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 All right, so we got Ariane with us here from Kraken.
0:05 I'll give a quick overview who I am. So Nick Smart, four deployed engineer here at Collide, started back in May, background in geophysics and GIS. Was it Chevron for five years?
0:16 Yeah, I'll go. Ariane New Jersey, I am working in midstream commercial at Kraken, I've been there for seven years now I started in
0:26 production accounting and
0:29 reporting, and then got into more of the midstream ops and commercial world, and that's where I live now. Thanks. So Kraken was one of the first projects that I took on whenever I started here at
0:44 Kaleid. It'd be great to hear Ariane kind of your background, this initially used case at the midstream contract review. What did that look like before Kaleid? What did your day-to-day look like?
0:55 How much time did you spend on that process?
0:59 Yeah, so really, we started off, our CEO
1:03 was interested in getting into the AI game, and so we got with Collide, and basically got a couple of our key members that worked on some tasks to brainstorm ideas that we could maybe hit the ground
1:18 running with. And so one of the tasks that I thought of or the use case was our mid-stream contracts, specifically focus them on the scope of just our gas gathering and purchase agreements.
1:34 And putting that into Collide, doing everything they said, chunking, vectoring, and being able to ask questions and chat with it.
1:44 Basically the way that would work right now is, sorry, before Collide was all of those agreements are just in a file explorer structure saved in, and however we've come up with our precedent of
1:57 saving contracts and agreements and drafts and partially executed copies and fully executed copies and everything like that. And we just have to basically follow the maze and find the agreement
2:10 you're looking for and maybe go through five or six different agreements until you finally find the one that has the info you need. What were some sample questions? Like if someone's like, Hey,
2:19 Ariane, can you go X? Like what would that request look like? Yeah, a really good one is, and something with, I've pressed on a lot, we've worked a lot on with Collide, is
2:33 analyzing dedications and acreage dedications in contracts. That's a big part of why the contracts are made in the first place, or one of the main exhibits. And so, a question I might get is, you
2:47 know, this, we need to tie in some new wells to a, a gatherer is the acreage that these wells are producing, is that acreage under any existing agreement? So we drill our own wells, we inherit
3:03 wells, but by way of acquisition as well. And so we, sometimes we may not, we might not know the top of our head if that acreage that is fully a white space has been dedicated under some other
3:15 contract that we inherited maybe. So that's a question that's always a frustrating once they have to go and look through And you don't even know where you're looking, just all of your contracts to
3:26 see if any of them that you don't know have happened to have these certain sections in the unit in their dedication. And I'd imagine that those questions, right, that's not a part of your
3:36 day-to-day, where these are just things that you're naturally thinking of, you're kind of stopping, focusing on the requests, going through file folders, et cetera, to try to - and scrolling
3:47 through pages and pages of documents, find that answer, pull together a couple of different documents to compile and then give that back, would that be a fair statement? Yeah, that's exactly
3:57 right, because you also don't know, let's say, oh, you do find the answer you're looking for in one of the contracts, well, what if it's in another one or another two or three, which has
4:05 happened? And so that's, I mean, it could segue into that. Collide has been able to do that, is say these agreements show the section you're looking for under this certain amendment, this
4:17 original agreement, and give you the exact answer you're looking for. Yeah. Maybe we can talk a little bit about like what that process looked like. From start to finish, we all came to us with
4:28 this use case, we engaged, we got access to the data. I think something that might be really helpful for the folks here is like, what did that, what did that user engagement process look like?
4:39 What did the iteration look like?
4:42 We had a lot of back and forth with looking at chunk size, getting feedback.
4:51 to you as a customer being able to work with us and make sure that those answers are correct. Yeah, the way we handled that was, first I started off with, I said, we narrowed the scope to just
5:04 gas purchase agreements with the future plan of going on to water and oil and fresh water and everything too. But now we're in the scope just to gas to keep the variables low. The initial part of
5:18 the project was even smaller than all the gas agreements It was just a certain gather, or I think we started with two, maybe.
5:26 One with a lot of agreements and amendments and stuff, and then one more that had just a few, just to narrow that scope even more for the initial testing. So, started with that. Got the
5:37 agreements in there. John and the team did their work to a ragget so we could ask our questions. And then basically just plugged away asking questions that I would wanna ask or I would get asked
5:49 about our agreements. and got answers, and then follow up questions, stuff like that in the chat, and then having the communication with collide after asking these questions and getting answers on
6:05 the feedback that was provided, that I could provide, that was a big part, was having the ability to basically thumbs up or thumbs down a response, give a comment
6:19 to how accurate your response was, and then afterwards we can look at this whole log because I can't remember everything I've asked and the good and bad answers I've got. We can look back at the log
6:29 and see which ones are good, which ones are bad, and figure out how to fix them. I mean, I have to give you all a ton of credit, that was incredibly helpful for us. As we're going through,
6:37 we're trying to improve the product for you. You taking the time to give that feedback was incredibly helpful for us as we help to drive and improve the product You know, John and I spent a ton of
6:47 time going through the feedback that you left. And we were able to really diagnose further, okay, this is the feedback that Ariane's giving. Let's group and chunk the pieces of feedback that we're
6:57 hearing and ask questions about, okay, these seem like overlapping, like opportunities for us to improve. Where do we start? You know, one of the top ones was, you were limited to only like
7:09 three results, right? And so that was something that we thought, okay, we need to go back and expand this to where you're getting all of the relevant items. The other was the chunk size, right?
7:19 John mentioned that earlier. We were able to sit down and review and say, okay, how do we give already and even better answers over time to give more relevant documentation, sift through things
7:30 that might be relevant, but wasn't really at the core of what you were asking. So I have to give a ton of credit to you for like being a great design partner in that way. And that's also led into
7:42 other use cases. Maybe we can talk about those for a couple of minutes as well. Yeah, for sure. Which ones do you wanna get into? We'll do the stripper well first. Sure.
7:52 Yeah, so like you said that this project has led us to see the way that efficiency can be improved by utilizing collide. And so the collide team was interested in doing workflow optimizations or
8:07 automations. And so I basically, I got with my team and asked them, I posed the question, what are the most frustrating, frictional workflows that you guys have that just piss you off to do them
8:22 every day or every week when you have to do them? Because let's stop those. Let's make your life easier and happier. So sorry, thank you. They came back with a few that we're working on. One of
8:36 them that I think the team's pretty close with is stripper well applications to the state of North Dakota. So what that is is there's a requirement, a threshold, that for production that a well,
8:53 if a well is within, then you can apply for stripper well status or designation with the state of North Dakota. They send out to the tax office and then if that's approved by the state, you get a
9:04 lower tax rate. Well,
9:07 what it takes to create those applications is first figuring out which wells are even applicable to be able to apply for this exception So we kind of figured out that part, just a simple Excel model
9:22 that we just refreshed with production numbers and then we get the list of five to 10 to 15 wells that month that we can apply for strip or wall status on. Was the step after that that was cumbersome
9:37 and that is because every single one of these wells that you identify this month needs a unique, well not unique, but an individual letter application sent to the state, and this is a physically
9:49 mailed letter,
9:52 and what that is is we've over the course of time, we've got a precedent of what the state wants to look for in these letters, so basically we've got all this header information and certain pieces
10:04 of production info and data related to the individual wells that the state needs to see in order to approve the exception.
10:13 Well, that you have to go and find all of this information from different places, different resources, and put them into one letter. This is for every well, every month that you identify. So
10:27 that's where the engineer
10:29 and my team that does this said, you know, that this part seems like it's, you know, like we said, it's repeatable, it's frustrating, and it's not really brain intensive. You just have to go
10:40 to different places to put translate information to a word document and then print it out and mail it. We went through this workflow with Collide and it seems like it's a perfect use case for the
10:53 workflow automation. So now I think we'll get to the point where we can just input the list of wells and maybe a couple other pieces of information that are already in our Excel workbook. Input that
11:06 into the workflow automation, tell it to spit out the applications. And here we go, we can download the applications that we need and mail them Or even have a mail service, mail them directly for
11:18 us just with the click of a button to approve them. And I think this is really a great testament to the slide that Todd showed last, where we want to start small, right? We had great success with
11:29 the marketing contracts, oriented a great job of asking his team, you know, like, what else is pain in your ass? What else can we try to fix for you? And starting with something that seems small
11:39 and insignificant, but over time, the more of these smaller tedious use cases these small wins add up really quickly. In long-term, we see that there's going to be a ton of continued opportunities
11:53 that are going to kind of continue to pop up. Arians become very well-versed with what we're building here at Collide. We were over at Kraken this morning talking about traditional use cases with
12:03 their land group. One thing that Tom, you might be interested in, the third-party forecasting tool, the mapping integrations with that. Maybe we can talk to that one a little bit more as well.
12:15 Yeah, so that one is a task that I've done the past and the other one the engineers and my team does now. It's basically getting information from a couple of different places, one of them being our
12:27 forecast output from Aries is what we use. And then also our development map that comes from the reservoir development team. And basically putting that information together to create a schedule of
12:40 wells with IP dates and then a volume forecast as well. And this is on a call it third party
12:50 or partner, gatherer or purchaser basis. So the idea is we send our partners, forecast updates every quarter, every month, depending on how much work we're doing with them, so that we can have
13:05 intelligent conversations about what's a plan going forward. So the way it's always been done is we get the volume forecast and we put that into the format that we're looking for with only filtered
13:19 to the wells that are potentially going to this gather, and then the other piece of it is, what's important to them is they're pipeline companies, they need to come and connect to these wells where
13:29 they need to know where the wells are going to be located. And we're talking about PUDs, these are perspective pad locations.
13:37 So we have to assign perspective locations to these pads, which are in this stage for things that are further on in the timeline, they're very arbitrary. You just look at where your unit is and our
13:49 development team kind of just throws a marker on a map and calls that good until it gets into the scoping stage. But those initial pad locations are still important. There's a little bit of work
14:01 that goes into them. So that's the first pass that we want to have the conversations with the midstream companies about
14:07 So that process of getting those locations into a table of STRs, section township ranges and quarter quarters for them is very tedious because we just manually through human input, look at the map,
14:24 find the perspective pad, find the section township range and quarter quarter of it and translate that into the table It's a very man-hour intensive process that's very frustrating when you talk
14:39 about some of our gathers hundreds of Wells' perspective in our drilling program that we wanna talk to them about. Supertime intensive process. And really, just yesterday, we had an awesome
14:51 follow-up conversation about that, where Ariana and I were talking about the potential to automate this process, very similar to what we've done for the regulatory workflow. Take information in
15:04 different Excel sheets or XLSB files, and then port that over to the respective locations and see if we can read these PDF files that were generated through SpotFire, and then these well pad sites
15:17 were manually added on using PowerPoint. So the question that I had for Ariana was, okay, well, where is this PDF being generated from? He says, Spotfire. Like, okay, where's the source of
15:31 these PDFs? I'm sorry, where's the source of the well pad information? And this is manually edited in a word editing processor. PowerPoint, yeah. And so the follow-up question was, okay, well,
15:43 let's try to dive deeper and explore and see if there's a way where we can add spatial data and have these well-pad locations added where there's more metadata associated with it. And then rather
15:56 than us trying to go scan a PDF document and have there be document intelligence to read through all this information and parse it out, if we were to use the origin document, so whether that's a
16:09 shape file, whatever format we need to pull from from Spotfire, what if we use that as a reference instead, which seems to be like the path that we're going to be using moving forward. Yeah, so
16:20 just investigate into this workflow some more with the development engineer and come to find out he's actually doing like half of this process already in a GIS software and just not doing the rest of
16:34 it in there. So to where we then have to use the maps manually. So basically that spurred a conversation of, well, can you basically complete this task to the end, which means placing the
16:46 perspective pad location markers into the GIS as well. So I can just take that, give that to collide, and then collide can use this map file to create the forecast that I'm looking for. So I guess
17:00 just a point to just thinking about the workflows can make you figure stuff out that you didn't know before Absolutely, and helping to, I mean, small step that he takes that's gonna make your life
17:13 in our development process much easier. This also opens the door for us to be able to start pulling in some of these JSON files or shape files that so Ariane's gonna be providing me with the packet
17:22 of documents within the next week or so. And then Michael and I are going to begin really like our first full fledged FDE project and see, okay, how can we go through, and first of all, complete
17:35 the completion of the Excel file output. Ariane said, Hey, here's the ideal output that we'd like to see with specific formatting, which is easy for us to pursue. We know exactly what we need to
17:47 have as an end output. Michael and I are then going to sit down and think through, Okay, how would we want this information to be presented to the user? Is there a visualization element where we
17:59 can highlight or zoom to specific well pads when you ask questions about it? So this is really, I mean, for me, I'm extremely excited about this. I'm going to geek out over GIS for a couple of
18:10 weeks, but we're, again, we're really excited about this
18:15 opportunity to start with a small scope of work, right? But this is going to add continual value back to Kraken and to other use cases that will happen in the future. Yeah, just to throw another
18:27 one that we talked about that That's a quick question from sitting over here with Plains All-American.
18:40 Well, let's see if I did, and if not, then ask your question. I'm sitting there going, OK, if we can automate the projected volumes by marketing agreement so that we can go talk to the
18:53 mid-stream company, we can automate that, see how that saves you time and gets rid of a headache. Does it become a constructive tool in being able to iterate with the mid-stream company and planes
19:08 goes, Hey, I can't take all those volumes, can you give me three less wells or something because you can do it quickly, is that going to lead to potentially more agreeable outcomes between the two
19:21 of you? That's exactly what I was thinking. It's like, I think it becomes a partnership, yeah, and then we just start splitting the difference on the capital costs. We're both saving by getting
19:30 more efficient, right? Absolutely, yeah That's a big part of what happens and it's kind of something that can't happen fast enough because it's such a cumbersome task to do. But it's actually,
19:42 it's usually the opposite. It's not taking less off. It's more of, hey, we can do more out here. Do you have anything else over here, too? So when we say, oh, yeah, we do. Let's go through
19:52 this cumbersome process of making you another forecast with all this info that we didn't put in the first time. So yeah, that's how we would do it. Add on more and show even more to the third
20:01 parties. And if I'll guarantee those volumes, do you not 50 cents off, or whatever the case may be? Exactly, yeah, let's negotiate with the story.
20:12 For sure.
20:14 Any questions?
20:20 It's really cool, the marketing contracts. I assume, are you going to be able to put the GIS part of layering the acre of dedication back over that? That's what I was getting to before Chuck cut
20:31 me off. Yeah, that's exactly the use case I was talking about. I forgot about it. Yeah, I've talked to Nick and the team about that, is those dedications mapping those is another extremely
20:43 cumbersome process that we take on every blue moon just because it takes so long, but it adds so much value. So, yeah, being able to visualize all the dedicated acreage in a certain contract,
20:56 everything that was added on in future amendments, and then overlaying other contracts, 'cause like I said, some might have the same sections dedicated, some might have different ones, some might
21:06 have well bores only, some might have acreage, some have just lands, there's so many different ways to be able to label it and visualize it that it's It's. You can't grasp it as well when you're
21:19 just looking at the table and at the end of the agreement. Well, and to add more context to that too, when we had these discussions, you were saying, Hey, I'd love to be able to spin up some
21:28 more maps, but y'all run lean. So you can't just ping your GIS analyst for every single mapping request you have. So if you could throw somebody together really easily in a collide instance and say,
21:38 Show me this particular map location with these particular elements, and I want it to be these specific colors, you could have that done You don't have to ping that person anymore, you can get work
21:50 through much faster. Yeah, yeah, exactly, that's huge. We have one GIS analyst, and she works under our reservoir development team. So like, we don't bog her down with that stuff. She's
22:01 focused on other things. So any type of maps for our midstream agreements that we want, it's just, we have to build a man-hawk ourselves. How many different gather in gas industry? Yeah,
22:13 absolutely. Do you have, are you in the dozens? Uh, I'm. Gas gathering partners, we've - It's like the different dedication agreements. Oh gosh, every agreement has dedication. I mean, we're
22:25 in the probably close to 100, this is my guess. Yeah. Those, these are those, but we really agree that some priority volumes, like yeah, certain part of our yarns, and so trying to optimize
22:37 that with those mid-screen partners. Exactly. Yeah, that's very cool. Yeah, yeah, exactly You'll end up with primary dedication and then secondary, and then sometimes tertiary or areas of
22:49 mutual interest. It's like, hey, let's talk about it, but we're not necessarily obligated to gather them. It gets super messy. Yeah, so the current search that you do on the, what's currently
22:59 working, like your search is, hey, give me this section or this quarter section, and like what, what, about these documents, have that in there, or how does that actually work? Yeah, so
23:09 types of searches that will happen right now or I might ask like, well, this is actually sample I got. Yesterday I had a new amendment come in for some new wells that we're getting tied in. There
23:19 are parent wells in the existing unit Which are on an old contract that we inherited? Well, these new wells they're longer at laterals, so they have another Section in there so but these new wells
23:32 we were going to add to another another a different agreement a newer one Well, I got an amendment to get from the from the third party to get those sections added to the acreage Well, I saw one
23:42 that didn't make sense. It wasn't in that acreage. So I went and collied and asked what Agreement is this section township range? Under and and it took me exactly to the amendment that it occurred
23:55 it existed on that was from 2021 That's already dedicated to them. So it did not need to be on this on this amendment. So that's an example of Of asking what agreement is this section under and that
24:09 I go the other way as well like what? But what sections are dedicated under this agreement? Those are, those are simple ones. The old way you can do anything that you just have to go through all
24:19 of our tickets. Yeah, exactly. Do you have the dictionary where you can ask well name, and it goes to the same PI. It's got places all the way through the section township range, or have we not
24:32 gone that far yet? We haven't gone quite that far, so some of the agreements have well names in them that are associated with the sections that are getting dedicated. But some don't, they just
24:42 have a big long exhibit of like 100 rows that labels every single section being dedicated.
24:50 So we haven't done quite yet. We talked about it, but we haven't assigned the metadata of unit sections, spacing unit sections to wells yet. That's something we've talked about that I would like
25:03 to do as well, because we have some agreements that don't have the well names with them.
25:11 Visualize it very easily. Right. So I do want to say that as Chaka asked that question, my deaf team here, that's busy talking to me. Yeah, my deaf team just looked at me. Just with this, with
25:23 the, I don't like smiling. Oh, great, now we have to kind of build this like my tomorrow as well. Awesome. We're demoing that QIS stuff at happy hours. Yeah.