Chuck Yates Got A Job

0:19 Hello everybody welcome to ProFract. This is so cool out here. All right guys as an ex-finance bro they would never let me out in the field because the finance guy would always walk up push a button

0:33 hey what does this do and I'd mess something up so this has been cool that y'all actually have me out here. Tell me what you two guys are doing together. I got the demo in there and thought it was

0:43 really cool. Matt what's up? Yeah well at profrack probably seven or eight years ago we started working on frack automation and you know we put pro pilot 10 out about two years ago and you know

1:00 eventually evolved into pro pilot 20 where it's fully automated and what we've been looking for is a partner for real-time data. We want to know more about what's going on down hole and looking at

1:15 you know, what is some real-time data that we can collaborate with and provide automated responses and really take completion designs forward from where, for the last, you know, since the Shell

1:29 Revolution started, it's been a cookie-cutter approach where every stage is the same, very little variability from one stage to the next And with FRAC automation, you've got the ability to adapt in

1:42 real-time and make real-time changes that are predefined based on the information and the real-time data that you're collecting. But at ProFRAC, we don't collect as much real-time data outside of

1:58 our pump performance, pump rate pressures, but what's unique about Seismost, and I'll let Pano talk about it

2:07 here, you know, he knows it so much better than I do, but what you can see from a uniformity index, from perf efficiency. And how good is the frack? How well is each perf cluster taking fluid

2:20 and sand? And how much better can it be if we specifically target these parameters and make real-time changes to deliver this? And more importantly, how do we give the tools to the customers so

2:35 that they can define it and predefine what they want to happen rather than coming to them and saying, We know this better than you do. Like, we've got it We'll fix your problems.

2:48 That's not what we are. We want these tools to be fully utilized by our customers so that they can come in and push completion design to the next level. And I think that capability now exists

3:04 and it hasn't been something that was available previously. I mean, I'm old enough to remember you'd sand down on a fracked two days to get the report to find out what happened. Are you telling me

3:18 that doesn't happen anymore? We can do stuff in real time. Yeah, good point. So let me say a couple of words about seismos. So what we bring to the mix, so we are a technology company, focused

3:29 on acoustic sensing. And essentially, we enable real time data for the subsurface and for the fact business We always had this vision of pumping being automated, of machines being self learning,

3:49 and essentially being in an adaptive mode and constantly learning more and more as a job goes.

3:57 The partnership that we announced has to do with closed loop fracturing is how we call it. And essentially, it is correcting the pilot docs. We had surface-based companies, the pumping companies,

4:09 performing a subsurface task that was not even being measured. So essentially, we're closing the loop by enabling real-time data feedback from the subsurface. And it integrates smoothly with all

4:24 the FRAC automation that pro-FRAC has enabled on the surface side. And what a closed loop means is we have now a situation where a pumping company is performing its task, with all the automation it

4:37 comes along with, you can see in real-time the performance of the FRAC, that data feeds a completion logic, and that logic, which can be pre-configured by the user, as always, essentially

4:51 dictates some of the changes that essentially have to be done on the pumping side so that you can optimize performance. So now that I can do things in real-time with real data, do we have enough

5:05 evidence to know, am I saving money on the FRAC? am I increasing performance of the reserves post-frack, or am I doing both? So, most of the work that we've done for over the last two years with

5:21 Frack Automation has been really focusing on efficiencies, utilization, and cost structure. Let's optimize the equipment for best performance. Let's get these components to last longer. But now

5:34 that we've got this partnership with Sizemos, now we have the ability to go in and target well performance. And that's, for me, that's where it gets really exciting. I think, you know, why do

5:47 you innovate? Why do you build stuff? And I think the big reason is, like, my customers really don't care how much it costs me or how much my expenses are or how much it saves me. They want to

6:01 know, okay, why do I need this? What does this do for me? And I think the answer is we can go and

6:10 work with our customers to advance completion design to get better results. And, you know, this is everything from uniformity index, per efficiency, as well as, you know, getting real-time data

6:23 from like disposal wells and, you know, allowing for us to delineate the third bone spring.

6:32 There's a lot of risks with these additional benches. And if you've got real-time data and the ability to act on that real-time data, we can start opening up new benches and helping operators

6:44 delineate these formations. So as far as an uplift and well results, I don't want to get in and make promises, but I'll tell you this, you give the customer and their engineers better tools, they

6:57 will figure it out. And that's our goal here is let's give them better tools, better execution, and work with them on how to maximize the result.

7:10 Yeah, if I may add a point or two So I think another way you put your question is. That's real-time data matter when it comes to savings or production. And there's been sufficient evidence in the

7:20 industry. Hess, for example, published a paper of the recent Eurotech conference where they show that if you can monitor fluid distribution and just marginally improve it, let's say by 10, you

7:31 get an extra NPV of 300, 000 per well. So by all means, knowing how many clusters are taking fluid, by all means it does correlate to production. And that's one thing How you approach the

7:43 situation, of course, may vary according to the customer strategy. You may have a stage, for example, where 50 of the clusters are taking fluid. What are you going to do? Are you going to fight

7:52 it with hoping that you're going to add some more clusters? Are you going to cut volumes a little bit? I mean, you don't want to over capitalize and essentially inject the same design volume. But

8:05 to 50 of the clusters is going to lead to problems and frack hits. And that would leave to significant savings, like just saving, for example, 10 of volume. At a given, a frack job is going to

8:15 save you100,

8:18 000, ramping up on some faster and save you another56, 000. Adjusting friction reducer can save you another20, 000. So depending on the strategy of the operator, you can have both benefits on

8:30 the production side, but also on the cost-saving side. And I think what we are enabling now with the closed loop fracturing is for the first time you have visibility on the millisecond. So you can

8:38 see that during the stage. And then it's on you how you act on them how you act on the data and which strategy you're gonna follow. Interesting, 'cause one of the things I always thought we had an

8:49 issue with is completions of reservoir would be in their own silos and not talk to each other. And a lot of times this was your analogy, completion recipe. And you'd come over here and you'd

9:04 optimize it for cost and you know, I saved200, 000 on the frack And reservoir is going, but we needed that much water. to get the reserves and stuff. And so, is there an ability for potentially

9:20 the reservoir person to be working in real time along with the completion people? Yeah, I believe so. I think one of

9:32 the things is when you get prescriptive on a per lateral foot standpoint, you're adding in complexity. You're adding in a lot of variability from one stage to the next, and you're opening up

9:46 opportunities for risk, for failure, for issues. And

9:53 so when you come in and you start introducing this complexity, it usually comes at the cost of efficiency. Efficiency is that the industry is fought so hard for. And that's where automation and

10:06 real time data and a closed loop

10:11 usability of that data

10:13 allows you to have that complexity without sacrificing the efficiencies that the industry has bought so hard for.

10:21 If I may add to this,

10:24 I think you do have those engineering silos, we still have them today, and we're doing our best to bring them together. But I think as we're moving to tier two acreage, this is no more just an

10:33 engineer's discussion. The CFOs come in, the COs come in, you know, the whole strategy needs to change a little bit. Whereas optimization before was not necessarily the first priority thing. As

10:47 we're moving to that tier two acreage, it's no more an engineer's decision. That's what I'm saying. It's not forgiven anymore. So really being able to have real-time data and acting on it is

10:57 really becoming a C-level now type of decision. And that's the approach we're taking with the closed loop fraction So Panos, when I was getting the demo in the in the data van there, somebody

11:10 brought up the concept. of an independent audit, what does that mean? Yeah, yeah, that, first of all, I need to congratulate ProFrak for being open to the idea of an audit, 'cause despite the

11:22 partnership, we as a company are in the measurements business, was to think of it. So regardless of how these guys perform, we're gonna read what the performance is, we're gonna benchmark it,

11:34 we're gonna provide a number, and maybe good, and maybe less good

11:39 So I think Kudos to ProFrag being open in that level of transparency, 'cause for the first time now, you have a pumping company that is great enough to say, you know what? I can do my work, I'm

11:50 confident enough, and I can be measured. And I'm okay to situations where I'm not perfect, but at least, you know, through that loop, I can make an adjustment, and I can correct for that

12:01 imperfection.

12:04 Matt, the other thing I heard When I was in the data van, as I heard. supervised and unsupervised closed loop. What are those two concepts? Yeah, so unsupervised is, you know, we take the

12:21 completion schedule and then we also have the dynamic configuration so that whatever feedback we see while we're fracking from and whatever real-time data we see, we execute a set of orders and we

12:36 start executing based on what the customer is configured and what they've required of us. And so there's a, before a job starts, we go in, we work with them, we program everything, we walk

12:48 through it. There's a customer configuration tool that allows them to go in and pre-configure how they want us to respond based on what events or what data we see in real-time. Now, of course, on

13:04 the supervised side, Under no circumstances do we want to take control away from our customer. If our customer wants their consultant to have the ability to get in and control this and make

13:18 adjustments to it, they still have control over the fracks. They can still step in and make adjustments to for whatever reason, for any reason that they want. But with the pro pilot is we've got

13:32 full documentation of what was changed, why it was changed, and full visibility So in unsupervised, think of it like a Tesla with the full self-driving.

13:45 These cars will pretty much do the whole job for you. They'll drive you home. But if you need to step in and take the stairwell, you still can. So

14:01 we live in a world now of AI And my running joke is I'm old enough to remember when we call that statistics.

14:12 AI, I've told my kids that I've lived through free game changers, the internet, and the reach of the net to connect the world. I lived through the shale revolution which totally changed

14:27 geopolitics, the world, our economy, et cetera. And I think AI is actually gonna be bigger than those two put together. How are y'all using AI? How are you thinking about AI?

14:42 So I think when you go in and you start looking at automating a lot of your back office, I mean, I think everybody's working on that or should be working on it. And I think it's the amount of data

14:55 that you can process and the number of processes you can automate is just incredible. But what I'm really excited about is we collect so much data, it's nuts I know Sizemos is collecting 20, 000

15:11 data points. per second, is that right? And then on our side, we're collecting 4, 000 data points per pump per second. You know, we go up to Silicon Valley to try to raise money, and I tell

15:24 people out there, we use less than one percent of the data we collect as an industry, and it blows their minds. It's just so much data that like, you've got a team of engineers at your corporate

15:37 office and your maintenance department, and they're processing data, they're looking at reports, they're looking at all kinds of things from different angles. But you also don't want to rob your

15:47 operations team of like the tried and true operational experience that they have, and there's a lot of tribal knowledge in this industry, some of it's good, some of it's not. And I think by taking

15:58 a fresh look at it, processing as much data as you can, and what AI allows you to do is to get organized with the way that you process that data.

16:11 One of the things we've seen with the automation, we had a fleet in the Utica that was pumping 110 barrels a minute. We had 22 pumps in line for redundancy. We had an extra six pumps. The problem

16:23 was is that the way that

16:27 your pump operators are trained is they balance that load across all 22 pumps. And so pumping five barrels a minute with 22 pumps to get your 110 rate, they consumed 1500 gallons of diesel an hour.

16:42 But if you optimize it to 16 pumps, and you pump a little under seven barrels a minute, what you end up with is only consuming 1100 gallons of diesel. And you don't put engine hours on those other

16:56 six pumps. You don't put hours on your transmissions on your other six pumps. I'm not changing valves and seats or packing on the other six pumps. So it reduces the cost structure for us, And it

17:08 saves the customer. 400 gallons an hour, which on an annualized basis is about72 million a year, just because we ran 16 pumps at 7 barrels a minute instead of 22 at 5. And so this is challenging a

17:24 lot of the way that you train your pump operators. I can do a software update now and target specific outcomes on this equipment and run it a specific way because it saves the operator money And on

17:38 top of that, with the AI, we can come back in and do the proper analysis to actually prove it. Because there's a lot of noise, there's so much data, there's a lot of field tickets that operators

17:52 have to sort through. And the way that the industry calculates substitution is like, Well, how much diesel did we consume?

18:01 How much gas did we pay for? And it's the analysis on measuring this stuff in real time. just the industry's overloaded with data. We have too much data, and we don't have the ability to process

18:15 it all, and that's what's so great about AI, is it's changing the way that we think about all the data that we're collecting. If I may add a few points, and I'm gonna try and make it a little bit

18:26 more specific to the closed loop fracturing that we're offering. So these guys pump, then come seismos, we take the real-time subsurface measurements One key differentiation is those are

18:39 physics-based. This is really important, 'cause it's not no more something that is inferring on an outcome, it's a direct measurement of the outcome. Those direct measurements are feeding a

18:50 completion logic, and that's where AI comes in. 'Cause you now have some completion logic that says okay, I read this, cluster efficiency is 60. What should I do, right? Should I add the

19:01 friction reducer? Should I increase rate? Should I do this? And that's where the AI component comes. because as you track more stages, that specific acreage for that specific operator, taking

19:11 into account knowledge, maybe from other wells, you build knowledge, so you pass now, so the completion logic incorporates AI, pass feedback to the pumps, again, to the pro-track pumps, and

19:22 those are making an adjustment. But guess what? The whole system gets wiser and wiser and wiser. So we move to the next stage, we'll pump again, we take measurements, physics-based, really

19:31 important, big differentiation, and then it feeds again the AI-powered completion logic. But now it's more wise. It's one stage, wiser than before. You understand what I'm saying? And it keeps

19:41 moving on. And that's the whole purpose of an adaptive system. So we're training the machines of tomorrow. That's what I like saying. We have a pump that is getting wiser and wiser other pumps.

19:52 I make two points about AI, 'cause I'm spending a lot of time doing it. One, I think it's very important that you understand that the human being is still the subject matter expert. AI can

20:05 identify Correlation the human needs to tell you causation on stuff. I think that's really important I think the other thing that's really important and here's what I want to get your take on is We

20:20 have the tribal knowledge you talk about the old crusty guy and just put his hand on the machine and say well It's broken because of this He got there because of all the grunt work he did throughout

20:33 his career to build that tribal knowledge I worry with AI doing a lot of the work and the more front work How do we create the old crusty guy or gal that we need in the future get thoughts on that?

20:49 Yeah, that's an interesting question

20:53 There's gonna be both like if we think of the oil field of the future I mean we all kind of agree that it's navigating towards the state of autonomy almost humanless, but but we cannot take it to

21:05 that extreme There's always a need for the expert. Profrag has been very vocal. Like the power and the decision making needs to be left in the hands of the customer. So the way we've designed this

21:17 closed loop fracturing, whether it's supervised or unsupervised, the end decision stays with the customer. Even in the unsupervised mode, which kind of runs automatically, again, it's on the

21:29 customer 'cause it's the customer who comes and pre-configures the completion logic. And think of it as a real-time decision tree that says, If I read this, I'm gonna do this. If I read something

21:38 else, I'm gonna do something different. So by all means, it's controlled by the expert you're talking about. Yeah, and just to touch on that, I don't think AI is gonna replace, it's coming for

21:53 jobs. They say AI is gonna replace humans. I think humans with AI are gonna replace humans without AI.

22:03 When you look at the tribal knowledge, you need people who've seen failure modes before, that they understand that I know what this data says here, but there's a root cause that's being missed here.

22:20 Some of the data isn't gonna be given, there's no magic bullet, there's no silver bullet here. This is to help make use of as much data as you possibly can, but you're still gonna need guys that

22:33 are coming in and saying, Hey, I've seen this before.

22:37 I know that, I think with Elon, he talks about trying to optimize workflows that shouldn't exist. There's problems that come from symptoms that shouldn't exist.

22:51 Where the root cause happened way upstream, and I think that's where you're gonna need to travel knowledge, guys, to come in and say, Hey, I know that this is saying this, and we need to fix

23:02 that if that's how we're going to do it. but we shouldn't be doing it that way anyway. Interesting. So how did this still come together? What was kind of, is this a Y'all are on a dating site and

23:16 hook up, was it a first of a blind date? How did this happen? Yeah, so we've been working on automation for a long time. We've had pro pilot out in the field for a couple of years. And as we

23:29 looked at the development path, it was all about fixing our own needs and fixing costs and utilization. Again, how can we be better at what we're doing? But this whole time we've had, we've been

23:41 really looking, what can we do for the customer? What can we, besides going faster and charging less, what can we do for well performance? And so we knew that we needed to find a good data

23:53 partner to work with. And

23:58 Alan Smith made a recommendation to Panos, to us, like, hey, you guys need to talk. These guys are working on some amazing stuff. If you had the ability to act on that stuff in real time, I

24:11 think you would have a real differentiated offering that could substantially improve well results. This would be a powerhouse of a combination. And I think he's exactly right. Yeah, so by all

24:25 means, it was a common board member that enabled the introduction, but at the same time, we as a size was because we've been growing very fast, we realize we're at a time where one plus one can be

24:35 five. So we were seeking the right partner. And there's very specific things that we like to be pro-frag. And it's not random that we are collaborating with pro-frag. So we like, we're like their

24:48 attitude. We like that they're addressing the market, they're go-getters. We also like the fact that they're understanding that industry is moving towards a certain direction and they want to be

24:58 ahead of the industry and doing things. So, and most important. It's a thing we mentioned before, they're transparent. So they're open to the whole audit idea. Like, they're gonna frack, and

25:10 they're gonna be measured, and they're okay with that, and customer's gonna see this. So there's very specific things we like with ProFrak. So yes, it was enabled by a common acquaintance, but

25:20 it's actually a partnership that we were very much seeking for. So Panos, I'm gonna put you on the spot first for kind of our wrap-up question. Hopefully I haven't screwed up bad enough You'll have

25:32 me back in five years when we're here. What are we talking about in five years that maybe no one's thinking about in the audience? Particularly those guys root over there talking. But what are we

25:47 not talking about today that we're gonna be talking about in five years? Yeah, not at the cost of people, 'cause we discussed this a little bit. All I guess is claiming at the new energy mix,

25:58 it's fair share Is it 30, 50, 80, we can always debate on this. In that new energy mix, every shortage of energy, including only gas, has to be safe and in some state of full automation and

26:12 autonomy. So what I think is gonna change in a few years, you said five, it could be 10, is the so-called fully autonomous field. I don't think it's gonna come at the expense of people. There

26:23 won't be any people at the field, but there's still gonna be plenty of people in Houston. But that's the big change that I see happening. Yeah, I would be willing to go and say like, we'll be

26:33 talking about this in 10 years, that the industry still has 10 years of inventory left in 10 years because of the innovation that this industry is known for. We're always finding solutions. There's

26:46 benches that haven't been delineated because they've got technical challenges that they have to resolve before they can properly delineate it. And I think it's gonna be solutions utilizing real-time

26:57 data to solve real issues that allows you to do that And I think that - You know, what we've seen over the last three years since 2021 is productivity per foot and well degradation from inventory

27:12 quality has been coming down. But I think that with dynamic completion designs and real-time feedback and acting on that real-time feedback can turn the tide and actually start showing improved well

27:27 results. We saw a step change in well productivity when we went to slick water, when we went to in-basin sand and increased the amount of fluid and sand this pumped per foot. There's so many

27:39 innovations that have happened through the years. Going all the way back to the Barnet, we had a customer that stubbed the well out at 2, 500 foot instead of the 5, 000 foot lateral. And we

27:51 convinced them to let us go ahead and do the full frack that they were gonna do on the one mile lateral. And it ended up being the best well that they had. even though it was only 2, 500 foot. And

28:04 so when we came back in and tried to convince them to do a one mile lateral, they thought we were trying to screw them the whole time, but they actually did it. They let us come in and pump twice

28:15 as much sand, go from five stages to 10 stages on this lateral. And you saw this massive improvement in well results. And it set off

28:27 a massive change across the industry where it was chasing pounds per foot, fluid per foot. And I think we've reached the diminishing returns of what you can get by just pumping more fluid, more

28:37 sand. But when you started getting in and looking at what can you do in real time? And how let the formation tell us how different this stage needs to be than the last stage. And what does that do

28:54 for well performance over time? And if you can - if you can change a statistical base and into a commercial bench or a statistical bench into a commercial bench, how many more locations are there on

29:08 that bench? How much delineation can we actually do here? And I think there's so much opportunity that when you look and you hear all the noise about there not being enough inventory, it's nonsense.

29:23 Just like we did last year, the year before that, every decade, we will continue to innovate. And in 10 years, we're gonna be talking about how operators only have 10 years of inventory left.

29:35 All right, guys, did y'all see the last podcast we did?

29:40 Yes. You did. So Matt and Panos are gone now. What did they mess up?

29:49 You know, I think it's gonna be hard to follow up with that, but we're gonna give it our shot. There we go.

29:56 I wouldn't say they messed up a whole lot. I think they did a pretty good job. All right. Only because this is being recorded. All right, so tell me about the partnership. How do we measure,

30:09 let's get

30:11 more granular. How do we measure what is good,

30:17 what is bad?

30:20 How are we thinking about that? I mean, that's a great question. I think back to what Matt was talking about, how many data points we get, and how much data we have to analyze, it's enormous.

30:31 And so, we've been at this for a long time, and so we have a team of engineers that look at analyze and

30:39 assess the data points from screen out, screen out preventions to

30:46 the type of chemistry that we're pumping in the water quality. I mean, it's all a big deal and it's super complex So when we say it was good and what's bad, I mean it's

31:00 it's trying to filter out the bad and keep the good and use that to help train our systems. I mean, that's at least from a surface perspective from the pro-frag side. I think Steve will probably

31:08 have some additional comments on the down-hole, but.

31:13 Yeah, and in terms

31:16 of how do we define what's good versus bad, I think in a lot of ways, we're starting at ground zero. 'Cause a lot of what we're trying to do today is not something that's being done on a consistent

31:25 basis And so it's about raising the standards a bit. That's what this whole partnership is aimed to do, is how do we get away from looking at things from a price per stage perspective to focus on

31:39 outcomes? And so this whole thing's about

31:43 how do we get better outcomes out of the work that we're doing on location? So I'm an oil and gas company, I'm the client, and we're talking about this How much are we doing on the - front end in

31:57 terms of level setting expectations, walking through these things. Is it still early days and we're figuring that out? But what are those discussions look like?

32:12 I mean, from our side, we've already been on some work together already to where you already kind of tested the waters. There is an appetite for the automation plus the real time insights I think

32:24 the industry is still absorbing a lot of it. And so we're at the front and center of it. So you will see some building the bridge where it's going to take some time for it to have mass adoption.

32:34 And again, more case studies, the more success stories you have through that process, you're going to start to see that. What it's going to boil down to is I kind of led with that last answer,

32:46 but we're creating this new standard. And for those that aren't doing it, they're going to be left behind And so it's either going to be the odd man out

32:57 Yeah, there's a lot on the front end. There's a tremendous amount of work that we do. Whether it's analyzing the historical performance or putting together a completion's design, whether it's, I

33:11 think Matt alluded to it earlier, like why are we doing this? I mean, completion's designs have changed drastically since the Shell Revolution. And they've, whether it's more profit per foot or

33:24 fluid volume per foot, we've optimized it to a point where we are today.

33:30 And now we need to take a prescriptive approach. And so that prescriptive approach is very complex. And so what we can do is take all of our data that we have accumulated since inception of the

33:46 company, put all that together to help build the matrices for our customers to help make those decisions, not to make them for them, but to give them some. background of what we know and what we

33:57 seem, so then they can make better, more informed decisions with our tools. So I'm the client. We're gonna talk a lot on the front end. How does the discussion go between a supervised and an

34:12 unsupervised closed blue? What do I need to be thinking about pros, cons?

34:20 Good question. In both cases, the goal here is to leave the final decision to the customer, whether it's them making the decision on the spot versus them building some pre-determined, pre-defined

34:33 logic that follows a sequence of checks before it makes a decision. At the end of the day, it's putting the power within the customer for that workload. But to better define closed loops,

34:47 supervised versus unsupervised, I look at it a lot in the way of cars, so closed loop when you think of what is an unsupervised, what does that look like? You can kind of think like a self-driving

35:00 car. You know, there's sensors on that car that's kind of guiding you and telling you what to do or even in some cases with adaptive cruise control. You know, it's alerting you that you're about

35:09 to smash into the person in front of you, right? And then when we look at supervised, you can think of that as just regular cruise control. Like if you didn't take some action then you would

35:18 likely have an issue, whether it's braking, et cetera. And so you can think of frack in the same way.

35:26 Yeah, I think it depends on the customer and their comfort level and their knowledge of the tools that we have. And so the more we can educate and show and open up and give them the tools to make

35:40 the decisions, I mean, I think the more comfortable that they'll get or they will be. And so when we think about supervised versus unsupervised that maybe we start one way and we work our way into

35:51 the

35:53 unsupervised. Not really we, I guess, so back up. It's the customer who's building and designing it and making those decisions to get to that point and that level for their wells, Phil. And it's

36:06 a further add to that. There's two things that we're accomplishing here. One is we're taking the automation from the pro-frag side, which is gonna help with a lot of the consistency on the

36:16 execution part of it. And then it's we're taking the subsurface intelligence part And so instead of sitting in a frack van, looking at treating pressure and asking yourself, like, am I losing

36:29 perforations? Is my pipe friction higher than it needs to be? Having these measurements, having this insight to kind of help you guide you through that process. So I'm like, what's the right

36:40 knobs to turn? Is what we're after today. So going back to my days as a finance pro, we used to do early stage assets and Sargent. I talk about I've done 125 lease and drills in my career and we

36:59 would often drill the first horizontal well in accounting. Those are the kind of assets we like to do. And so one of the things that I found really important in doing those type projects was making

37:16 sure we measured something and isolated a variable one at a time when we went forward. Because often what we do is we have a company that would want to over optimize the second well and we got a

37:32 vastly different result. And you have no idea because of the three or four or five or 12 different variables you changed. We added more profit. We changed the sand type, et cetera. And so it was

37:47 incredibly important

37:50 literally just change a variable at a time so we could measure and monitor that. And so that's one of the things as we were talking about supervised versus unsupervised, I understand the importance

38:02 of real-time data, but at the same time being able to have a certain prescription followed so that I can measure it would seem to be pretty important.

38:14 Yeah, I mean,

38:17 that's really the reason that we've kind of come down to where we're at, as you know, prescriptive completions design.

38:24 As an engineer, you only want to change one variable so you can control the system as a whole and know and measure what's going on and turn the knobs the right ways so you can measure the results.

38:38 We don't always get to do that like you were talking about, but when we get to prescriptive, we can and we will and we are. you know, historically it's been okay, well here's our completions

38:49 design, here's our lateral, you know, it's not perfectly straight, it deviates, and so it changes, you know, and so we may not be able to get a great proficiency in a stage, and we need to be

39:02 able to make a call whether it's, you know,

39:07 reduce the amount of resources for that stage, or cut it off and have those resources to a different stage. So, you know, being able to measure and control the variables is key. Chuck, I think

39:19 you said the right word. It's a lot of variables. I don't think we have a real good handle as an industry of what's going on behind pipe. It's a difficult challenge to really nail down. And not to

39:29 mention, you know, as you transition from heel to toe, there's different rock mechanics going on. There's different things happening that are outside of our control. But if we can be surgical at

39:39 surface with how we execute, if we can be

39:44 really constrained, what's happening inside the World War. It puts us in a better position to at least get to the point where we can predict what the outcome is going to be. And I think that's what

39:52 we're after.

39:55 So walk me through maybe an anecdotal story or two on something we've seen in terms of saving cost andor increasing performance from the partner's show. Yeah, it can be something as simple as being

40:13 able to identify if you have a plug issue, like a plug failure. You know, where is that fluid going or are you over-capitalizing the stage that you previously treated? Even if you have good

40:23 isolation and you can tell that you're only treating, let's say, 50 of your clusters, do you still pump the same amount of fluid, the same amount of propant when you're only stimulating half that

40:32 interval? I mean, there's substantial cost savings that can come into play there. Whether it's taking the fluid that was designed for that stage and reallocating it another stage just performing

40:43 well. You can think of it that way. When it comes to some of the other variables that you mentioned, whether it's getting a better handle on pipe friction and per friction, look, there's a lot of

40:55 things that can happen even during the perforating process.

40:59 Not to touch too much on perforating, but that's kind of the wild, wild less on what goes into the perforating and how we get a handle on, or my holes being shot where I want them to, or they the

41:10 right size. How does that go into what I designed for? How does that impact it? The better handle that we can get on those things, the better chances of us having success.

41:23 You know, again, it's back to giving the customer a tool to make the right decisions so they can enhance, or make their wells more efficient. And so whether that be through, you know, a

41:36 perforficiency, downhole tools, you know, real-time water quality measurements to adjust chemistry, you know, real-time on the fly. you know, or like I alluded to earlier, reallocating

41:49 resources to either less or to a different stage. And you know, at the surface, we've got the pro pilot and, you know, on average, yeah, the past year we've seen it across our portfolio, on

42:03 average of seven minutes of stage savings, just on the automation of the equipment so we can get to rate faster and maintain the velocity throughout the stage. And so now with size most, we can

42:15 enhance that performance by, you know, bringing the loop closed, so. So do we have some items on the roadmap that we'll see next year, two years from now, three years from now? And, you know,

42:31 you're a publicly traded company, so go ahead and just lay it all out there and we'll deal with the SEC later. I mean, I think Matt and Pano, you know, I had some really good. responses to that

42:46 question,

42:49 it's difficult to see, when we're in the middle of it looking right at it, but we know that we're gonna continue to have technology and use technology to help advance and help our customers. And so,

43:04 what's down the road? I mean, we're right now looking at stage by stage, inter-stage efficiencies, and I think that's gonna continue to get better and smarter and faster as we use AI and machine

43:17 learning and build these AI algorithms to help us continue to innovate. So, I mean, technology's just gonna get, yeah, it's gonna continue to make us way more fisher. Look, I think when we look

43:32 back five years from now, you're gonna see that FRAC is gonna be very similar to how directional drilling operates now, to where it's fairly automated.

43:41 I talked about it previously, where we're trying to raise that standard. that's the ultimate goal here, you're gonna see that, whereas although our industry is fairly slow to adopt changes and

43:52 adopt new technology and adopt new ways,

43:56 you're gonna see that progression. As far as the old-and-guess goes, we're pretty resilient as an industry, and we've shown that we can innovate when we need to, whether it's commodity pricing,

44:08 that's pressure, et cetera. But I like to think of us as a cockroach. We always come out on top, we always find a way to survive. But that's what you can think, I think you can look forward to.

44:19 Yeah, no, it was interesting. One of the, I forget which formation it was, but we literally found out as we were drilling a 10 foot difference and where you landed the lateral led to 35

44:35 differences in EUR. And so, geo-steering became incredibly important and so being able to turn around and do that with fracks by interval is going to be pretty interesting. And to add to that, I

44:54 mean, we've been pretty fortunate with Tier 1 acreage. I think Palano had mentioned in his talk to where there is room for error with Tier 1 acreage. You can maybe do not so good of a job and still

45:05 have a pretty good well. But as we start to get into that fringe acreage, the room for error is going to be far less And for the economics for it to make sense is also going to be tighter. And so

45:16 what we're doing now is extremely important for the success of

45:20 that. You know, anytime I talk to folks outside the industry, they don't believe me when I say, Yeah, I don't know, we're getting 10 of the oil in place, Sal. I mean, they kind of have a view

45:33 of it will well as a bloom, you pop it, the air comes out I'm like, No, it's 10 best case. Does the partnership Potentially have ramifications for refracts. I mean, at some point, don't we

45:47 have to go back and frack every well we've ever fracked before?

45:52 Yeah, I mean, look, the possibilities are, you know, whatever we can dream up, and however we think about it.

45:59 You know, when you think about that of the 10, I mean, that's a staggering number, considering how much we produce in the country versus just that So, you know, the way that we get more and

46:13 enhance that is, you know, stage by stage, getting very prescriptive and intentional about the decisions we make. And now we're beginning to get, you know, basically get some eyes down whole to

46:27 give us, you know, an idea of what we're doing. And if we're, if what we're doing is making a difference.

46:35 Yeah, I mean, you touched on it If you look at the number of wells refried, you're over here the

46:40 last couple years, it's

46:43 and so I do think people are catching on to that. Sometimes I ask myself why more wells aren't refracked 'cause the economics are there. I'm sure there's reasons for that. But you talk about having

46:53 a whole separate problem. Like now you're going from shooting through one casing, through shooting through two. And so

46:60 the problem just becomes that much more difficult to make sure that you're doing a good job at it.

47:07 What is maybe a misperception that a client has that you would love to be able to convince them of? Interesting question. That's why I asked

47:21 it. I think everyone,

47:25 anytime someone's suggesting to do something different, obviously the first thing that comes to mind is what is the risk of doing the different thing? Like a lot of times what we're going to be

47:36 focused on is the negative versus the positive And so the biggest thing - I would like to see overcome from some of the customer standpoint is in order for us to continue to innovate and push forward,

47:46 we're gonna have to take risk at some time. And our job is to minimize, you know, what the impact of that negative risk could be.

47:57 Yeah, so, yeah, I think what our tools do, what ProPilot does, you know, with size most in the closed loop, you know, it's fully customizable to the customer because nobody knows the world's

48:11 better than that. And so, you know, we're here to help provide, yeah, smart tools to help make their decisions faster, quicker and more efficient.

48:25 I like it, go ahead. I'll go ahead and say Chuck, the other way to think about it is, you know, we're blessed as an industry to have a people with a lot of experience, but as you talk to some of

48:34 these people, a lot of times you'll hear them say, like I've forgotten more than you know, like that's a common thing that you'll hear people say. You had brought up in your discussion with Pano

48:41 and Matt about AI. One of the

48:46 things about AI is that it doesn't forget. And so it's able to use that data that learnings to continue to optimize the whole process. And so that's one thing I see AI coming into play as we talk

48:58 into how do we ease the burden on how customers think of what we're trying to accomplish.

49:06 One of the things I'd love for, and maybe we can do it right here on stage, I'll represent EMP world and y'all are the service companies. Is I do wish we could take more of a partnership approach

49:21 on things? Like you said, let's share some risk on this,

49:26 because we both have the goal of us doing better. 'Cause at the end of the day, if I drill better wells, I can afford to pay you more And I think too often we were way too adversarial when we had

49:40 power. We'd stick it to you guys on pricing. When you guys had power, pricing might be a little rough in our direction. And I do wish the service companies and oil and gas companies to sit down

49:55 and partner better. Am I just jaded and cynical? Or am I right about that? Yeah, I'd, look, at the end of the day, we are a service company, that's what we do, that's what we love. I've been

50:07 doing this for a long time and I'm super passionate about it Like, all the tools that we've built, you know, it's been in the making for a really long time and, you know, I mean, look, it's a

50:19 tool that we build to help our customers get better. Like, so, you know, in order to continue to get better, we need to be able to collaborate and partner with the ENPs to truly know if, or win

50:32 the knobs that we turn are a sectors and working the right ways. So, I think, you know, Those types of partnerships are a key, or especially here today and in the future.

50:48 No, I think you're absolutely right. Like Sizemos itself has a lot of its success to contribute to our relationship with Hess. They help kind of steer us in the right direction as far as where does

50:58 a product need to be? What does it need to be able to do in order to have value? And so those type of relationships are extremely important for the continuing to push the envelope on things like

51:11 that we're trying to do here with Brofract. So we're not going to always have all the answers. And so having their knowledge base too will certainly expedite that learning curve. So maybe to close

51:24 it out, I'm going to turn the table a little bit from the last podcast. The last podcast, I told you I was really worried about the loss of tribal knowledge and how do we create tribal knowledge,

51:42 if AI is doing a lot of the menial work that we all started our careers doing, I'm gonna say a different hypothesis this time and you take it any direction you wanna go with it. I'm gonna say the

51:56 loss of tribal knowledge is really good 'cause we're gonna get young, smart people with AI, looking at data with fresh eyes. And an appropriate answer to that is you're an idiot, Yates But what do

52:10 you think? I mean, I think there's definitely gonna be a transition of roles and I think they alluded to it earlier. Yeah, we've been, you know, the

52:20 tribal knowledge that we have, we've been taking and building into our pro-pilot. So we've been doing that for years, you know, collaborating with all the teams that know the engines and the

52:31 transmissions and the engineers and you name it. I don't think that tribal knowledge is ever gonna leave. I think it's gonna continue to grow It's just going to be in a different - aspects of what

52:42 we do, in order to run AI you have to go AI and you have to be able to create the prompts and run the systems and so we're going to continue to build and develop that and then that's going to become

52:54 tribal knowledge as we continue to innovate and add more technology to the stack.

52:59 I see the tribal knowledge as something that's still going to be there. The only thing that you'll remove is maybe the bias of the human. The human is always going to have some bias, whether it's a

53:10 bad experience or a good experience. It's always going to weigh heavily on how they decide things, but I would echo a lot of what Larry said as far as, you know, when I think of it from the

53:20 seismos perspective, more so we're going to steer AI in the direction of how do I free up more time so that my talented people can continue to be excellent and so take that busy work away, allow

53:34 them to spend more time on the things that are going to make us better I love it. I always say nothing can be more misleading than personal experience, and so do I need it. Guys, I think this was

53:48 great.