All right. Well, welcome, guys. This is our our first episode of of Energy Bites. You're one of your hosts, John Calfan. Bobby Neelan. And uh we're here today with Talal from MI4. Hello. Appreciate you being the first guest. We're excited. me. Yeah, nice to be here. Don't know that I could ask for a better first guest, actually. Thank you. So let's
Let's jump right into it. Uh so just tell us about yourself. Who are you? Where kind of what's your what's your story? How'd you get to where you are today? So um my name is Talal Nimi and I'm the founder of MI4 Corporation. And we're a software company, a software and data company for oil and gas. Um I started MI4 around 2000. And initially um I was just doing consulting for different industries, software consulting. And um for the first three or four years, I was mostly heavily involved in building um litigation, mass litigation type um software that didn't didn't have a solution off the shelf. So I was doing something very custom. And then in um when that project ended, I got into the oil and gas industry by accident to help a friend with something. Um it's a very usual scenario. An engineer who was very self-taught developer moved to a different company and everything that he had built became now like, you know, um an urgency that nobody else could take care of. So they called me to come in and help. And um I really liked the the difference in the momentum and the pace and between working in a, you know, legal um field versus the oil and gas field. So I liked it so much that I focused entirely on oil and gas. So ever since we became dedicated to oil and gas, 90 90%of our work We're still a consulting firm, but we also develop our own so since then we've developed multiple products for the oil and gas industry. Um, and here we are. Yeah, now that's it's one of the reasons I think you're such a great first guest because you've had kind of both sides of it, right? You've got internal building your own software plus you you guys have done quite a bit of consulting and you kind of get a feel for what other people do and what problems there are in the market and stuff like that. Um, how how did you get into kind of tech and and stuff like that? I mean, were you interested, you know, in electronics and stuff as a kid or how how did that kind of evolve? Yeah, so I started when I was a kid, you know, when I was 10 years old and was you know, I remember Commodore 64. It was I I wanted a Commodore 64 and I was just nagging my dad to get me one. And this was, you know, back in Lebanon before I moved to the US and it was pretty much the price of a car and my dad didn't understand why I I needed it so bad, but I was so persistent he finally got me one. And I just started tinkering. I was always a tinkerer, you know, I'd take take the VCR apart and try to put it back together. Uh, so I got into programming, you know, just for fun at first playing games and trying to make a game and things like that. Then I got my my bachelors in computer science, and I was very geeky into, like, I wanted to to continue down that track and get a PhD in computer science and do more research like type work. But I moved to Houston in late '90s, and I got my masters in computer science at U of H. And while I was there, I mean, I started getting more exposed to, you know, real work versus, you know, academic stuff. And I was actually, you know, initially I was focused on some projects that were AI related and if you want to get into that, there was an AI winter. There's a lot of, you know, ups and downs in that field. And then so I switched when that wasn't working and What was the what give us a time frame? What year was that? this was in 99 and I was involved in a project that was trying to
process 3D seismic data using computer vision algorithms, which is part of AI. And identifying reservoirs, things like that. That's crazy it goes that far back. No kidding. Yeah, and it probably even goes even further back than that probably, but it just you know there's those are still hot topics. Yeah. So I want to hear more about AI winner once we get to some of these AI questions, but keep going. So then my my real passion was you know databases data data related. So I focused on that. And most of my consulting work was initially just building databases for for companies for for internal purposes and things like that. And What were you using? What databases at the time? Mostly SQL Server. Yeah. A lot of access databases for small things, you know, which which became came in really handy when I worked with a lot of companies that have Aries because it was all access at the time. Yeah. And being able to take all these access databases and convert them into SQL Server databases was a very you know popular thing so that you know got me a lot of work early on. Cool. Yeah. How did you how did you kind of make the jump or transition to kind of consulting and and doing your own thing? Well it wasn't really like a planned thing. I was initially you know after I finished my my degree I was working for for a company and trying to you know get more experience in that sense. But I had already had a lot of experience being a freelancer. And my first project, that was, you know, was like writing code for SCADA controllers for a jet like real-time data using Visual Basic and was you know chemical injectors inside jet turbines that like GE was you know using so that in real time they're cleaning the turbines and reading the sensors to see if the cleaning efficiency was So it was kind of like PLC type stuff? Yeah, I was very it was it was you know very basic for me it was pretty easy but yeah. But for them, when I saw how important it was that we were able to do this and you know it really changed, because initially, you know, you're... and this is something that I think is important to hear for for aspiring developers is that: some people or some media might make it sound like, "Oh, it's so complicated to to to build software." Well in reality it's really if you break it down into smaller you know bite-sized problems it's very straightforward to to do. So, while I was working at a company here in Houston, and you know, when there was a a financial issue crisis you know, after the dot com bubble and Enron, and it caused a lot of companies to go out of business I found myself looking for a job. And I, I started reaching out to my friends who, you know, were like in related IT stuff, and got introduced to this law firm that was handling mass litigation on a massive scale; and they were doing everything they could to keep up with the, you know, Access databases and things like that. So they brought me in and I for several years I like built a application for them that was to allow law firms across the country to collaborate on mass litigation cases before you know in in a secure way. Right. So it was pretty cool but yeah. Do you feel like I mean you were just saying that building software can is actually pretty straightforward if you just have the right kind of processes in place. I mean, have you do you think that I feel like it's almost one of those the more things change the more they say the same kind of thing. I mean like are you there's still probably fundamentals that just are the same as they've always been and maybe some of the technologies change a little bit, but Yes. that's that's something that's very important because um um people who are new to software development they think that oh I got to learn this language to become a developer. In reality what you what I think you you should learn is the basics. Go go learn the fundamentals uh computer science 101. You know that that's basically learn about what happens inside the computer when you when you're writing code to do something because that really helps you later on in ways that are so subtle that you may not realize it, but uh because for example if you're debugging a problem that you're having and you're not very familiar with how memory works or or there may be a leak or something like that, but also learning about data structures and different you know algorithms and things like because once you do that you can then learn how to learn new languages, you know? So um I'm not really an advocate of one uh Technology. techno your stack over another or one language over another. I definitely have preferences, and I think everybody has different every developer has different preferences and likes and dislikes. But I think at the end of the day, it's just a tool for a task, and it's important for you to be educated to know which tools are more suitable for which tasks so you can pick them. And very often, I find myself using a tool that I don't really like, but it is the right tool for that task, you know? Like Java for example I I really hate Java. Everybody knows that. So um yeah, but like overall since you asked at work we've always been mostly in a Microsoft shop Yeah. and there's a lot of convenience in sticking to a platform um that that gives you all the tooling and all the tool set that you can use and that also has enough community you know support behind it. Or even just hiring too, right? I mean to be you bring people onto a project or to even just have them jump right in. I mean, Yeah, no. I mean, that that's something that even we kind of run into with some of our the projects that we're working on on the dev side, right? Like no-code tools are great if you can find people that know how to code in that or dev in them. But if you can't, then it doesn't it doesn't doesn't make much of a difference. What uh I'm curious too cuz I didn't you know I just had never thought about it until uh like I was at Hivecell really. But you know, people think a lot of people think you know, dev and coding is it's just Python or C or whatever it is. But then you know, when you you have to take a step back and understand the architecture, right? And like the server, you know, that's to me as someone who has never been formally educated on the kind of software dev side, it's like okay, you start writing little scripts and little things here and then you're like okay, well now I want to do this by itself and I want it to do it on a schedule or every time this thing occurs, it's like okay, well now I need to spin up a server and and then I now need need to figure out what all of that means and then you've got like the frameworks and the OS's and all the stuff stuff down at the hardware Yeah. The hardware layer and then yeah, the different platforms, whether it's ARM or uh Yes. and so it's I think that's one thing that a lot of people either on the outside or just starting out don't really understand that it's a very deep uh kind of Well, a lot of things that are attractive for people that they don't appreciate and I think they get into also as well. Yeah. And I think it also goes back to uh why are you trying to learn this again? Because um if I was for example a petroleum engineer who's trying to learn you know, skill up and learn um how to use uh coding tools and things like that uh it's for a completely different purpose than if it was someone who's trying to build software to sell, right? So there's very different paths for those two things. Um and um like since you mentioned you know, the the the DevOps that go around building software. If I was building something internal that only I am going to use, then it's really pretty straightforward. I could, you know, write a single file, you know, of code and have a task task scheduler on my machine and then now I have an automated thing, right? And and and I've had you know, I've had stuff like that built internally that only I would use so it would run something for me on my machine, but know that when my machine is not working that this whole thing is basically shut down, right? Or let's say I'm I'm writing this little script that will hit an API that has you know, prices or I I want to keep my own spreadsheet updated. If that API that has the prices on it is now down, then what's going to happen? Am I going to get alerts? Am I going to be able to get caught up when it's back up and retrieve the values that I missed? So these are all things that you have to think about when you're building commercial software where that may may not be that important if you're building something personal, right? sure. Yeah. Yeah, and even like some of the like the testing that that are on your code or lack thereof, I mean testing, documenting. Yeah, No, that's that's very very true. Security is the biggest thing, right? So when when you're building so the way I I was thinking about like what what would be a good example, if you're building something personal or something internal for a small company that only you're going to use, it's like you're building your own dirt bike, right? It's loud and but it gets the job done. It takes you from point A to point B and you know how to fix everything and you can see everything that's in it. But if you're building a commercial application and that's what you you got to think of it like you're building a limousine bus that should ferry people safely and securely over public roads. Now it's not on your property anymore. So you got to you have so many other things regulations that go into place, and you have to worry about other people's safety and data safety. And so, that's that's I think a good example. And the size, the scope of what you need to build changes so much. Yeah. I think it's something John and I learned pretty quickly when we were working on a web app at Reservoir Data Systems. Um and you know, admittedly probably had no business running that project. But we were the people that you know, that were tasked with it and learned a ton from it. Um but just think people underappreciate how hard it is to write software that is easy to use. Well, especially at a at a you know, commercial or pseudo commercial scale where it's going to be used by people that you can't directly talk to. Like just I mean GUI, you know, UI UX design is so important these days, right? Like you can have the best software, the best model, the best algorithm, whatever. But if your GUI sucks, it's unintuitive or it's not geared to towards the user, right? I think that's a big problem in the oil field. It's getting better. But you know, you've got all these platforms that they want field guys to be entering information or going through a process or a workflow or whatever. And it's like, well if if it was written by a developer who has no idea about what their day-to-day looks like and what they have to deal with and what is when they're supposed to be doing this process and you know, do they have connectivity? Does it need to have offline all those different variables and nuances? It's not there's a much lower likelihood of success than if you go talk to those people and really understand it, right? Absolutely. And And this is why I believe that um experienced engineers who become software developers have uh because they have the subject matter expertise, can be much more successful at it than you know, developers who have to learn the industry, right? So, um one of the things I did initially when I started working in the oil and gas industry, I was a consultant and um I was embedded with the client and we still are. We work, you know, in their office in the same room sometimes with with with with the users so we understand what they want. Otherwise, you just cannot; there's so much that can get lost in translation and, and you cannot read their minds or understand how they're using the system or out in the field, you know, on site when, when you're developing something that's going to be used while they're drilling, you know, things like that. Um also for like a new developer what it so you mentioned what what goes around building commercial software. It's very easy to make to make someone realize what goes in because what what what you can do is you can build an API that says this is an API that when you call it it will give you a certain data right? You can write that in a few lines of code it's very simple it it hits an SQL database and gives you back some data. It works now how do we make it commercial? Well now we have to first build the testing around it we have to build a deployment pipeline so that whenever we make an update we hit a button and it it deploys a new version without shutting down anybody right? We have to do CI CD continuous integration um then we have to have Right don't forget that too that's another now we have authentication and authorization to make sure that this and then encryption and then we have to do redundancy so what happens when this data center that your API is in goes down all your customers have to wait or do you have a failover plan in right? Is it seamless or is there like you know what's your time to recovery so there's so many things that go into that. Even just something as simple as how much data are you going to return to them? Is it going to be free, or is it going to cost? Like, how much? That's a whole another thing. We were having that conversation earlier this week in the office talking about who was it? Twitter just came out, and now they're charging for their API. And it's like, well I, was I don't know who it was, but it was on a podcast, and he was talking about how, you know, there have been these entire platforms that were basically built on Twitter's free API that were essentially abusing it, right and? every people don't realize every time you call an API, it money. Yes, yes. it's like that's a whole another thing to consider that a lot of people don't even think about, right? It's like, oh yeah, I just need an API. It's like, well, if you're And good API design with like rate limiting and you know And documentation. You got to do rate limiting and throttling and things like that. And what I mean, most people realize is that first it costs money to develop anything and then it costs money to host it. Right. And then and maintain it, keep updated. And then it costs money to for data that goes in and out of that, right? So of course I like the idea of free APIs and that would be great. But at the very least you have to recoup your costs. So what do you do then as? your if you're a big company like Twitter is I don't know about that, because I mean, you know. But if you're small, if you're if you're, you know, oil and gas software company, where it's a small market with sometimes tons of data that people need to transfer in and out. Um you can either bake that price into your general price so that it's you know for taken for granted. But then you're penalizing those who don't use the API and it's a very small percentage of users that actually need the API. I just recently posted a a poll on my LinkedIn to see whether people prefer Excel versus you know APIs and or exporting a CSV at the So I I plan on offering all of the above because why restrict the you know just let everybody get what they need. And um I really wasn't planning on charging for the API and I still don't think I am going to even though my team thinks that we should because we need to be able to justify supporting it and and you know things like that. So it's very tricky. It's a you know it's something the whole pricing thing is very tricky. Yeah, absolutely. I mean those aren't those APIs are not the APIs that they're using to interact with the app generally probably either, right? I mean it's not you know you have So yeah, we have internal we have internal APIs that we developed and then the public facing API is going to be completely different. so it's a separate product. Yeah, so it's completely separate product. Yeah, that's that's something that most people think no, if you have an internal API just let us have at it. You can't do that. Just give us access to the database, right? Oh yeah, actually I what, what I end up doing, because that was very common request and and I end up explaining to them that for many reasons (security being number one, but the many other reasons), you cannot have access to the database because you could shut the whole thing down by just making too many reads, you know, locking and things like that. So what I end up doing is for those clients that are willing to, you know, to to have their own database, I replicate one for them and it's a read-only database that they can just, you know And you host that or or do some you sometimes do like the replication from yours to theirs? I give them a choice. So if they want to have their own database, that way I don't charge them anything. I just push the data to it. And if they don't know how to do that or they don't want to bother with it, then I host it for them and I just pass the costs to them. I don't really, you know Yeah. Yeah, it's whatever it's to host Yeah. Yeah, that's like sometimes like maybe 100 bucks a month or something like that for for I've got a question for both of you because I'm I don't have a a dog in this fight at the moment, but I do, at least from a theory perspective what. How do you see...so given that, you know, Twitter and it looks like more and more companies are going to start charging for APIs, how do you see that kind of playing out, you know, in the marketplace, at least in the energy space, right? And you know, my back Bobby knows my stance on this is it's, you know, most of the time in oil in the oil field, the data that we're dealing with, at least from the software side, is technically the operator's data. It's somebody else's data. The flip side of that is you're doing manipulations, transformations, etc. to make that data more valuable. So then you're adding value to it. But then at the end of the day, you know, is it their data so they should always have the right to it? Is it their data, so they should always have the right to it as long as it's, you know, they're not abusing the API or something like that? Or is it, you know, some kind of hybrid of: you can get, you know, so many returns or so much frequency for free, and then if you want it faster or more often, or you want, you know, all of the data back to the beginning kind of a dump type deal they pay extra for it. I'm just curious kind of what y'all's thoughts are since y'all are kind of on opposite sides of the the market. So, first I'll explain how I've dealt with this with in our, you know, business and then I can give you my opinion on on the rest. Internally what we've done we've so we've had clients come in different shapes and sizes. So some of them have no IT staff to handle these kinds of things so they want us to just host the database for them and they have full access to it as much as they want, no limits and just they pay the hosting cost for it, right? Like 100 bucks a month. That enables them to have a simple connection string that is we secure it so that only their IP address can, you know, and then they can plug that into their Spotfire or Power BI and then they can yeah they can build or put it in their own data warehouse if they have one. So that's that's the easiest. I have other clients where they wanted something on premise. So there's many different ways to do that. We can either you know feed it directly or just send them a daily backup of the whole data. We've done that as well because I strongly believe that this is the customer's data and they should never feel that they don't have access to it because I've had clients come to us from other solutions where they were made to feel as if, you know, this isn't really possible. Oh you want all your data you know we can't. It's in the cloud. We can't really you got to go into the application and export every Excel file Excel and therefore you can't have it. I know this is just an assumption but I don't know if some of that has to do with the stickiness of the application. Like if you make it really hard for these customers to export all their data maybe they won't leave us. Maybe they'll just, you know, and I deal with this every day because I have customers coming to us from different applications where the hardest thing is for them to give us their older data because they don't have an easy way of getting it out. Mhm. And uh I don't want us to become experts on other people's databases, so it's really hard for us to uh time consuming to import that data and figure it out how to put it back together from different reports and things like that. So uh so what we do is we try to give people as much access to the data as we can without breaking the bank because uh if you if you give an API to someone and uh tell them it's unlimited, right? Whether it's free or whether they're paying for like a nominal fee, um there's a difference because someone on the you know Matt Herman's Slack channel asked was was was uh saying API should be free. There's no additional cost for getting data out of an API versus getting data out of the application. And I said that's wrong and here's why that's wrong. When you're using an application to get your data out, there's guard rails. The application controls how many requests to make. You know maybe the data can be cached, can be given to you in a very economical way. Or done in batches, right? batches, right? Versus if you're using the API however you want, you can make as and people tend to then open up the you know the faucets and just like let I I one example uh we we wrote a custom application for a you know big client here in Houston who had uh they they were paying for service uh public data from one of the big vendors that you know they buy data for the whole country. And uh they have an API but it was very hard to deal with, so we wrote a custom application that um pulls data from that API and puts it in their own on premise. Um they wanted all they wanted way too much data and very more frequently than necessary and we try to keep it simple by saying look you can get you know the the the full data and then just get incrementals and Right. No we got up you know so people do abuse the you know that that that API and it ends up being a very expensive uh you know cost for yeah. Yeah I was talking to someone at one of those vendors you're talking about. He works them now and he's saying on the API just how it's actually not the companies you think that are abusing. It's like, you know, the BPs of the world actually have good developers that pull it properly and incrementally and like the, you know, they may pull a lot because they just the volume but that it's actually like some of these smaller people, these small companies that have inexperienced developers or like an engineer that knows a little bit of Python and they're pulling everything every time and it's just Yeah. So, yeah, there's definitely that side of it. I mean, I'd say from my side, I mean, obviously as on the operator side, we want everything as cheap as we can get it, right? Um to kind of shift it a little bit like I feel like I'm almost like going the other way on APIs. Like I used to like the idea of APIs, but hey, there's a lot of bad APIs out there. Like and there's not really any kind of good standard. I mean, there's some of them returning XML, some of them returning You know, it's just And then I mean, some of them just bad. Like one of them it was returning not enough data and they're like, sorry, we can't give you that. You can see it on the front end, but we actually can't access some of that data for you through the API. So it was like, well, this is useless and I just wasted two days of development, you know, against it cuz I can't get what I need out of it. Um Documentation is a very underrated thing in my opinion when it comes to APIs cuz it's like it's very clear when you start messing with them, when you find a very good documented API and then when you find a really crappy one where it's just like here's the call and here's one of the potential thousand returns that it could have or like here's here's one example of some of the parameters that you can pass, but we're not going to tell you all of the other parameters even though they do exist. Or just like random stuff like that. say on the data engineering side, like I deal with some APIs and some of them I don't feel are made for actually for like data consumption as much as like they are for integration. They're built for integration. So then maybe it's great for passing data back and forth between these applications, but then for me to get it and it's heavily nested and it's pretty gnarly, then it's like it's a pain in the butt to pull out. Um I mean, I like where some people are going again like to your point like if someone can just share a SQL server database with me, awesome. We're great, you know. But again, it depends on the size of the data too. Um but I mean, now a lot of people have like the Snowflake and that and that is making data really easy to share across I think we're going to see a kind of a just a change and maybe not just Snowflake, but people sharing like their data lakes, if you will, or being able to share like parquet files or stuff across, you know, just making those things accessible to for people to query against or pull over. Like it's almost like we're going to go back to the FTP days, but with slightly different technology as a middle man, but Tell me more about that just for like the general audience or, you know, the C-suite manager level person that generally speaking kind of understands this stuff but doesn't necessarily know the technical side of it as well. What kind of what what do you mean by that? Okay. So I mean I think we can talk about Snowflake and I think it'll be a similar paradigm, but like so Snowflake and a lot of these new modern day warehouses, the store is just separated from the compute. So what makes it really nice and we've done it with one of our frac vendors and we're looking at doing it with maybe with some of the public data vendors, not that a lot of them have it, is they can do these simple like data shares. So, you know, company X shares data with company Y and within like three buttons, you know, I tell them here's my account information, blah blah blah. They put it in now their data is available to me in a database on my side of the world. And now when I hit it, you know, I pay for the compute against it, but they're covering they're paying for the storage. And the storage is generally pretty cheap or very nominal for, you know, most people now. So they're still hosting it. They're hosting the data. And then you're paying for the compute to pull it into your instance essentially. But then, you know, so now like that frac data that we have across like I can just query it alongside everything else in my data warehouse and I don't need to do any API calls to get it. I don't need to have a nightly process. When data is added to their database, it's available to mine. Yeah. You know, so really reducing I mean but again people are some people are pretty proud of those. I mean you you're going to pay for it, but then you sit back and say, well, how much data engineering time would it take to, you know, not just only develop against it and get all these tables, but then also to have them available in real time and you know That's a very valid point because I think a lot of people me included, I used to be on this camp as well, is like, oh, well, they've got an API, right? Like that'll be easy. And then it's like well, okay, anytime the API gets updated, you now have to update your call. Or anytime they switch the like the thing I deal with on the marketing side a lot is the OAuth, right? So like each social media company has a different OAuth uh you know process for using the API and they constantly change them, which is like okay, great from the security perspective, but I'm just trying to get my Google Analytics data in an easy way without having to rewrite this stupid call every six months. Yeah. So and I think where it's going to go though because like one thing that Snowflake had most of their stuff is in proprietary format, but now you're seeing a lot more like I mentioned parquet. There's like Iceberg, which is like a thing parquet with some metadata around it and now like Databricks their whole thing is that their Delta tables are are open source. So like you mean now I think you're going to see more of that where like they can just give you access, probably do that data sharing with an open format. And now if you want to hit with Python or R or you know some SQL engine or Spark or whatever, then like that stuff's there for you to do. Yeah. And uh you know I wasn't too familiar with the Snowflake, but after your talk about it on on that you know last the the data yeah uh a couple months ago I think. Yeah, we I I went and looked at it and I looked at DBT and it's very impressive at you know what what what you can do especially the the low cost of of being able to just turn it off when when when you're done and uh um and I think that more and more companies are realizing the benefit of of having these abilities to to just do the data set once and share it. And uh recently Power BI just you know announced that now you can have uh shared data sets between organizations. So what you can do is you can prepare the data sets and uh and back to your point about the data engineering piece, that's another reason why the public API has to be different than the private API because it's a completely different data model. And also the databases that we host for companies uh very few actually want the relational database. Most of them want a analytical database. So what we do is we we flatten it out for them. And instead of getting you know 100 tables, they get five or six tables and it's very easy for them to to process. So they don't have to know that person ID equals ID the person's table and those And and and you know like recently for example I have a client who's buying a new property and they just send us a whole SQL database of a third-party product. It's it's so hard to decipher that the the model because everything is gooey and and yeah. And there's no clear there's no actual relationships between the tables. So you have to kind of do esoteric things to figure out whether or not this is a So the data model is very important and this is why like I we spend so much more time on it to make it as simple as possible. So if I you know hand it to someone they can just look at it and understand what it is. Yeah. Well that's better on your side you know it's kind of putting that effort up front so that you you know less like back and forth. when we bring in a new team member we don't have to sit there for six months trying to explain to them where all these you know they can they can just pick it up in a week and yeah. Oh or yeah even on the client side I mean Bobby and I spent a lot of time you know we were very adamant with our web app that it had a a good functional useful API and I mean I wrote a bunch of the documentation on it just by using it and even then with with what I thought was very clear very straightforward you know it returned whatever a thousand records and then had a pagination number or it returned the last value of the last record or I'm sorry the last Unix timestamp of the last record and then you could just append that to your call to start and and just loop through it but something as simple as that we spent hours and hours on calls with people to the point where I just built a little collab a collab note book with it that had the template in there so I could just send it to people because Yeah. Pagination is is one of the you know hairiest things in APIs. There's an easy solution for it. Instead of having to force people to use timestamps and things like that one of the suggested ways is when you're when you're writing an API and you want to give a result set back that is paginated, in the in the response you put the next URL. Mhm. So all they have to do Oh, just the URL. That's really smart. so they take that and post it as a as the next and you just keep going until Feed a timestamp variable. Yeah. Yeah, and that's and that way you can also do the previous so you can, you know, rewind a few. So that's yeah. No, that's slick. So I mean, I think you know, because you talk about sharing SQL server databases, I mean, in that you generally are Microsoft stack. Can you talk about some of the benefits of the Microsoft stack? I mean, because I feel like it's pretty ubiquitous, especially in the oil and gas industry, like whether it's just how well it integrates with AD and you know, just you know, most people have SQL server so that SQL server replication is more straightforward or Yeah. So I, like I said, I encourage people to use the best tool for the job, right? And sometimes you only have one tool, so there's not that much to choose from. Other times you have so many and you have to do a lot of weighing to figure out So to give you like the the the wise why we why we chose to stick with with mostly Microsoft products was because there's so much help out there for, you know, how to solve certain problems. And when you're building commercial software, you cannot dig it. It's, so, one of the, you know, for example, arguments that people make: "Well, I want to use this such and such free open source solution instead of paying for a Microsoft product, because if I want to fix something and the all source is there, I can go..." Yes, but are you are you building a product? Are you going to fix someone else's, you know, source code? You don't have time for that kind of stuff when you're when you're on a on a schedule. Also you need support. You need to be able to to integrate with other, you know. So a lot of the things that we initially were building were integrating with office tools and, you know, things like that. And it just makes more sense. Now I I want to address them like piece by piece, right? So the the the the RDBA the the the relational database software management system, RDBMS, SQL Server. Initially, when I built the first commercial product that we had, production here, I wanted to make sure that I could sell it to any company regardless of which sequels, which database server they were using. And the first version of Production Year that came out in 2007 was on premise. By 2009 it became a SaaS solution in the cloud. But in 2007 I built a data obfuscation layer so that I could be, you know, database agnostic. And the API itself does not know which database server it's talking to. Yeah, exactly. And so that I can map these different functions to return the same data set regardless of where they're getting the database from, the the data from. And because I knew that some of my clients, because I would have done some consulting for them, use Oracle, for example. It ended up being that this was kind of a waste of time because up until now everybody is still okay with SQL Server. And even the company that had Oracle they had a SQL Server as well, right? So, so, and yes. And then, once we moved it off premise, then nobody really cared what the but the fact that, if you look, well, um, even back then when we started, if you look at the tooling around SQL Server, Oracle had this horrible user interface for, like, comparison to SQL Server Management Studio, for example, where if you're writing queries and you want to see the results and analyze the query path and the plan and things like that, you had this tiny little window where you can just write the code and you have to copy paste it to see it the whole thing. It just it was just a horrible user interface and I don't know if they fixed it since then. This is very old for me. But also the cost was a big factor. The the the the features in it that you know how easy it is to do replication and you know failover or clustering and things like that. The the ability to to how hard it was to train someone to become a database administrator where they can administer the backups and all these other things. All of that played, but also the biggest piece was that if you're going to a company where most companies had, you know, started something in an Access database going from an Access database to a SQL database was almost a no-brainer. Yeah. You can pretty much do a, you know, plug and like shift and lift or as they you can take someone's Access database, replace the tables that they had in Access with links to a SQL Server table. yeah. And they wouldn't even and and most of the time the query, the syntax is almost identical between. There's a few changes you have to edit, like replacing double quotes with single quotes and things like that and the joins, but you can take all these queries that they've built, like I'm talking about Aries for example, where they had engineers who have been building queries on top of Aries. and stuff. if you use so taking that and converting it to SQL is almost instant and they don't have to learn anything new cuz everything is still there. Right. So that was also very helpful. Now the analytics. So initially when we started SQL Server SQL Server 5 2005 Okay. didn't have a good reporting engine or SQL Server Reporting Services. So I had to use Crystal Reports which was back then a big commercial product. Still still hanging around. It was like pulling teeth. Yeah. They had like 500 different versions of their you know I'm exaggerating but so many different versions of their DLLs and if you're left-handed you have to install this DLL and if you're right-handed you have to install this. And if you install both of them on the same computer it will blow up. it, yeah. It was so time consuming and painful. But we had to use it cuz back then it was the best, you know, option we could get. But then once there was a new version that came out of SQL Server Reporting Services they were so much better. We immediately switched to that and we're still using it for the paginated tabular reports and things like that. And the integration of that with SQL Server to be able to schedule and send out reports and so again tooling when it comes to development, we've been using Visual Studio for ages, and it's been constantly improving. They made a free version, community version, Visual Studio Code, you know, all that stuff. So, we really it's not about that we like always wanted to stick with Microsoft product. It was always the better choice for us, right? We really had no reason to to go in in in and but to keep to make sure that we are always doing the right decision, you have to educate yourself about what else is out there. Yeah. You know, so you have to be familiar with other IDEs, Eclipse and other things. And so we try to do that, you know. I'm not a big fan of trying to learn a new language just for for the sake of, you know. But I think you should know that that language exists and what it can do Mhm. so that when the time comes where that language makes more sense, you at least know about making that decision and you're not because I've seen a lot of people stick to their guns Right. and this is what we use and they end up it ends up being a disaster sometimes because they're using the wrong thing. Like Rust, most people have probably have no reason to use Rust, but if you were going to have to learn C#or learn Rust, it sounds like Rust may be the best choice or or not C#, C++, sorry. Like if you had to do something embedded, you know, it's starting to sound like Rust may be the best choice if you were going to Absolutely. If you never had to go that deep, maybe that's the best choice now. That's not for me because I already know C#, C++, but so to me, I don't see the benefit of learning Rust unless maybe there is something there that I cannot do in C++, right? Yeah. I mean, even tying it back to something that I've had lots of conversations with you about, something as simple as like R versus Python, right? Like they're both similar in a lot of aspects, but they do very different things, right? Python's kind of super set of that. I mean, like just I mean it's it's still claim to fame is it's, you know, probably not even second best, but it's second best at everything, you know, like Yeah. whereas like, you know, there's definitely tools that are better you know, there's better web development tools, there's better this and that, but I mean the if you learn enough Python, you can be dangerous in a lot of things and at the very least prototype with it. Yeah. I want to know your favorite either tool that you've been using forever or new tool that that you've recently come across that you're you're using these days, either cloud, software based, or My favorite that so Power BI for example is one of my favorite tools right now because it's amazing how fast you can you know visualize something and show it to someone and it will save you time versus having to explain something, yeah? To give you an example, we were having a discussion about a new field and whether or not the import of the prior you know third party application was okay. So to take that and generate a field schematic in Power BI was just a few minutes versus us having to present it in a table version and you know just one example. No, for sure. So that's a and it's so easy for anyone to learn it and the desktop version is free, right? So That's So we're, we're, we're currently creating a solution for a client where they did not want to pay for the embedded Power BI to go into the, the application, but they wanted to have the benefit for internal purposes. And they didn't know that if you just wanted to use it internally, you can just run it on your desktop, and it, and it connects to your database, and, and the application didn't really need it like they need it internally. So so so that's a good tool to to know about. Another tool that I've been using for ages is called SQL Compare. It's from Redgate. I just just used that last week. It's yeah. So we and it's one of they have so they have many tools, but this is one of my favorite one, SQL Data Compare and SQL Compare. One of them compares schema, one of them compares data. If you're synchronizing databases or if you have the same database that you're using for multiple clients like we do and you want to make sure that let's say you added a column and then every database It's replicated correctly. Yeah. It it just it a click and you see everything. That's pretty slick. Yeah. This is a great tool. I bet that would be very very useful for operators. we're partners with Redgate. So but there there's there's other competitors out there that do the same thing. Yeah, I mean yeah, I literally just had to use it last week cuz we're going through a migration from one version of our ERP to another and like this new one is kind of a they've actually merged they they had talk about software development practices. They didn't really have the best they had like different code branches for each operator they worked with which was not good. So they're this next version is basically them merging the best of class of all those in, but basically was able to take our, you know, previous version of the database and compare it to, like, the UAT version we have now. And I was able to just basically confirm that, all right, these came across and that the data types were the same and, you know, we could feel good that, you know, there wasn't any breaking changes. They they they also have one of their tools is a dashboard that shows you all your databases between development, testing, and production and shows you if there's any drift. So if one of them starts to drift off of the schema it will it will That's pretty cool. Yeah. Shameless plug to Power BI, cuz I'm also a big Power BI fan boy. And I think it's a great... it's a great introductory[], like all the way from full-blown production, like Talal's doing, to even just, "I'm an engineer and I'm getting interested in, you know, data science or data manipulation or things like[]." I use it all the time for data manipulation because it's easier for me to use their GUI than it is to code it. Right? really good thread I saw last week or whatever so on on the upstream Slack. Um forget is it Zach Warren I think brought it up. But I mean again because Spotify is still pretty ubiquitous and and there's good reasons for it but like at the same time I mean I think and I think one thing to your point about using Microsoft tech if it's not good yet just wait cuz it will be. I mean and like they just rapidly improve things. mean even Power BI in the last three four years like I remember trying it out right after they bought whatever company that was and kind of started distributing it under Power BI Microsoft's license and stuff and it sucked. It was terrible. And even I mean within that year I was using it on a fairly regular basis. You know, their integrations with Python and R and all that stuff like makes it very very handy. Um and very useful in my not not the best for uh high frequency time series data. I will throw that out there unless you want to find a a custom plot, but But I mean even for them, I mean like the visualization stuff is fine. And the thing is it's it's good enough for most. I mean like there's some GIS capabilities that be something to be desired for me and that's why a lot of people haven't made the move, but like just I mean Power Query and then just like the data sets and data flows and like the data engineering behind it. They They can that citizen data engineers or data scientists can use like and be dangerous and add value to the companies and um no it's a great tool. it's catching up. There's a lot of so also full disclosure I'm our BI partner as well. Um we're we've been tracking it since day one. And uh we were using Spotfire for our clients because that's what they had. But then as soon as Power BI started to meet some of these, you know, basic requirements like you said, we immediately switched. And we even developed our own visuals for it, you know, for for what we wanted to do. Um but then before long the built-in visuals met all those missing features so that now the visuals that we developed are maybe not needed anymore, you know, that kind of thing. Um but yes, GIS is one area where it's still kind of behind. Even though they have the ESRI map where but you have to have like a a separate paid, I guess, account for it. And I think I just saw recently do have a map map box at one point had a had a pretty decent plugin, but the problem with like the lot of plugins and you know um but I mean I think got the Azure maps thing and I and I just found out recently I think they're looking at doing multi-layer stuff through Azure maps. I mean it again, if you don't like it, just wait, you know, another year. Look at their That's that's true, but the true power like you said is in the background what happens in with with the data mart, with the data verse, with the Power Query. I mean there's there's a new feature coming out every week you know and we're like we we have to keep up. So when, you know, one of the things that we used to do when Zach had a Zach Warren had this Power BI meetup group, and we would have to spend a lot of time every month to do like a five minutes of what's new in Power BI and try to compress it all in there. That was a lot of work because of how much new features they have released in there. like the tooling around it now with with like tabular editor and and DAX Studio. I mean, there's some really good developer tools, you know, around it too. I mean, I I think on a similar side, I mean, talk about some of the low code just kind of shifting into that. Have you all had to do much with Power Apps or like yeah? So but in in in the sense that if I need something very simple that we're not it's not something that's client facing and we want like, you know, to trigger something or to to monitor process or have something that can interact with. I'm not a huge fan of low code or no code simply because most of what we do is commercial and we have to have full visibility, full control, you know, things like that. But they do come in handy in a lot of like, you know, small scenarios like that. I wouldn't I wouldn't use them though to build an application and I know this may rub some people the wrong way. I'm not going to name any names, Jeff. But I just it may work, right? And you have to put a lot of effort into it. So to take a low code from what I've seen, to take a low code platform and build a complete app on top of it, it's not like it's not like what the brochure says. It's not an overnight thing. Yeah. You have to make you have to work a lot on getting it to look exactly like an like a commercial application and to make sure that you've, you know, secured everything and and be able to onboard new customers on it without a lot of work. So of that, you're still at their mercy. And by the by the time you're all set and done, you're not really that far behind having to build everything on your own where you have so I think Power Apps to me is a really good or, you know, Zapier, insert, whatever connector, trigger workflow you want to use, but to me it's a very good like quick and dirty internal integration or like even just a proof of concept in my mind, right? Like if you are at a big company and you are an engineer and you're trying to figure out this workflow, or you have this idea and then you see that you can do it through Power Apps, set it up, test it, and then give that to a developer who can actually go write it. Right? Like it makes it much more straightforward and having instead of having to like write out, you know, your business case and do it just on paper. It gives it gives you the ability to kind of show them and demonstrate it. We have we we've got five more five 10 more minutes left. Um I want to pivot and ask you about AI, ML, GPT, and all things uh forward facing with that and kind of the energy space. What are you know, what do you see what are you getting questions about? What do you see people doing? What are you what are y'all looking at doing with some of that stuff or Um this is this is one of those things where we're trying to be very cautious and wait as much as we can to make sure that we're not investing in in you know, in a bad way. Here's why. Uh first there's a lot of solutions out there already built, but they don't all scale very well. And when when it comes to all companies, our clients, there's a big difference between a small operator and a big operator and what they can benefit from or afford in terms of these kinds of solutions. So as far as I'm concerned, what I would like to what I'm actually planning on on getting involved in is ways to use the existing data that we have to offer the user some intelligent insights, right? And if you go through our websites and brochures and everything, you won't see a single mention of AI or ML and because we're trying to be very careful. We're we're very serious about what we do and the last thing I want to do is use buzzwords, you know. So I would give you ubiquitous insights that are so seamless that you wouldn't even know that there's an intelligent algorithm behind this that's you know kind of coming up with that but what I like to call it is and this is just a term that I coined kind of gut instinct analytics. So if you have an you're looking at a chart and we gave you an insight and as soon as you see it your gut tells you that this is correct right this is so as an experienced engineer oops I poked the microphone as as an experienced engineer you know that it's telling you this is what happened here you know because we think that there's a hole in the tubing or something like that. So without without trying to sell that aspect so that's that's my stance on it here's why there's a lot of snake oil out there there's a lot of great solutions I and I know some of the people who are developing these great solutions but there's also a lot of snake oil out there and I just never really and and like I said I started a long time ago and I you know we we were writing our own Yeah. you know and you were writing yeah using Lisp and using other types of solutions to you know platforms to to create these algorithms and every time when there was an AI winter was because the there was a lot of buzz Right. And then, and then a lot of products were created, a lot of tools were created, but then they failed to deliver on, you know, on the P L at the end of the day when, when you want to benefit from it. And what happened is the funding dries up, and when the funding dries up, then the research stops, and then 10 years later it starts over again. This time, we're seeing way better, you know, solutions, as you can see, especially with, with the, what you, you know, chat GPT and the self-driving cars and all the different aspects of it. But as far as, as far as the oil and gas industry, there's, there's things, look. So, I've had a customer once ask me, "I want a dashboard that just has green lights, and then when something goes wrong in any of our, I want a red light." And that's as simple as, and do you know the, the, the amount of uh intelligence behind that, to, to replace everything with just one single AI that tells you when something is going wrong across 10,000 wells? And this is, and I want it in a year. Yeah. Under a year. I've also got all these uh I've got 20 years worth of data sets that are all unstructured, undocumented, and have errors full of them. And some of the some of the and some of the data they were getting was unstructured data and the PDFs and the so the expectations and the reality and that's that's something that the Yeah. Yeah, which is which is why we weren't really rushing towards something like this. Yeah, and then you hear about companies that pitch AI or whatever and then it's either a bunch of nested if statements or it's a bunch of people in you know third world country like This is a dog. This is a you heard about the the doorbell where where it turns out to be people classifying things and and then they said we're just training the system for a year. Well, now you've got I saw there's a clothing company that just came out and it's it's it's clothing that disrupts well-known computer vision models. So instead of identifying you as a person, it identifies you as like a banana or whatever like there's now there's this whole counter right like we're swinging both ways now. but to summarize I think there's a lot of promising things you can do and some of companies are already doing them. Uh unfortunately what I've seen so far is that they're not affordable for everyone. They're just you have to be really big to be able to get that kind of and you have to have a massive amount of data. Uh but if we can do something that we can then make it available to everyone without them even realizing that hey we have a machine learning thing running in the background that's giving us because again like I said I don't I don't like to hype and No, for sure. No, no, yeah. Smart I mean Our industry has Well, because then it cheapens the real work that you do as well, right? I mean Yeah. Well, and it I I like it to be like the flash feature in Excel if you've seen it like where Right. They click and it fills down. Yeah, it it learns from what you're doing and then it mimics it without you even realizing that this is Yeah. I will say, pro tip to anybody that's interested in that stuff, even just using ChatGPT to debug and or document code is a very nice little life hack because it it documents code better than any human. What are you using
GitHub Copilot yet? Not yet, no. I I'm keeping an eye on it, but yeah, it's I've seen other similar stuff like very small projects that we're trying to do that before. It's very promising. No, it's it's been pretty good for me. I've talked to a couple of people using totally different language. I, you know, mainly in Python, but another guy who's using Go and it's just like you can literally write comments and it will spit out Yes. you know, pretty much probably 90% functional code. You know, like and then but again, it still takes that subject matter or that expertise to be able to say this is good or bad, you know, but even while you're writing your code, the suggestions have been really impressive, too. I think to me that here's the thing. As as a software developer, of course, I want to do everything I can to make my life easier, but at the end of the day when when let's say you're hiring a new developer, Yeah. right? You want to know that this developer is not starting to depend entirely on code generated by, you know, by an algorithm and because you mean you learn from making mistakes, but you have to learn from your own mistakes. You cannot learn if someone else is making mistakes and you can't catch them, right? So so people are depending on this code being correct and that kind of scares me a little bit. Yeah. And I mean now they're just relying on that and not Stack Overflow, so. That, too, right? So so it's great to have all these tools as long as there's Yes. Yes. You have the expertise to Yeah. All right. We've got a few more minutes, so I'm going to ask kind of our last question and then Bobby we'll get into the speed round. What's one piece of advice you give to people looking at getting into the energy tech community or development community? I would say that wish you knew back in the day. Uh I wish I knew back in the day. Okay. now. Or now. Yeah. Uh, I would say that learning about the the data is just as important as learning how to code. And I see that a lot of people just think that that's secondary or, you know, and I think it's it's just as important if not more important. Yeah. Understanding the data, understanding data structures, understanding why uh you know, why something looks wrong on the chart that you're trying to create or something like that. Cuz you know, I'm assuming that we're talking here to a lot of people who are going not to be just doing software development, right? Because yes, you can be a UI software developer and not understand a thing about databases, but that's not what our audience, right? No. I mean, I agree 100%. I mean, I'd say like our data team at Grayson Mill, all of us including my manager Corey are former reservoir techs. And so we got exposure cuz on the reservoir engineering teams you get exposure to the drilling completions, you know, obviously reservoir engineering, like even geology kind of stuff. So I mean, now when we're actually working data problems across the business, like we understand the why. And I think understanding it, whether you're doing VI you know, type stuff or you know, developing the software to understand the problem is probably the most key. And the second advice I would say is that um try to leave your ego outside the door because that helps you learn and and don't get attached to like one language or you know, that it's not it shouldn't be a personal thing. Right. Yeah, it's a tool to do a job and use like I said, use the best tool. I mean, don't use a wrench as a hammer. Yeah. Yeah, exactly. Even though you do it in the field too, but Um Sometimes you got to do what you got to do, but All right. Um we'll go real quick through the speed round and then uh now we'll just it's been great having you on. Um So what's your favorite cloud to work in? Um Azure. Yeah. Um favorite database? SQL Server especially. Um Open source tool. Open source tool. I would say um this is a tough one. Uh there's so many that I know. Yeah. Uh, I'm hoping that OSDU is going to be, you know, the the yeah, GitHub, a lot of other things that, you know, we use a lot of stuff. But uh, um, there's something there though: when you're developing commercial software, you want to reduce your, um, you know, dependencies uh and footprint. So as much as open source tools are great and they're all over the place, uh we use them maybe to help us internally, but we try not to depend on them in the in the final in the end product cuz that's a scary thing. I mean, you've heard about the one uh script on the you know, NuGet package that brought down half the internet one one time. So and you know, there's ways you can mitigate against that by like hosting everything locally and things like that. We're going to have we're going to have a full open source show with a bunch of libraries. There was there like the NumPy library was calculating things wrong at one point and invalidated a lot of like scientific Yeah. So so you have to be very conservative and very very careful when you're choosing dependencies on other tools and things like that. You want to do one more diamond the last one? Yeah, my my I I really want to know what's your favorite either current or all-time video or board game. Oh man. Um my all-time favorite would have to be StarCraft. Yeah. It's a classic. and if you if you're familiar with it uh we there was this period of frustration and wait for the second one the you know, the second StarCraft to come out and my friends and I this was, you know, maybe I don't know, 10 15 years ago they kept promising and and launch date next year and then pushing it by a year. So this is very relevant to software development cuz now I'm I'm afraid to like, okay, are we going to say our next sprint is going to be in in April because we cannot miss that deadline. So yeah. People will just stop believing you when you when you For sure. Yeah. No, we've been there. StarCraft. Absolutely. Well, thank you so much for for being our first guest. How how can people find you if they want to mi4.com I'm on I'm on LinkedIn. This is the best way to kind of, you know, and my website is mi4.com m i and the number 4 .com Perfect. Appreciate it. Thank you for having me. Yeah, thanks a lot.