Subscribe
Share
Share
Embed
Welcome to Chuck Yates Got A Job with Chuck Yates. You've now found your dysfunctional life coach, the Investor Formerly known as Prominent Businessman Chuck Yates. What's not to learn from the self-proclaimed Galactic Viceroy, who was publicly canned from a prominent private equity firm, has had enough therapy to quote Brene Brown chapter and verse and spends most days embarrassing himself on Energy Finance Twitter as @Nimblephatty.
0:00 This talk here, it's going to be based on well-search and regulatory filing automation. Colin did an awesome demo earlier, so we're not going to demo again. But let's go ahead and introduce
0:11 ourselves. My name is Michael Cortez. I've been oil and gas for 20 years. I was mostly on the operation side. Frank Hans, he meant Han, Hal Burton, spent most of my career at BHP Billton when
0:24 they bought Petra Hawk and all those assets Spent the last five years with them offshore Trinidad, Tobago, Gulf of America. I've been in Tekka the last five years now, previously with Corva AI,
0:37 Director of DevCenter and Colin and Todd and Kineshius. They got me over here across the parking lot to collide.
0:49 My name is Buck. I guess I'm working on 17 years now. First 10 was with a major operator as a drill and supervisor superintendent, mostly. Mostly deep water, remote area, exploration wells,
1:02 which is important because now I'm in production. So I didn't actually deal with any completions until maybe two and a half, three years ago. After that, well control engineer. So blowouts,
1:14 fires,
1:16 some modeling, stuff like that.
1:20 Went back to consulting on the rig in the Permian Basin and now after a serendipitous conversation with our leader who asked me what I know about production and completions. I'm now a production
1:33 engineer, so.
1:36 Which I love 'cause when I met Buck and asked him about his past career and he told me he was a driller, never did production. I hope he feels a little bit of pain when his driller is drilled out of
1:49 his own now and crooked and his completion team, they set the Packers the wrong depth leave crabby fluid in the annulus. he gets to deal with all that now. I mean, you learn a whole lot, right?
2:02 You know, if you know anything about wells and picking up packages and assessing risk, probably the biggest existential risk that exists in the oil field in terms of black axle operations is
2:15 sustained casing pressures, sustained annulus pressures.
2:21 And that is, you know, you look at the old wells, you look at the new wells, the designs have changed That's all part of that evolution, right? How to mitigate that? So everybody's kind of
2:30 conscious of it. It's not sexy business, but it's a problem we're gonna have to look at, so. Heck yeah, yeah, leading into not a sexy business. Let's talk about production ops. I know Joseph
2:47 is production ops. Any other production ops in here? EP? Well, I
2:52 can't wait to disappoint you.
2:57 Let's talk about production ops a little bit. Let's talk about the budget and production ops versus, say, CapEx for drilling. What kind of restraints are you dealing with there with the outside?
3:03 Great question. So coming from a major operator and
3:14 going to, it's a very small operator. We have 500 total well boards. Right now we're 1, 700 barrels oil a day with some gas, maybe 1, 900 equivalent So we are extremely small and we run
3:28 extremely lean and we're very new. And we are looking for certain types of properties to bolt onto our stuff to increase our production and all that. So we have a collection of funds and each one of
3:39 those funds has a different set of stakeholders. So when you're talking CapEx and drilling,
3:47 I don't know if this is like a feather in my cap or what, but I've been a part of like three of the most expensive wells that this super major ever drilled. So money was no object. We were
3:58 thoroughly convinced that our CEO was reading our daily drilling reports. So, yeah, yeah. On this side of it, the AFE conversations, the facility expenditure conversations, those are all, I
4:16 walk 10 feet, I talk to our CEO and he says, Can you make it cheaper? And that's about it So for us, you know, the
4:28 delegation of authority guideline is like an order of magnitude lower than what I'm used to. So a lot of adaptation there, but now after two or three years of being in it, you kind of take the
4:43 shape of the container you're in, right? So we're making it work. We've got a good team. Awesome. The reason why I asked that is because production apps I think a great great use case here for
4:58 what we're doing. Running super lean Joseph, like almost, when I met in the first time he had like a roadmap of all the things that Collide is gonna be doing. And I loved it, just like open my
5:10 mind up. So doing more with less, right? So let's talk about that a little bit and how you're using Collide to do those things. So let's just hit up on well search Tell me how you're using well
5:24 search in Collide right now on your side. So what we've done as far as the implementation is we've given you guys a small fraction, we've given you access to one of our funds, and it's right there
5:35 in the city of Midland.
5:38 It was actually the only well data that we had that was already electronic. We buy stuff that hopefully makes 10 barrels a day till Jesus comes. So it's all ancient, right? This stuff that we got
5:55 happened to be from a major operator had a lot more, it was a lot more robust. A lot of it was already electronic, so he's a year. Let's start here. So we fed them, we fed, collide that
6:08 information. And now,
6:12 for example, today, we've got a rig at one of those leases. And
6:18 we're dealing, our GRR has changed, so we're dealing with, as we start pulling rod pumps and stuff like that. We've got to get longer to get tubes and all the things that go with that. So just
6:28 yesterday, hey, give me a well bore history, or give me an operational history of the Isabel 117H. And it says, hey, spud date here. Here's your work. It went through our
6:41 well bore diagram that we adapted from the previous operator and it pulled out the well history. These were the failures. This is when it was worked over. Here's what you have to look forward to.
6:50 Fortunately for us, it's, you know, none of our stuff is really complicated. But to have that stuff at your fingertips, as opposed to like the last fellow was saying, having to go to however
7:04 many different documents and search for it, I mean, it's, that's probably a pretty conservative estimate, roughly a third of my time looking for information. And to dovetail onto that and kind of
7:14 go off track a little bit,
7:17 I think that for those that maybe, maybe don't see this as more of just a way to automate workflows, it's really easy to say, well, we, you know, Dropbox has a pretty good OCR, like I can
7:31 search Dropbox and it's really good at pointing me to the right document. Well, looking for the information and actually using the information are two different things. And I can find the
7:41 information and that'll save time, but actually being able to use the information is where the real value is And that's, you know, that's, that's. That's really where I started to really see the
7:53 promise in what's happening here, because, you know what? I mean, how many of you guys have been on site? Well, how many have you been on a rig or a completion site or anything in the field?
8:04 Yeah, more than half of us. How many times have you asked yourself? Well, if I only had to do my job,
8:13 if I only had to do my job, would it still be even a full-time job, right? With as much time as I've been looking for information? Now, imagine if you have a tool like this that can say like,
8:24 okay, well, not only can I find the information as like, hey, we need to pull this rod pump. You know, can you give me a procedure? Can you help me highlight the risks? Or has, you know,
8:33 based on what we have for well tests, has anything changed in the field that we need to be aware of? Obviously we're not there yet, but it's certainly within reach. I mean, you can see it when
8:43 you think about what this kind of platform can do So, no, it's good.
8:51 So with the foundation and what Colin showed earlier with the foundation of search and then document creation and then a Gentic workflows, where do you wanna see it go next book for you and how you
9:07 use it? Well, search, get an information and do you wanna see
9:13 it move next into creating a procedure that will go out to the field or what are you looking at, what are you thinking about next? Man, how much time we got?
9:25 What I would, you know, for us, and this is, again, specific to our application, like we don't write procedures for every workover we do because they're very simple. Mostly rod pump changes or
9:37 changing out bad joints of tubing. So that stuff is not complicated for us. And we have good field guys, we have great relationships, we all understand what's going on and how to do it What I
9:49 would like to see is.
9:57 more help with the diagnostic piece. And that comes from feeding accurate information to the system
9:60 that it can assimilate
10:02 and then use. And I've spent actually quite a bit of time thinking about this, like before I fall asleep at night, really what we need to do to get there, what we would need to come up with a
10:17 technical library that would be a knowledge base for our instance of Collide to pull from, or as Collide evolves and it's able to ping the broader community, 'cause right now our iteration of it is
10:32 wind specific. But if we were to feed it, say our own technical library on how we wanna operate, or any company for that matter, your own specific way you wanna operate, well now this machine can,
10:45 it knows how you wanna do things and it will structure its answers that way.
10:50 So if we want to put bounds on certain problems to say like, all right, the diagnostic piece of this, like we think we've got gas interference or we, this well is going down, we found a hot
11:01 polish rod today. Well, it'd be really nice if we can ask collide, hey, do you see anything going on over the last 30 days, 60, 90 days that would lead us to believe that this is anything other
11:13 than a pump failure? And it may look at it and say, well, you know, we're seeing this on the trend This may be indicative of gas interference or this may be, you know, whatever it is, water
11:25 intrusion or what have you. Do you have a, you know, maybe it comes back and says, do you have a current dyno or do you have a current fluid level? Just those probing questions that, you know,
11:39 as an engineer, when you hear about something's wrong, first thing you wanna do is okay, well, what does it look like? Second thing you're gonna do is you're gonna start asking probing questions
11:48 And then when you find the spot that - the spot that gives, well, then you dive in a little bit deeper, right? And I think that the only thing that really would be lacking other than, um, cousin
11:59 Jamie over there getting after before happy hour would be for us to give us to give it the right tools and us to give it the technical basis we want to, we want to be operating from. No, awesome.
12:13 Awesome I think, I think just on that point, so I think that's a really good one in terms of kind of level set and state of AI today. I think AI
12:24 does a really good job of finding correlation, maybe even stuff you don't appreciate doing it every day. It still comes down to buck though, being in the subject matter expert to say, is there
12:37 causation there, right, you know, so
12:42 Chuck, would you even say it can be a thought partner? That's good.
12:47 I've got, after my third glass of wine, the night at the happy hour, I'll start waxing with philosophical about the future. But I do think sooner rather than later, like if we're doing this next
12:59 year at this time, I think we're talking about AI being a thought partner on things. Hey, Bob, I noticed productions down this well. Have you thought about this? You know, have you thought
13:10 about that? It's gonna move from answering, finding your data, answering your questions, do you think, yeah, that's the thing? Thanks. And that's gonna happen sooner than we realize. Yeah, I
13:23 mentioned today, I think it was the Michael, or maybe it was Mr. Neely over there.
13:29 You know, really at the end of the day, every one of us is just like a light that shines on shit we think is worth looking at, right? So, and somebody else mentioned, you know, they get a list
13:40 of things they have to navigate, kind of before they really start their day.
13:46 It would be really good if you had something like collide to help interpret some of that so that you could really shine on the things that are worth looking at, right? And you don't have to spend
13:57 that time making the decision on do I pursue this, do I table it, is it a non-issue?
14:03 That I can absolutely see something like that, like you say, in the very near future. You know, just building on that and you said something earlier about, you know, the assets that you guys
14:15 look for,
14:18 what is when look like as you maybe go acquire more assets? Like, what does your team look like with Collide now and building out workflows, agentic workflows? Are you thinking that you guys can
14:32 run Staleen and run leaner
14:35 as you grow or as you get more assets? Yeah, yeah, I do, because Well, for one.
14:46 You know, the way I like to think about this is as a non-technology-centered person. I mean, at the end of the day, like, man, I sit on five gallon buckets and turn wrenches. That's how I cut
14:60 my teeth, right? But what I think about it is like, okay, well,
15:05 when I hired on with wind, we talked a lot. I focused a lot on, well, what does wind look like in five years? We're doing however much production a day, what happens when we add a zero to that?
15:15 And then I started thinking back to, okay, well, what systems do we have in place? Are those systems fit for purpose now? Are those systems going to be fit for purpose as we grow?
15:26 How can I intelligently design some of this stuff, but also not burden me and ultimately the people that will follow me so that they don't look back and cuss my name and all that wise that this way,
15:37 well, that guy did it, this is a problem So what I think is that the real value for Collide is effectively what we're doing is We're going through this evolution on how to optimize our company
15:52 structure by incorporating these types of tools, right? So before all this stuff existed, you looked at the work and then you looked at your workforce and you say, okay, well, I need two of
16:01 these, three of these, maybe one of these, and they all have different roles. We're doing the same thing. We're just doing it with computers, right? So how I think about it is in a perfect
16:15 world, this thing very quickly takes the seat of an engineer or a tech in some capacity, right? And it's not just the tech that when I go talk to that tech, they're now occupied. It's a tech that
16:28 can work for all of us at the same time. We can all ask questions. It doesn't take time off. It doesn't gripe about having to do something late at night. But all of a sudden it fills a seat that
16:37 we would otherwise be paying a person for, Sounds like AI took that job. It didn't really take that job. It provided an opportunity for potentially those other people to do other things. And for
16:47 me to maybe give more resources to my family, right? So all of this is an evolution on how to optimize, how the resources end up falling out kind of on a macro scale. So to me,
17:04 when this thing gets to where it can do some of the diagnostics, some of the analysis, some of the correlation and causation finding,
17:13 and help me focus on the things that really matter. Well, then that's, yeah, of course we can stay lean at that point. Well, it doesn't do me any good for our pumpers to drive by the wells that
17:26 are working every day and always let me know that they're working. I expect them to be working. Let me know when they fail and then let me know how profitable those barrels are so that I know to get
17:36 that money there as soon as reasonably possible. So, yeah, absolutely, I mean. Yeah, that's good. And I like the office base. What does it look like for a pumper? How do you think a pumper is
17:52 gonna interact with something like collide or down the field? Are they gonna check the scores of the game and then drive up to a well and say, Hey, here's the situation here. What's the procedure
18:06 or how do I bring this well back online? Maybe you have a new bumper out thereand the
18:13 guy's been working that field. He finally retired and he's gone. You know, what's that look like to you? Well, think about it like this. And this is industry agnostic. If you have people
18:25 working for you or with you and they have a problem,
18:31 how do you want that conversation to go? Do you want people to just constantly give you problems? No, no Well, I think, you know, when our. When our pumpers come to me, like we're close enough
18:42 and reasonable enough and understanding enough to where when they have a problem, they'll propose a solution as well. Or they'll provide me a little bit of data and say, man, I think it might be
18:53 this, or maybe it's this. And it's okay, well, send me your data and then I'll do some digging. So I'm gonna incorporate what he's told me and then see if I can find some other data points and
19:02 maybe we come to the most probable solution
19:07 In a perfect world, our
19:10 pumper pulls up to a site, say he's got a hot polish rod or something like that and he's just doing his diagnostics and he may say collide. I'm sitting over here at the bus barn 2620. I got a hot
19:21 polish rod.
19:25 We appear to have lost pump action. These are the pressures.
19:30 Can you help me with the diagnostics, right? So when he calls me, he can say, Hey, this is what I found.
19:40 you know I did some cursory glance this is this is what we think the probable situations are right the probable failures and then I can I can with a higher degree of confidence say okay well did you
19:54 ask collide did what it what it collides say or what what are you seeing what did you feed into the system and then we can have that discussion and we're a little bit further down the road then just
20:04 he calls me up and say I got a hot polish ride we've got to get a pump truck tomorrow whatever you know we can already be a little bit further down and because he has access to that knowledge and
20:15 access to that data eventually he's going to be more competent too so when I think about our grades of pumpers you have the brand new green guys wet behind the years don't know a whole lot still
20:25 learning so what is the goal for them the goal is if you see something that is not within these bounds ask for an adult right you're more senior pumpers may be able to come by and be like, No, man,
20:37 you just got to choke they'll know, right? But what do we really want is when they have problems, either they can diagnose it themselves and it's a non-issue, which frees up everybody's time up
20:49 the chain. Or when they call me with a problem, they already have some probable outcomes ready in hand and we're already further down the road to get it fixed. So I really wanna build PumperBot.
21:03 This is another three glass of wine grant I go on And my questions for you, I've got two of them. How much tribal knowledge on the well? Do you think it's in the Pumper's head that never gets on a
21:20 sheet of paper or in a database? That's number one. And then number two, waste to text is scary good. I mean, calling the tune that up there is amazing. We record all our meetings that collide
21:36 these days. And you can not only ask, what did Colin say? You can say, what did he mean by that? And the answer you get back was scary good. Do you think a pumper would be out there talking to a
21:47 bot? If it was helpful, if it was helpful. I mean, you know, what's the best compliment you can get in the oil field, at least in the upstream?
21:58 That dude knows his shit, all right? So if the tool knows his shit, course he's gonna use it as a resource. We would have to train them and maybe break some habits, but yeah, they absolutely
22:09 would use it, especially the new guys, because they're looking for something to hang on to, right? But you mentioned tribal knowledge around public. We have an example from this year. We had a
22:20 guy that was working at one of our fields that spent most of his career at this field. He knew when to flow what wells, what's going on here? He knew all about the structures. He knew all of the
22:30 history. I mean, he knew, all of the stuff about the existential threats 'cause this stuff was like built in the 40s. He knew it all. How could you possibly, possibly, download all of that data
22:45 and make it useful? You just can't. So somebody's gonna have to learn. So what did we do? We transitioned from a guy that was retirement age to a guy that's in his mid 30s. There's no way he's
22:60 gonna know everything. How do you do that? Well, what we can do is we can feed something to like a collide that may retain some of that or a Nick, I've been using the hell out of that granola app.
23:12 You sent me, I mean, yeah, it's incredible, man. That's on my phone. I used it in a restaurant in a business meeting yesterday and it got everything. The restaurant was loud, it was great.
23:24 But yeah, I mean, if you can
23:29 have those interviews capture that stuff especially when things go wrong.
23:34 and make it available to everybody, it just has to be there. You don't have to necessarily prompt them to use it because sometimes, somewhere, I mean, you'd rather be looking at it than looking
23:44 for it, right? So if they don't know, and they don't, what's the quickest place I can go to find out? Well, Pumperbot, it might just be a line, you know? Yeah, if you had like, how
23:57 everybody's doing well reviews? Yeah, I could get with the Pumper's and I sit down and go through the well. You have the well that's creating, like, did you have a recorded version of that? Yeah.
24:06 That was transcribed. Holy cow, that's a lot of knowledge. Yeah. You get a fraction of that knowledge base 'cause I'll always talk and then shit, Oh, well, if you do that, this always happens.
24:17 Mm-hmm. That well failed, you know, seven times this way and this is what you have to do. If you had some of that captured in the well review, that could be a really cool way to catch that.
24:26 Absolutely. It seems to me that would be so useful in safety as well. Oh, yeah if
24:31 you can ever get that. Yeah. on that, obviously, on the chemical side, you're gonna turn around and stuff like that. It's huge, you're about to go in here and do XYZ. And even if you start
24:43 giving a visualization of glasses and you start putting it to actions and now they don't see it, it's really cool where my mind can go. Every facility is a little bit different, right? So, I mean,
24:55 coming, so well-controlled days, I mean, this will apply to the further downstream than us side, but like, it didn't, it doesn't matter me what the flow rates on this well are If it's a surface
25:05 well, like, I know that at the end of the day, like, I still have to kill this well. And these pressures and fluids, they all relate a certain way. Well, whenever we would hear stuff like, oh,
25:15 well, we're capturing all these JSA's and you can go in and live edit the global JSA and everybody rolls their eyes. Because, I mean, it's so diluted at that point, but every facility is
25:28 different, every well is a little bit different. way that that stuff is really useful, because everybody already knows how to do the job. The only way that stuff is useful is if you have specific
25:38 information to that application. If you are, let's say you're in a facility and you happen to know, hey man, that valve is a little bit touchy, you gotta watch it. It doesn't quite work like
25:51 what you used to. How do you capture that?
25:55 Well, that's the knowledge that sits up here. That's the guy that just retired, right? And then we're all gonna learn that lesson again. A little scary. I make outlandish statements, 'cause
26:06 that's just kinda what I do. And I've been saying in meetings that 80 of the tribal knowledge in this industry never makes it on a sheet of paper. And the chief operating officer of a multi-billion
26:19 dollar EP company said, Oh, it's 95.
26:25 Yeah. I'd say that's probably accurate. When that happens, you must be able to save again. Yes.
26:34 Yup.
26:37 Awesome. Well, I'm super excited to be working with Joseph and Buck on the production ops side. I think there's so much opportunity there. Go tell your production ops colleagues about what we're
26:48 doing over here. We're gonna have an awesome, awesome roadmap. It's gonna be really exciting.