, we also have our power and renewable energy applications that are enabling the energy transition. We build analytics for siting renewables, energy project optimization, grid asset management, and support energy traders with real-time insights. These are critical tools for navigating the shift towards a more sustainable and decentralized energy system.

So, how many products approximately are under the Aniverse brand? I would say, across the different product lines, we probably have 25 plus products.

Okay, that’s a lot of products. How much of it is decentralized and how much is integrated into one platform that „rules them all,“ like Prism?

We really centralize around our data. While we have around 25 different products, the centralization happens at the data layer. As we receive and create data on our side, that information gets pushed down to our data layer where the different applications share and enrich the data. For example, our Open Invoice product might create or receive data from our partners, and that data can then be leveraged to provide better analytics in our Prism application.

Interesting. So, if we go into your data lake and review the data, what would we find there? Lukáš mentioned geographical data—do you also cover GIS data and similar information?

Yes, we have well data, production data, forecasting data, intelligence data, invoicing data, revenue data, joint interest billing data, and a lot of information about the power grid. We also have trading information and commodity data. So, really, we have a comprehensive set of data.

That sounds like you are an energy data powerhouse. How do you use that data, or more specifically, how do your clients use it in your software? How does it help?

Where can we find Aniverse?

Actually, we can say that the ultimate goal of our customers is to predict their return on investment. In practical terms, that often means helping companies figure out where to drill the next well and being able to forecast how much it will produce. Sometimes, we can do that up to 40 years into the future. That forecast is what we call a decline curve, and it usually starts with high…

Let me know if you'd like me to continue or adjust anything!

The ion is strongest at the beginning, and then it gradually drops over time. However, some real-world events can throw off the prediction. For example, if a company decides to refracture the well to boost production, or if a competing operator moves into the same area, the original curve no longer holds. Our platform can detect those changes, factor them in, and automatically recalculate the forecast, which can save a lot of money for our customers.

We are also doing similar work in the power and renewable energy sector. One big challenge is predicting electricity prices. These prices are driven by historical trends as well as the weather, especially because there is so much renewable energy from solar power. We have some models that work quite well for seasonal patterns, but short-term fluctuations are much trickier.

With the rapid growth in solar power, we are seeing a new challenge: negative energy prices. This happens on sunny days when there is more energy being produced than there is demand to use it. As a result, the price drops below zero. To keep up with these trends, we have to continuously retrain the models.

How many models do you have?
When we talk about models, is there one that knows all, or are they specific?
I suppose not. These are specialized models for each use case and each industry.

That's correct. We have thousands of different models used throughout different products.

One thing I wanted to add, too—Lukasz mentioned a number of use cases. We also have a big use case around our source-to-pay and order-to-cash processes. This is where, again, the operators in the field are drilling. When they start drilling, they need to purchase equipment, move materials, obtain services, and parts—essentially everything they require.

We provide a full source-to-pay platform for our customers, allowing operators to transact seamlessly. So, whenever they start drilling, they need a supplier. That supplier will either provide parts or perform work. Consequently, an invoice needs to be created so the supplier can get paid. This can all be done on our platform. We also help them find the suppliers they need among thousands available in the market.

We have over 40,000 suppliers transacting on our platform every day.

That's really interesting. So one part of the universe is like Amazon for the entire drilling sector?

That's exactly right. We help them manage their contracts, maintain their price books, and essentially handle the full source-to-pay side of the business within Inverse.

Additionally, on the trading and risk side, we support commodity traders by providing real-time market data, forecasting tools, and historical pattern recognition, helping them make confident, time-sensitive decisions such as…

Here is the corrected version of the text:

When to sell, at what price? We give them the information to help them decide when to sell, when to buy, that type of stuff, yeah. We don't actually make the trades or anything like that, but we provide them with the information to help them decide when to trade.

Interesting. From my perspective, the more you go downstream into trading, into invoices, into the marketplace stuff, the more common and business as usual it is. What really amazes me is the upstream part. How big is the upstream part, like drilling the oil wells? Have you been to the oil well with your background and stuff?

I personally have not been out there.
Me neither, unfortunately.
Unfortunately. But how big a part of the whole business is it? From my perspective, what you're talking about is getting smaller and smaller in the market industry, the economy aspect of it, the business side. Is it really important?

Right, it is still a very big part of our business. It's probably over two-thirds of our business.

Okay, interesting. Why I'm asking is, with all these propositions and products, it must be really hard to prioritize what to develop, what kind of products stick together, where to grow focus, where to invest, and what kind of company to acquire. So how do you look at that?

So our products like Prism, which focus on the energy analytics side of the business, such as where to drill and all that type of information, still make up a large portion of our revenue. But we're investing heavily now as the transition is happening into power renewables and really investing a lot there to build up that side of the business.

So, starting about five years ago, if I looked at our revenue percentages, power renewables was in the single digits. Now it's a double-digit part of our revenue. So it’s growing and is our fastest-growing revenue segment by percentage.

Great. What about Brno? So with all these product propositions, Lukas, what do you do in Brno? What's Brno's role in all of it?

I work specifically with our Prism EX team and our mission is to make complex data feel simple. All the data sources that we just talked about, like land parcels, transmission lines, power plants, and all the analytics for wells — we want to serve them in a human-friendly way. To do that, we use everything from intuitive visualizations, dashboards, to generative AI, which we will probably talk about. In general, we just try to be creative. It’s also a multidisciplinary area. On one hand, we need to understand the energy industry, but on the other, it’s still modern web development. So we use all the product management principles, etc.

So is Brno focused only on Prism?

Brno is actually focused across the board. They’re working on all our different products from Prism to our open invoice product, our energy trading, and register. They’re working across the board — across multiple products and…

Let me know if you want me to continue or polish it further!

Here is the corrected version of the text:

Layers in Brno, that's really interesting. Coming back to prioritization and the acquisition of a lot of different products, and you, Rich, being SVP of Engineering, that must be really hard technology-wise, I suppose. Each product was based on slightly different technology—there are a lot of programming languages, architectures, stuff like this. So can you walk us through that? Are there some principles? What are the most common languages you hire for in Brno, and things like that?

Yeah, absolutely. As you can imagine, as you mentioned, we do have a large product set and we acquired a lot of those products through acquisitions. Therefore, we inherited the technology that came along with them. So we are actually using a lot of different technologies. Python is obviously a big one. We continue to use Python, and that’s probably one of the most popular languages we are using. We also have a lot of stuff in Java and C#. On the front end, we use all the major frameworks—React, Angular, Vue. We have products that have been around for 25 years, so in those products, we'll also have technologies like JSP still actually in use. So basically, we have pretty much every technology stack you can imagine.

When we acquire companies, we centralize around the different products—using GitHub, for example, where all the source code is managed. We have similar pipelines for building and deploying the software. We have common QA practices for testing and automating our products. We do a lot of test automation, so we try to centralize around some test automation tools as well. But yes, we have a wide variety of technologies. You name it, we have it all.

Obviously, we also leverage a number of the big platform providers; it might be AWS, Azure, or Google Cloud. So we support those environments and operate within them as well. We are containerized.

Do you still live on-prem?

We do have some on-prem components, especially with certain clients. We are a hybrid cloud company; some of our products that have been around for 25 years still run on-prem, but they have also branched out. Even those products often have a cloud presence now.

Because we’re such a large data company, we have a lot of data processing—we have ETL pipelines, and use a variety of database technologies. We use relational and NoSQL databases, some columnar databases, some that live on-prem, some in the cloud. We even have very high GPU-based databases. So on the database side, we have it all as well.

So there’s not just one single place where all the data lives—it doesn’t sound logical given all the different data uses and use cases, right?

Well, you have the systems that provide the customer their experience through the UI. However, as we push the data down, we…

If you'd like, I can continue or polish the remainder as well!

Tady je opravený text s úpravou gramatiky a stylu pro lepší plynulost a srozumitelnost:

We do have a common data layer where our data resides. This data layer is shared across products like Databricks, Snowflake, and various AWS offerings. It sounds quite extensive.

When we look into the details, Lukas, considering your previous experience, your product portfolio wasn’t that complex, with fewer use cases and a narrower client base. You cover the entire energy sector, which is a much bigger challenge. So how do you approach this with your team and from a product management perspective? How do you prioritize?

As Rich mentioned, the whole data and process environment is very complex. One of our main goals is to help new users get up to speed quickly — that’s essentially the mission of our team. For example, we introduced a homepage as a central landing point where users can find templates to help them get started faster. We also have dedicated consultants to guide new users, but we don’t want anyone to feel dependent on them. When designing new features, we always ask ourselves: „If I were brand new to this, what would help me feel confident?“ That’s why we streamline workflows, add helpful tooltips linked to the glossary, and make features more intuitive.

For example, we recently completely revamped the filtering in Prism from scratch to improve intuitiveness.

If we talk about the homepage — when I open the app or web browser and land on the homepage — how personalized is it? Is it like a Google homepage?

Not really. The idea behind it is to show recommended templates that users can use once they start creating a dashboard, so they don’t have to start from scratch. At the same time…

Could you name a few of these recommended templates? What do users typically use?

For example, rig analytics, forecast studios, things like that. Users can also see their saved templates, so anytime they need to access something frequently, they can get back to it quickly.

What Lukas described really captures the Prism experience. Across the board, we have a common authentication platform for our customers, enabling single sign-on. They log in once and can access Prism, our EnergyLink product, or our Power Renewables products seamlessly.

If you visit our website, inveris.com, and click “login,” users with credentials land on a gallery that displays different products and solutions available to them. From there, they can choose which product to enter. For example, if I’m an AP clerk, I’ll access our open invoice platform. If I’m a geologist, I’ll probably go to…

Pokud chcete, mohu pomoci s dokončením poslední věty či s dalšími úpravami.

Zde je opravený text s vylepšenou gramatikou a stylistikou:

Prism: So Lukáš, how much domain expertise in oil rigging do you need to prepare a great template for rig professionals?

Fortunately, creating these templates isn’t specifically my job. Our team’s role is to make them accessible. We have dedicated experts in the area who prepare these templates. That said, having domain experience with our products is pretty important.

As we’ll discuss later regarding AI, this is where we’re really trying to help our customers navigate our applications more efficiently and quickly, with less need to understand some of the complex domain tasks. This is where we’re currently leveraging and will continue to leverage generative AI to assist them.

Instead of having to write queries manually to retrieve the information they need, customers can ask questions in natural language. When a question is asked, our system generates the necessary queries behind the scenes to pull the relevant data.

Interviewer: Is the generated query written in natural language or something like SQL? Or does it depend on the product?

It depends on the product, but most likely our generative AI solution creates SQL statements to extract data from SQL databases.

When looking at the architecture or the distribution of roles, I see the product software engineering side focused on many things, including numerous machine learning models and prediction algorithms.

Before we dive into generative AI, how much of that involves data science?

That’s a great question. We have been focusing quite a bit on software engineering, but we also have a large data science team. I actually lead an engineering team, but I also oversee our corporate data science team, corporate generative AI team, and corporate research and innovation team.

These three corporate teams work closely together. The data science, research and innovation, and generative AI teams collaborate on various solutions.

Our data science team has created many of our models and plays a very important role in the business. Traditionally, they’ve developed machine learning models, and we continue to leverage ML for analytics and insights on our platforms.

We are also transitioning more toward generative AI, but we aim to maintain a balance because there remains an important place for traditional ML and AI models.

From my perspective, not only do you cover many use cases and the whole value chain with multiple products, but you also manage the data layer: having all the data, insights, predictions — effectively the intelligence.

Pokud chcete, mohu vám text také přeložit do češtiny, případně upravit pro konkrétní styl či tonalitu.

Here is the corrected and polished version of your text:

Machine learning models, stuff like this, run through the back end of the application to the front end and user experience. So, how do you look into the ecosystem? What’s the relationship to other data tools and data stack tools? How do you see them? How important is it for you and your clients to use the front end and the software side?

Yeah, I think our front end really provides insight into their data. Because the data is so complex, trying to work with it on their own is difficult. We have a wide variety of data that we can pull together, and we allow our customers to bring in some of their external data sources and merge them with our data, viewing everything within our application. Lukasz and his team spend a lot of time building very rich experiences for our customers to help them get to the insights they’re looking for.

So that’s where domain expertise and client-specific knowledge play a role?

Exactly. I can take my drill data and put it into my BI software, but I would actually miss the insights and industry-specific details because I don’t know what I’m looking for—that’s what PRISM does.

Yes, and our product team is amazing. They work very closely with our customers. We’re not building these products in a silo. We meet with clients almost every day, showing them prototypes and developments. We engage with them, ask for their input. When we release products, we often do so as market tests, allowing customers to start using them, playing with the features, and providing feedback through the application. This way, we can look at analytics and metrics and decide, “Oh, this is going well, let’s focus on it and expand it.” If something isn’t going well, we reach back out to customers to understand why.

When Rich mentioned metrics, I actually like that we have recently been moving towards a more data-driven approach. For example, we identified success events that customers perform within the product and measure how many they complete over time. This helps us gauge how successful we are in delivering new features.

Would a success event be something like building my own dashboard?

Specifically for PRISM, it’s centered around workbooks.

What is a workbook?

It’s the space where customers create their dashboards and save them—you can think of it as a project within PRISM.

Instead of visualizations?

Yes, exactly. Events like saving, sharing, opening, and exporting workbooks bring the most value to customers.

What did it mean to implement such a data-driven approach to product development? It sounds great, but I assume it was hard work.

We can track the product engagement score over time and retrospectively assess how successful efforts were in improving it when implemented.

Let me know if you want the rest of the text corrected as well!

Opravený text:

Implementing new features. So, we can prioritize based on our assumptions about how this can raise these metrics, and then we can validate those assumptions. So that's…

The goal there, like PES, is what you're trying to raise. That's the main goal — that’s like the KPI or OKR of your team. So how do you do that? Give us some insight into what you're working on right now or what you have been working on for the past six months. What are the challenges? As I mentioned previously, we were working on the filtering system, which was quite complicated. This way we optimized how fast the customer can get to the dashboard view they need to see. And actually, this also taught us a lot along the way. We were collecting feedback from customers. We use AppCues, which is another tool for surveys and collecting satisfaction ratings from customers. We also learned that we need to ask better questions, because at the start, we put a simple prompt saying, “How do you like the new filtering?” and we got a few answers such as, “I hate it,” because some customers simply don’t like change. So first, we learned that we need to ask better questions to get better feedback. And second, we need to teach them how to use the new features and where the value actually is.

So, what's the better question then? What’s a good question? I think, in general, it could be something like, “Tell us about your experience with the new filtering.” Asking more open-ended questions and balancing detailed surveys that customers usually skip with simple questions that lead to yes or no answers.

So, do you do qualitative research as well?
We do. We do that through regular calls with customers when we ask about their experience. We usually let them talk rather than preparing a detailed script. We simply ask, “What keeps you up at night?” and they start talking. Then we can go into detail based on what they are struggling with.

Yeah, it sounds a lot like a manual for a first date, actually: ask open questions, listen, let them speak.

So, having space and opportunity to work on such a large scale and important product, what are the lessons learned you would like every product manager or somebody building their own product to implement in their development?
I'd say one of these would be flexibility in setting up the right processes — to see what’s working and what’s not, and to be able to accommodate that instead of blindly following the latest trends, such as…

What is the latest trend that doesn’t work from your perspective?
I'd say, for us, it was the Shape Up process that the rest of the company followed at the time we started, because we were the newest team and had to deliver UX improvements really fast. We realized that Kanban would work better for us, but then we needed better planning. This is where the shaping part is actually really strong from Shape Up. So we borrowed that — we groom every bigger epic with the team before it goes to development. We do backlog triage regularly in a small…

Here is the corrected and cleaned-up version of your text for clarity and grammar:

We were working in a group, and we also had to go back to story points because we needed some high-level estimates to create roadmaps for stakeholders. What are story points? Story points are used for estimating tasks; it actually comes from Agile.

Okay, so you have it all?
Yes, we made a hybrid, a bespoke solution for our team.

Exactly, it's a mix of a bit of everything, but we tend to pick the best parts from each approach to fulfill our current needs instead of quickly following one standard doctrine, I would say.

Yeah, so what about you, Rich? What kind of frameworks do you like?

I find it amazing that a small team in Brno can design their own internal processes, workflow, and prioritization scheme. From my experience with global companies, it's usually „one size fits all.“ Like, yeah, you have to use this framework no matter what—whether you’re the receptionist or the data scientist, you would use Agile because we run on this platform.

No, I don't believe that's accurate as a leader in the business. What we try to do is empower our teams. We break our teams into different responsibilities, and they work on different tasks. We want them to come together, usually sitting down when team formation starts, to decide how they want to structure their workflow. They also work closely with the product team on that.

There’s a lot of forming early on in the process—do they want to use Agile? Scrum? Kanban? It’s really about empowering the team to work together to decide how they want to organize. That’s what we as leadership want them to do because we focus more on results.

Results, exactly. So, do you manage results by KPIs or predictions like OKRs?

Yeah, we actually have goals that each team sets. For example, if they have a sprint, they’ll set sprint goals. Usually, there’s some type of goal set for the team as far as what they need to accomplish, and that goal is set by the team working on the tasks. Then we have KPIs to track those goals.

What are your KPIs, Lukáš? Is it PES or developing a new feature?

For products, it would be the product engagement score we talked about. But for each feature, we tend to identify some sub-goals that contribute to the main goal, so it’s a hierarchy.

Such as?

For example, measurements can include usage of the new feature, adoption rates, or NPS for a specific part.

So solving questions in product development, making customers happy and successful—it’s still like the view layer, right? They see their own data.

From what I’ve heard from Rich and also you, Lukáš, there’s a lot of data from really different sources. How do you tackle that? Like, how do you manage data?

Let me know if you want me to continue or adjust the text further!

Here is the corrected version of your text:

Management must be a really big topic in Envers, isn't it?
Yes, it is. You know, across all of our different products, I think data management is complex and one of the most important tasks that we have in order to be able to show our customers the different analytics and insights that we provide. We have to tackle the data, and this is massive data—we're talking petabytes of data from different systems bringing in new information every day. Well, maybe not every single day exactly, but quite frequently.

All these different data sources come in many different formats and schemas, so we have to handle all that. A lot of it is dirty data. Some of our customers' data is very well-formed, but other data we receive isn't so well-structured and has issues with the data's quality. So, we have to deal with all of it.

We can't just push back and say, „Hey, no, we can't accept this data because your data quality isn't great.“ We need to figure out a way to accept that data, work with it, bring it in, ingest it, and then provide value to our customers. And this all has to be done in real time at scale.

Also, one of the challenges is the metric system. For example, in the US, we use feet, which is completely different from the metric system. Since we have many European and Canadian customers, they would like to use meters. This is one of the challenges we are currently facing: how to handle that transition. We are considering recalculating measurements on the fly or creating duplicates of data. We haven’t yet decided which approach is the best from a technical standpoint, but this is one of the recent challenges we are dealing with.

Yeah, I wouldn’t have thought of that actually.
But yeah, we crashed on Mars because of it.
That's right.

But back to the data — you said clients are ingesting their own data into the platform as well?
Yes, they are.

Are they allowed to? Or do you have a marketplace or some kind of database of your data available from day one, like energy data that you connect?
Sure, yeah.

Can I use Envers without ingesting my own data?
You can. Different parts of Envers can absolutely be used without ingesting your own data. We have a lot of data from government sources, such as the Railroad Commission. So, there is a lot of non-proprietary data in our system that customers can access and leverage.

Looking at the data you use in energy, what is the most typical kind of data?
You have geographical data, well information — all the details related to a well — production data showing how the well is producing, land documents, lease documents, and revenue information showing past revenue. So, it’s essentially production data and related metadata.

Let me know if you’d like me to further polish or format it for a specific use!

Here is the corrected version of your text with improvements in grammar, punctuation, and clarity:

I guess, as well. So, we have, you know, mountains of data across the energy domain that we’ve got in our system. If you look at how digitalized it is from the perspective of data sources, how often is this data stored in Excel sheets? That’s where a lot of our data comes from—Excel sheets. Of course, we also work with data outside of Excel, but you know, Excel, text documents—we’re dealing with tabular formats like CSVs. Those are a lot of the data sources we work with. We also handle many documents that we OCR, transcribe, and extract data from, bringing it into our system. So if you can think of it, that’s probably how we’re getting our data into the system.

From what I understand, Anvers is actually the one software you need to run such a company, correct? Because you have it all—everything from production to invoicing.

That’s correct.

Okay, so let’s go to the elephant in the room. I told you not to talk about it earlier because I thought it would be an interesting part. Being in business so long, having deep domain knowledge, gaining customers’ trust, and working with these data sources—plus having data science and so many models—how does generative AI and large language models fit into all this? What has changed for Anvers in data science and product development in the last two years?

Yes, so as we discussed earlier, we’ve been working with AI for quite some time—traditional AI and ML models. Transitioning to generative AI wasn’t a huge leap. We wanted to be at the forefront; we jumped in early. Our data science and research & innovation teams reviewed the different large language models available, while our product team explored various use cases.

Our chief product officer considers generative AI one of the most transformative innovations since the invention of the fire. That’s a big statement, but a true one. It’s changing the world as we have seen in the last three years. When something that impactful comes along, you can’t ignore it. We didn’t ignore it. We have fantastic leaders who understood its importance and wanted us to engage with it early.

We’ve been working with generative AI for the last three years, and we continue to enhance our capabilities. But we’re not just using it to build new products; we’re also using it to help develop existing products, assist with sales, and support product management. Every department within Anvers uses AI in some way to build and sell products, and more. So we’re leveraging generative AI not only within our products but also internally to improve product development.

That’s what I’m trying to say.

Yeah, I think it’s a must. But let’s focus on the product side. So if I’m an Anvers customer now, like we are talking—

Let me know if you want me to continue!

Here is the corrected version of your text:

I've been working with LLMs and their potential for the last two years, at least three. But getting them into production — to have them actually be a useful part of a product that works and brings value — that's the hard part.

So, if we look at your suite of products, and the products themselves, where can we find generative AI?

Right now, within our products, one of the first things we did, as I mentioned, was to form a team to research LLMs. As you know, we're a data company, and one of our main areas of focus is intelligence. We produce a ton of intelligent documents daily. We have many customers who rely on these intelligence documents, which help finance companies determine M&A activities. They use the documents to make critical decisions about billion-dollar deals, such as whether to acquire certain assets or not.

For example, we provide executive summaries and research that analyze all the financials and recommend proper strategies and prices for acquisitions. We publish these documents in PDFs to our customers. These are large documents with a lot of information to get through.

One of the first products we focused on was providing access to this intelligence data, allowing users to ask questions about it. We built a pipeline to pull the data into our system, which at the time used a RAG-based system. This was presented through Prism or another UI, enabling customers to ask questions like, „How are tariffs impacting the price of oil?“ The system would answer based not on generic internet data, but on the proprietary data created by our intelligence team.

How was the experience?

Before LLMs, customers would have had to sift through many PDF documents manually to find that information.

Yes, the research papers were from your team, and now I can query them specifically with my own tailored questions. Really interesting. You keep amazing me. In all the propositions you've talked about, you also have intelligence and publish research papers.

Yes, publishing research documents is a huge part of our company.

Really interesting. So that was the first step — quite logical since these are written documents. You used RAG to retrieve and ensure the accuracy of the information. So what’s next?

Initially, yes, we used the RAG architecture for that. As we matured and learned more about the generative AI market, agents became a big thing. We actually built a multi-agent framework that we're now using to provide products like that. We're also leveraging many of the learnings, understandings, and capabilities we've developed from that.

Let me know if you need further refinement or a more formal or casual tone.

Tady je opravená a stylisticky upravená verze textu:

This was our first project, and now we’re expanding it to other parts of our business. It started with earnings, but now we’re applying it to earnings statements as they come out. We can also use it on contracts, allowing users to ask questions about contract content. There are various use cases, all evolving from that initial use case we developed.

Our goal is not only to generate new products but also to enhance our existing ones. That’s why we integrated this capability into our flagship PRISM product, specifically the Instant Analyst feature. This tool lets you ask a question within PRISM, then it identifies the relevant data sources to pull the necessary information. After retrieving the data, it associates it with appropriate workbooks. So, when the customer receives the answer, they also get visualizations to help them dive deeper and analyze it in the environment they’re familiar with.

That’s really interesting. I saw a demo of PRISM where the assistant actually performed queries, figured out which data was needed, and combined everything. Lukáš, what’s your take on this? You just finished the new filtering functionality. When do you think filtering will become obsolete if users can just ask questions in natural language, without having to click through categories or browse through a tree structure? How do you see your team’s role in the GenAI transformation?

Lukáš: Actually, I don’t think AI will replace filtering completely. Rather, it will serve as a complementary way to use the product. For some users, letting go of manual filtering might be uncomfortable because the AI remains a bit of a black box—they don’t know what’s happening behind the scenes. We’ve found that some users still prefer to set filters manually, especially for simple tasks where clicking a button is faster than typing a question. But for more complex use cases, AI can be very powerful. We’ve actually adopted the concept of agents, as Rich mentioned.

All of these agents are accessible through PRISM. One is called Intelligence, which acts as a virtual assistant to curated analytical content. Another is PRISM itself, serving as an assistant for the platform, helping users quickly answer any questions about data. We also have a specialized agent called Earnings, which summarizes earnings calls from public oil and gas companies, covering financial performance, operational updates, and strategic plans. Our customers use this to track competitors, validate assumptions, and stay ahead of market trends. As you said, this is indispensable — and I really appreciate how we approached it because…

Dejte vědět, pokud chcete, abych text ještě nějak upravil nebo přeložil!

Zde je opravený text s lepší strukturou a plynulostí:

So we started by identifying real challenges first. At the end of the day, every customer who opens the application is trying to answer a question, right? Sometimes it can be as generic as “How are my resources performing?” but other times it’s very specific, like “Where do I need to add an extra casing string in Texas?”

To help with that, we introduced the idea of multiple agents, each focused on a specific part of the workflow. Looking to the future, what’s ahead? You have the baseline — the agentic platform or agentic system. I suppose more AI agents will be added to the platform, covering more use cases. What is your strategic plan regarding this?

I can speak to that. One thing we’ve learned along the way is that as the number of agents grows, some users can get confused about which agent to use for what. To simplify this, we are now building a master agent — a single starting point that automatically connects you with the right agent to answer your question. This is one of our current projects.

At the same time, we’re developing new agents that will be introduced in the future. Think about all our different products — each will have multiple agents to help customers answer their questions, respond to requests, and improve workflows. Our focus is really on building additional functionality within our products, leveraging Generative AI to solve problems faster and help customers get the answers they need more quickly.

My expectation is that generative AI will become a natural part of every project developers work on. They will integrate generative AI elements into their stories in a way that improves the results of the tasks they’re performing — much like how connectors and APIs are used today.

Exactly. That’s a good way to describe it — another layer to consolidate insights, data, and use cases under one roof. Generative AI and agents offer a technical solution for this future vision.

Right. And as Lukasz mentioned, having an agent that can observe across all different platforms and products, understand the question being asked, and know where to find the answer — that’s really exciting. Of course, it comes with challenges, especially when managing such a broad portfolio with so much data and so many offerings. Generative AI technology might solve many of these problems. Good luck with that!

For the last part of this podcast, I’d love to ask you again about Brno, to return to our local relevance. You mentioned that you’re really fond of Brno and that you were among the first people to get started there.

Pokud chcete, mohu vám také pomoci s překladem do češtiny nebo s úpravou stylu podle cílové skupiny.

Here is the corrected version of your text:

We edited it and hired the first guys. You said that right now you have 110 people, growing to 200, possibly more, by the end of the year, which seems like really rapid growth. And once again, there are challenges and a lot of work to be done. Is there anything else specific to Brno and how it works that you would like to mention as the closing remarks of this interview? Because from my perspective, it’s really interesting what I found out about how you’re doing business and work in Brno. It seems really special and unique in a way.

Yes, and I think we probably both have some comments on that. One of the things we wanted to do when we set up Brno was to really establish the culture. We felt it was important that Brno wasn’t looked at as an offshore development or manufacturing location. It’s really hard and solid part of Enveris, just like any other office within Enveris. So it was important for the team here to feel like we’re not just throwing over the work from North America that we didn’t want to work on and just handing that off to Brno. That’s not the case at all. They’re working on all the new products we’re developing, all the new initiatives. They’re involved in the generative AI work. They’re involved in the latest and greatest work we’re doing on Prism, building out new products there within our Source of the Pay platform. So it was really important for them to feel as much a part of Enveris as anyone else at Enveris, as well as to have the culture that’s important to them here in Czechia.

So what’s the culture like? What’s it like to work for a global company out of Brno?

Yeah, I mean, I can confirm what Rich just said. You don’t feel like a second-class citizen just reporting to the US. There is a strong sense of ownership. We have a lot of freedom in prioritization, etc. At the same time, we also need to synchronize regularly with the US teams, of course. But on a daily basis, one thing I really appreciate is the flexibility, because we can work from anywhere. Some people work fully remotely from home. Others like to come into the office. And we actually have a really nice space in Brno. It’s open to everyone. I personally enjoy seeing people face-to-face, so I usually go in a few times a week. But it’s totally up to each person. And I personally enjoy working with my team. I’d say they are some of the smartest people I’ve ever worked with. They’re solving quite complex problems, and everyone’s actively working to improve things. Honestly, that’s not something you find everywhere, at least from my experience. There is also a strong sense of community. We have team-building events. Sometimes we just go out for a beer to chat about life. And we also get four volunteering days each year. Actually, in a few days we’re heading to a dog shelter near Brno to help out for the day. So, yeah, I know this might sound like something I’m supposed to say because I’m representing the company, but I really mean it when I say that I’m hap—

If you need me to continue or polish this further, just let me know!

Here is the corrected and polished version of your text:

Interviewer: Why did you choose Envers? Are there any lessons learned you would like to share, something that our listeners can implement in their companies? Or maybe some rituals or parts of company culture you find really beneficial?

Interviewee: In general, better communication. When you have such large teams, you need to clearly communicate the vision from top to bottom, so everybody is aware of their role and how they contribute. And, you know, better synchronization within teams.

Interviewer: Do teams need to collaborate?

Interviewee: I was going to say, I’m lucky enough to travel to a number of different offices, and traveling over here is always special. The vibe and excitement in our office is something hard to find elsewhere. So it really is a special place. To make it special, we had to build not just a software engineering office, but a full technology stack. We have teams working on infrastructure, back-end databases, security, product management, software engineering, and data science. So, everything you would see in a North American office, you also find in the Brno office.

Interviewer: So these are the roles you are looking to fill in Brno?

Interviewee: Yes, all over the technology spectrum. And as we grow, obviously, we need more recruiters, more HR team members, and eventually, other parts of the business might also move here to support the office as needed.

Interviewer: Great to hear that! Thanks for the compliments. Being a small European country, it’s always pleasing to receive compliments about our culture and nature. But most of all, thank you for sharing your experience and teaching us about the current energy sector and the wide range of products, propositions, and work Envers is doing in that sector. Good luck with implementing GenAI and making it the glue that binds the whole platform together. And good luck with hiring such great people like Lukáš in Brno. Thank you for coming.

Interviewee: Thank you. Thank you for giving us this outlet to share our story.

If you'd like, I can also help format this for a transcript style or another specific purpose.