Latest / Silicon & Shale Podcast / Episode 9: Acquisition Data, Field AI, and the Automation Build List

Transcript

Episode 9: Acquisition Data, Field AI, and the Automation Build List · Silicon & Shale Podcast

0:00
SPEAKER_01: Seriously, what are we even talking about?
0:02
SPEAKER_01: A gas shortage?
0:04
SPEAKER_01: By 2028?
0:06
SPEAKER_01: You gotta be kidding me.
0:07
SPEAKER_00: Pac, I'll tell you what, I heard that same song and danced back in 08, then again in 14.
0:11
SPEAKER_00: Every time there's a whisper of a cold snap or a new pipeline gets held up in court,
0:15
SPEAKER_00: suddenly the sky is falling,
0:17
SPEAKER_00: and every single time we find more gas,
0:19
SPEAKER_00: or we get better at moving it,
0:20
SPEAKER_00: or somebody drills a well nobody thought would produce.
0:23
SPEAKER_01: Look, have they met the Permian?
0:24
SPEAKER_01: Or the Marcellus?
0:26
SPEAKER_01: We are swimming in natural gas.
0:27
SPEAKER_01: Moving it is a separate problem, but the resource is there.
0:30
SPEAKER_01: This feels like a data problem wearing an energy questions hat.
0:33
SPEAKER_00: A data problem wearing a very expensive energy questions hat, yeah.
0:36
SPEAKER_00: Because every time someone spooks the market with a headline like that, it ripples.
0:40
SPEAKER_00: Investors get nervous, producers pull back, prices jump for a minute.
0:43
SPEAKER_00: Then it all settles down when reality actually happens.
0:47
SPEAKER_01: Right?
0:47
SPEAKER_01: It's almost like they forget the sheer ingenuity of this industry.
0:50
SPEAKER_01: Like, you give a good engineer a problem,
0:52
SPEAKER_01: and they're going to figure out how to get that gas from point A to point B.
0:55
SPEAKER_01: Even if it means building a pipeline with a spork.
0:58
SPEAKER_00: A spork and about 10 miles of duct tape, yeah, that's the spirit.
1:01
SPEAKER_01: Silicon and shale.
1:02
SPEAKER_01: We should probably explain why Derek is arguing with a forecast before breakfast.
1:05
SPEAKER_00: That's right.
1:07
SPEAKER_00: And we're diving straight into it because these forecasts, they keep coming.
1:10
SPEAKER_01: And they keep getting it wrong.
1:11
SPEAKER_01: We've got so much interesting stuff happening,
1:13
SPEAKER_01: from massive tech investments in the energy space to, honestly, wild M&A activity.
1:18
SPEAKER_01: Let's talk about what's actually happening.
1:20
SPEAKER_00: Yeah, let's talk about it.
1:21
SPEAKER_00: Because this gas shortage prediction, it doesn't sit right with me.
1:25
SPEAKER_00: Look at the last few years,
1:26
SPEAKER_00: we've seen record production out of places like the Haynesville,
1:28
SPEAKER_00: the Marcellus, even some of the smaller plays are humming along.
1:31
SPEAKER_00: You think we're suddenly going to run out of places to drill or waste a frack?
1:34
SPEAKER_00: I don't buy it.
1:35
SPEAKER_01: No, I don't either.
1:36
SPEAKER_01: And I think it's illustrative of a broader issue,
1:38
SPEAKER_01: which is that people are still looking at these things through
1:40
SPEAKER_01: a very traditional lens.
1:42
SPEAKER_01: They're not factoring in the actual pace of innovation.
1:45
SPEAKER_01: They're not factoring in how quickly we can optimize existing wells,
1:48
SPEAKER_01: or even how much more efficient our drilling techniques have become.
1:51
SPEAKER_00: You're telling me.
1:52
SPEAKER_00: I remember when a thousand-foot lateral was a big deal,
1:55
SPEAKER_00: Now we're talking about laterals that are two miles long,
1:57
SPEAKER_00: sometimes more,
1:59
SPEAKER_00: and the amount of gas we're pulling out of those,
2:00
SPEAKER_00: it's on a different scale entirely.
2:03
SPEAKER_00: The game has changed, but some of these analysts, they're still playing with an old rulebook.
2:06
SPEAKER_01: Totally.
2:07
SPEAKER_01: Drilling is only the first piece.
2:09
SPEAKER_01: The data carries the advantage, right?
2:10
SPEAKER_01: Better sweet spot predictions can cut dry holes.
2:13
SPEAKER_01: Then you optimize flow rates, trim downtime, and deliver supply more efficiently.
2:16
SPEAKER_01: We still need rocks in the ground, but fiber now has a lot to say about what happens to them.
2:20
SPEAKER_00: Well, at that point, I get a little bit skeptical.
2:24
SPEAKER_00: Because even with all your fancy data, you still got to get a rig out there.
2:26
SPEAKER_00: You still got to get pipe in the ground.
2:28
SPEAKER_00: And then the rubber meets the road pack.
2:30
SPEAKER_00: You can predict all you want,
2:31
SPEAKER_00: but if you don't have the cruise or the sand or the water or the permits,
2:34
SPEAKER_00: that gas ain't going anywhere.
2:35
SPEAKER_01: Okay, but what if the data helps you get those permits faster?
2:38
SPEAKER_01: Or helps you plan sand and water before the schedule forces you to pay a premium?
2:42
SPEAKER_01: Finding gas is one job.
2:43
SPEAKER_01: Streamlining the full run from subsurface to sales point is the larger opportunity.
2:47
SPEAKER_00: Hmm, I'll give you that.
2:49
SPEAKER_00: We've seen some of that for sure.
2:50
SPEAKER_00: Better planning, less waste.
2:52
SPEAKER_00: But it's still a human element to a lot of it.
2:54
SPEAKER_00: You still got a guy in a dozer.
2:55
SPEAKER_00: You still got a roustabout checking gauges.
2:57
SPEAKER_01: For now, yeah.
2:58
SPEAKER_01: But the trend is clear.
2:59
SPEAKER_01: Look at the M&A activity happening right now.
3:02
SPEAKER_01: Magnolia Oil & Gas, they picked up some wildfire energy assets for $650 million cash.
3:07
SPEAKER_01: Matador Resources did two all-cash acquisitions in 2023.
3:11
SPEAKER_01: Advanced Energy Partners for $1.6 billion and Encore Permian for $160 million.
3:16
SPEAKER_01: These aren't random land grabs, Derek.
3:18
SPEAKER_01: These are strategic plays for efficiency, for scale, for consolidating acreage positions.
3:22
SPEAKER_00: Yeah, no, that's a good point.
3:24
SPEAKER_00: And the cash part is interesting.
3:25
SPEAKER_00: It means they're confident in their balance sheets,
3:27
SPEAKER_00: confident in the assets they're buying,
3:29
SPEAKER_00: not taking on a ton of debt in a higher interest rate environment.
3:32
SPEAKER_00: Smart.
3:32
SPEAKER_01: I'll give you that.
3:34
SPEAKER_01: At that scale, especially with all cash deals, buyers need value and synergies quickly.
3:38
SPEAKER_01: At that point, the tech earns its keep.
3:40
SPEAKER_01: How fast can you integrate the new assets and get them onto the production
3:43
SPEAKER_01: and data systems you already use?
3:45
SPEAKER_00: Ah, see, now you're talking my language.
3:47
SPEAKER_00: Because that integration pack, this is the point where the wheels usually fall off.
3:50
SPEAKER_00: You buy a new package of leases,
3:51
SPEAKER_00: you get a bunch of their old production data,
3:53
SPEAKER_00: maybe some handwritten notes,
3:54
SPEAKER_00: some PDFs that look like they were scanned on a potato,
3:57
SPEAKER_00: and then you gotta figure out what's what,
3:58
SPEAKER_00: what's producing, what's not, what's got a good title, what's got a mess.
4:02
SPEAKER_00: The useful part starts where the real work starts.
4:04
SPEAKER_01: That work should be automated.
4:06
SPEAKER_01: If you're spending hundreds of millions,
4:07
SPEAKER_01: maybe billions, you need to process acquired data immediately.
4:11
SPEAKER_01: Find the synergies,
4:11
SPEAKER_01: redundancies,
4:12
SPEAKER_01: and optimization opportunities before closing,
4:15
SPEAKER_01: or at least within days instead of months.
4:16
SPEAKER_00: Months is optimistic sometimes.
4:18
SPEAKER_00: I've seen it take a year to get all the production curves aligned properly.
4:22
SPEAKER_00: Different reporting standards, different software, different ways of naming wells.
4:25
SPEAKER_00: It's a headache.
4:26
SPEAKER_01: Which is why SLB's investment makes so much sense.
4:29
SPEAKER_01: Schlumberger, right?
4:30
SPEAKER_01: They're pouring resources into building out their digital infrastructure.
4:33
SPEAKER_01: Data centers for high-performance computing and AI applications and energy.
4:37
SPEAKER_01: They're not selling software anymore.
4:39
SPEAKER_01: They're building the literal backbone for the next generation of energy operations.
4:42
SPEAKER_00: SLB, yeah.
4:43
SPEAKER_00: They've always been ahead of the curve on the tech side.
4:45
SPEAKER_00: I remember when they first started pushing their seismic processing stuff,
4:48
SPEAKER_00: everyone thought it was black magic.
4:50
SPEAKER_00: Now it's standard, but HPC and AI, that's a whole other level.
4:53
SPEAKER_00: What are they actually doing with it?
4:55
SPEAKER_01: They're building the foundational layer for operators to run their entire business on.
4:59
SPEAKER_01: Think about it.
5:00
SPEAKER_01: Massive subsurface models,
5:01
SPEAKER_01: real-time wellbore optimization,
5:03
SPEAKER_01: predictive maintenance for every piece of equipment,
5:06
SPEAKER_01: Even automating compliance reporting.
5:08
SPEAKER_01: All of that takes immense computational power and sophisticated AI.
5:11
SPEAKER_01: They're essentially building the cloud infrastructure for the oil and gas industry.
5:14
SPEAKER_00: So instead of every operator building their own data center,
5:17
SPEAKER_00: which is what some of the bigger guys tried to do,
5:19
SPEAKER_00: SLB is saying, hey, we'll build it, you plug in.
5:22
SPEAKER_01: Okay, it's a platform play.
5:24
SPEAKER_01: Smaller and midsize operators can tap the same computing power and AI capability
5:28
SPEAKER_01: without the huge upfront investment.
5:30
SPEAKER_01: That puts advanced tools within reach of more than the super majors.
5:33
SPEAKER_00: That's a good point.
5:34
SPEAKER_00: Because the mid-majors, the privates, they're the ones always scrambling for an edge.
5:38
SPEAKER_00: They can't afford a whole data science department.
5:40
SPEAKER_00: They're running lean.
5:41
SPEAKER_01: Right.
5:42
SPEAKER_01: Oil and gas isn't the only use.
5:43
SPEAKER_01: Dr. Neeraj Gupta at Battelle is working on carbon management and CCS research,
5:48
SPEAKER_01: including offshore carbon storage with U.S. Department of Energy involvement.
5:51
SPEAKER_01: That takes complex data-heavy modeling.
5:53
SPEAKER_00: CCS, yeah.
5:55
SPEAKER_00: Carbon capture.
5:56
SPEAKER_00: That's a whole different ballgame.
5:57
SPEAKER_00: You're trying to figure out how to put something back in the ground safely and permanently.
6:01
SPEAKER_00: That's a lot of geology, a lot of fluid dynamics.
6:04
SPEAKER_01: And a lot of data.
6:05
SPEAKER_01: Those reservoirs need integrity monitoring for decades.
6:07
SPEAKER_01: AI and high-performance computing become critical because nobody is checking
6:10
SPEAKER_01: an offshore storage site with a dipstick.
6:12
SPEAKER_01: You need continuous autonomous monitoring.
6:14
SPEAKER_00: No, you definitely don't want a guy with a dipstick offshore.
6:17
SPEAKER_00: I've seen some pretty creative field solutions in my time, but that's a new one.
6:20
SPEAKER_01: That tracks.
6:22
SPEAKER_01: Conventional production,
6:23
SPEAKER_01: acquired assets, and CCS all need serious digital infrastructure and AI.
6:27
SPEAKER_01: Digital capability has moved from a nice extra to basic operating infrastructure.
6:31
SPEAKER_00: I guess so.
6:32
SPEAKER_00: I still think about the guys out there, the drilling foreman, the pumper.
6:36
SPEAKER_00: They're still the ones making it happen.
6:37
SPEAKER_00: But I can see how having better tools,
6:39
SPEAKER_00: better information makes their job easier, or at least makes them more effective.
6:43
SPEAKER_01: More effective, safer, more efficient.
6:46
SPEAKER_01: That's the practical payoff.
6:47
SPEAKER_01: Good tools give people leverage instead of replacing them.
6:50
SPEAKER_00: Yeah, Baker Hughes definitely keeps a close eye on producer spending.
6:53
SPEAKER_00: It's their business to know.
6:54
SPEAKER_00: And it does create a cautious environment.
7:13
SPEAKER_01: But that also means the companies that are spending,
7:15
SPEAKER_01: like Magnolia and Matador,
7:17
SPEAKER_01: are doing it very strategically, looking for assets they can immediately optimize.
7:21
SPEAKER_01: It reinforces the need for that digital edge.
7:23
SPEAKER_00: Right.
7:24
SPEAKER_00: If you're going to spend a dollar, make it a smart dollar.
7:26
SPEAKER_00: Better data helps separate a useful investment from an expensive science project.
7:30
SPEAKER_01: True.
7:31
SPEAKER_01: Speaking of smart money,
7:32
SPEAKER_01: Navis sponsors some industry podcasts and is pushing laser scanning
7:35
SPEAKER_01: and digital twin technology.
7:37
SPEAKER_01: You can create a detailed digital copy of an acquired facility,
7:39
SPEAKER_01: walk through it virtually,
7:41
SPEAKER_01: measure it, and plan upgrades without traveling to the site.
7:43
SPEAKER_00: Now that's pretty wild.
7:44
SPEAKER_00: A digital twin.
7:46
SPEAKER_00: So a copy of a refinery, a well pad, or a pipeline?
7:49
SPEAKER_00: You could see clearly where everything sits and what condition it's in, all from your desk.
7:53
SPEAKER_01: Right.
7:53
SPEAKER_01: And not a static copy.
7:55
SPEAKER_01: Sensors feed pressure, temperature, flow rates, and equipment status into it in real time.
7:59
SPEAKER_01: You can run simulations, predict failures, and plan maintenance.
8:03
SPEAKER_01: That could change asset management in a practical way.
8:05
SPEAKER_00: So if I buy a new gas plant,
8:06
SPEAKER_00: I don't have to send a crew out there with clipboards and tape measures to map
8:09
SPEAKER_00: the whole thing?
8:10
SPEAKER_00: I get their NavVis file and I'm good to go?
8:12
SPEAKER_01: Pretty much.
8:13
SPEAKER_01: You can integrate it with your existing systems,
8:15
SPEAKER_01: identify bottlenecks, plan out your capital projects with incredible accuracy.
8:19
SPEAKER_01: No more, wait, that pipe's actually two feet longer than the old blueprint said.
8:24
SPEAKER_01: That kind of stuff costs millions in delays.
8:26
SPEAKER_00: Millions, yeah, or a whole lot of headaches.
8:28
SPEAKER_00: I remember one time we bought a small field and the schematics,
8:31
SPEAKER_00: they were just wrong, completely wrong.
8:34
SPEAKER_00: We ended up having to reroute a flowline because
8:35
SPEAKER_00: the original drawing showed it going through solid rock where it was actually dirt.
8:39
SPEAKER_00: Cost us a fortune in three weeks.
8:41
SPEAKER_00: A digital twin would have saved us that.
8:43
SPEAKER_00: I'm actually a little impressed by that.
8:44
SPEAKER_01: See?
8:45
SPEAKER_01: I knew I'd get you eventually.
8:47
SPEAKER_01: The goal is catching expensive physical problems before they happen.
8:50
SPEAKER_01: SLB is building toward that, and companies like NavVis help enable it.
8:54
SPEAKER_01: Tangible return is the test, not hype.
8:56
SPEAKER_00: Okay, okay.
8:57
SPEAKER_00: I'll admit that digital twin thing, that's got some legs.
9:00
SPEAKER_00: Especially for acquired assets where the original documentation is, shall we say, aspirational.
9:04
SPEAKER_01: Aspirational.
9:05
SPEAKER_01: I like that.
9:07
SPEAKER_01: On the E&P side,
9:08
SPEAKER_01: consolidation is pushing operators to use better tools on
9:10
SPEAKER_01: the reserves they already own.
9:12
SPEAKER_01: The operating goal is simple.
9:14
SPEAKER_01: Recover more value and waste less time doing it.
9:16
SPEAKER_00: Yeah, I can see that.
9:17
SPEAKER_00: The game has always been squeezing value out of every asset.
9:20
SPEAKER_00: Now there are more tools on the bench.
9:22
SPEAKER_01: Right.
9:22
SPEAKER_01: And those tools are getting more powerful by the week.
9:25
SPEAKER_00: Speaking of powerful tools, you've been talking my ear off about these AI agents building apps.
9:29
SPEAKER_00: Last week you mentioned it.
9:30
SPEAKER_00: What's the deal with that?
9:31
SPEAKER_00: Are we all out of a job now?
9:33
SPEAKER_01: No, no, definitely not out of a job.
9:35
SPEAKER_01: But it's wild, Derek.
9:36
SPEAKER_01: Corey Knowles and Grant Harvey,
9:37
SPEAKER_01: they did a live walkthrough on how to publish an AI-built iPhone app to
9:40
SPEAKER_01: the Apple App Store.
9:42
SPEAKER_01: The big takeaway was that building the app with AI is actually getting incredibly easy.
9:46
SPEAKER_01: The challenge is still getting it approved by Apple.
9:48
SPEAKER_00: So the AI can make the thing, but it can't deal with the bureaucracy.
9:51
SPEAKER_00: Sounds about right.
9:52
SPEAKER_01: The AI can write code, design a user interface, and generate content quickly.
9:56
SPEAKER_01: Then it hits human review, app store rules, and security checks.
10:00
SPEAKER_01: People are still crucial there.
10:02
SPEAKER_01: Remember the barbecue app I pitched for your midnight build list?
10:04
SPEAKER_01: The one that tracks brisket availability?
10:06
SPEAKER_00: Oh, the good BBQ joints between here and Midland.
10:08
SPEAKER_00: I'd still pay good money for that pack.
10:10
SPEAKER_01: An AI agent could build that app far faster than a traditional project.
10:14
SPEAKER_01: The breakthrough is its ability to understand the request,
10:16
SPEAKER_01: break it into parts, and produce a usable application.
10:19
SPEAKER_00: So if I wanted an app that, say,
10:20
SPEAKER_00: tracks every single permit application in Pecos County in real time
10:23
SPEAKER_00: and flags any potential issues, an AI could just whip that up?
10:26
SPEAKER_01: It could draft it, yeah.
10:28
SPEAKER_01: The front end, the back end, how it connects to the data sources.
10:31
SPEAKER_01: The actual data sources and getting access to them,
10:33
SPEAKER_01: that's still the hard part for a lot of things.
10:35
SPEAKER_01: But the app itself, it's becoming a commodity.
10:38
SPEAKER_01: This also ties into what Deep Ubi,
10:40
SPEAKER_01: Global Director of Technology for Startups at AWS, was talking about.
10:44
SPEAKER_01: It's the journey from an AI-generated prototype to secure, reliable, production-ready software.
10:48
SPEAKER_01: That's the gap.
10:49
SPEAKER_00: Okay, so it can build a sketch, but can it build a skyscraper?
10:52
SPEAKER_00: That's what I'm hearing.
10:53
SPEAKER_01: Sure.
10:54
SPEAKER_01: AWS Startup Advisor is an AI builder assistant that brings AWS architecture
10:58
SPEAKER_01: and practices into the process.
11:00
SPEAKER_01: It helps close the gap between a cool prototype and software that works reliably
11:03
SPEAKER_01: at scale, so teams can ship and maintain what they build.
11:06
SPEAKER_00: So it's like having an architect who knows all the building codes
11:08
SPEAKER_00: and structural engineering rules, even if the AI drew the initial floor plan.
11:12
SPEAKER_01: That comparison holds up.
11:14
SPEAKER_01: It helps make generated code secure, scalable, and suitable for enterprise use.
11:19
SPEAKER_01: Producing something is one thing.
11:20
SPEAKER_01: Producing it correctly is the standard.
11:22
SPEAKER_00: That's the key, isn't it?
11:24
SPEAKER_00: Doing it right.
11:25
SPEAKER_00: Because in our world,
11:26
SPEAKER_00: doing it wrong can mean a lot of zeros at the end of a very unhappy number.
11:29
SPEAKER_01: Agreed.
11:30
SPEAKER_01: And that leads right into this huge debate happening in the AI world.
11:34
SPEAKER_01: Open weight versus closed AI models.
11:36
SPEAKER_01: Dario Amode from Anthropic and Jensen Huang from NVIDIA are kind of the figureheads here.
11:41
SPEAKER_01: There was even a letter,
11:42
SPEAKER_01: Open weight's an American AI leadership,
11:44
SPEAKER_01: signed by over 230 companies and researchers,
11:47
SPEAKER_01: raising concerns about premature restrictions on open AI development.
11:51
SPEAKER_00: Open weight, closed model.
11:53
SPEAKER_00: What's the difference, PAC?
11:54
SPEAKER_00: Is it like open source software versus proprietary?
11:57
SPEAKER_01: Pretty much.
11:58
SPEAKER_01: Open weight means the underlying model,
11:59
SPEAKER_01: the weights that define its intelligence, are publicly accessible.
12:03
SPEAKER_01: Anyone can download it, inspect it, fine-tune it, use it.
12:07
SPEAKER_01: Closed models are proprietary.
12:09
SPEAKER_01: Only the company that built them has access to the inner workings.
12:12
SPEAKER_00: So, open weight, everyone can tinker with it.
12:14
SPEAKER_00: Closed, only the guys who built it know how it works.
12:17
SPEAKER_01: Yep.
12:18
SPEAKER_01: The argument for open weight is that it fosters innovation, transparency, and democratizes AI.
12:23
SPEAKER_01: Anyone can build on top of it, audit it for biases or safety issues.
12:27
SPEAKER_01: The argument for closed models often revolves around safety and control.
12:30
SPEAKER_01: If it's too powerful, you don't want anyone having the keys.
12:33
SPEAKER_00: I can see both sides of that.
12:35
SPEAKER_00: On one hand, you want everyone to be able to build cool stuff.
12:37
SPEAKER_00: On the other hand,
12:38
SPEAKER_00: if these things are getting really smart,
12:40
SPEAKER_00: you don't want some Yahoo to accidentally or on purpose break something important,
12:43
SPEAKER_00: like, I don't know, the grid.
12:45
SPEAKER_01: I can see why.
12:46
SPEAKER_01: OpenAI and Anthropic have even collaborated on some lobbying efforts
12:49
SPEAKER_01: in Washington around AI regulation.
12:51
SPEAKER_01: They want to balance innovation and safety and shape the rules before governments act alone.
12:55
SPEAKER_00: So the guys building the tech are also trying to write the rules.
12:57
SPEAKER_00: That's interesting.
12:59
SPEAKER_00: You think they can be objective?
13:00
SPEAKER_01: Nobody has a clean answer yet, but they're engaging.
13:03
SPEAKER_01: The race also extends beyond U.S. companies.
13:05
SPEAKER_01: Moonshot AI from China is expanding its Kimi model globally,
13:08
SPEAKER_01: so this is a worldwide contest rather than a Silicon Valley debate.
13:11
SPEAKER_00: Worldwide, huh?
13:12
SPEAKER_00: So if these models are open,
13:13
SPEAKER_00: does that mean our competitors overseas can take our tech and run with it?
13:17
SPEAKER_01: Potentially, yeah.
13:18
SPEAKER_01: That's part of the concern for some.
13:19
SPEAKER_01: But the counterargument is that if you don't keep up with open-source development,
13:22
SPEAKER_01: you'll be left behind anyway.
13:24
SPEAKER_01: It's a complex dynamic.
13:25
SPEAKER_01: What's clear is that the pace of development is insane.
13:28
SPEAKER_00: No kidding, I'm trying to keep up with my email sometimes.
13:31
SPEAKER_01: Trust me, I get it.
13:32
SPEAKER_01: Forget the big models for a minute.
13:34
SPEAKER_01: The practical applications are more interesting.
13:36
SPEAKER_01: Melissa Tokmak,
13:37
SPEAKER_01: founder and CEO of Netic,
13:39
SPEAKER_01: described an autonomous platform that manages services such as home repairs
13:42
SPEAKER_01: and pet care.
13:43
SPEAKER_01: If your air conditioner breaks, an agent can find and schedule the repair.
13:46
SPEAKER_00: So an AI that calls the plumber, now that's useful.
13:49
SPEAKER_00: I could use one of those.
13:50
SPEAKER_01: Right?
13:51
SPEAKER_01: It's taking the idea of AI agents that we talked about last week,
13:54
SPEAKER_01: moving beyond office workflows,
13:55
SPEAKER_01: and putting them into the real world, solving real problems.
13:59
SPEAKER_01: It's like having a hyper-efficient personal assistant
14:01
SPEAKER_01: for every single mundane task you can imagine.
14:03
SPEAKER_00: Okay, but what if the AI calls a bad plumber or one who charges too much?
14:07
SPEAKER_00: Can it negotiate?
14:08
SPEAKER_01: That's the goal.
14:09
SPEAKER_01: The agent learns from prior interactions, ratings and pricing.
14:12
SPEAKER_01: It weighs speed,
14:13
SPEAKER_01: cost and quality against your preferences,
14:15
SPEAKER_01: then handles the calls, scheduling and payment.
14:18
SPEAKER_00: So now we're ranking plumbers instead of prospects.
14:20
SPEAKER_00: I'm starting to see a pattern.
14:21
SPEAKER_01: You are.
14:22
SPEAKER_01: And it's already happening in enterprises.
14:24
SPEAKER_01: Good Karma brands, they deployed ChatGPT Enterprise to all 650 teammates on Labor Day 2023.
14:29
SPEAKER_01: Guess what?
14:30
SPEAKER_01: 65% of the company now uses AI daily,
14:33
SPEAKER_01: and they're reportedly saving 10 to 15 hours per week per teammate.
14:36
SPEAKER_00: 15 hours a week?
14:37
SPEAKER_00: That's almost two full days.
14:39
SPEAKER_00: What are they doing writing all their emails with it?
14:41
SPEAKER_01: More than that.
14:42
SPEAKER_01: It can summarize meetings, draft reports, do market research, and handle repetitive tasks.
14:47
SPEAKER_01: The win is removing mundane cognitive work so people can spend time on higher value decisions.
14:51
SPEAKER_01: That's measurable return.
14:53
SPEAKER_00: Okay, that's actually pretty impressive.
14:55
SPEAKER_00: I'll admit it.
14:55
SPEAKER_00: 15 hours a week is not insignificant.
14:57
SPEAKER_00: That's a lot of time.
14:58
SPEAKER_01: It also raises governance questions.
15:00
SPEAKER_01: Dr. Marcus Schmidberger, founder of Junto AI, challenges the need for a single chief AI officer.
15:06
SPEAKER_01: One executive can't oversee all of it.
15:08
SPEAKER_01: AI literacy and responsible use have to spread through the organization.
15:11
SPEAKER_00: So instead of one guy in a fancy office telling everyone how to use AI,
15:14
SPEAKER_00: it's more like teaching everyone how to fish with a new kind of net.
15:17
SPEAKER_01: I'll buy that.
15:19
SPEAKER_01: Darren Ward, a Chief Strategic Innovation Officer, also emphasizes executive ownership.
15:23
SPEAKER_01: Success starts when leadership commits to integrating AI
15:26
SPEAKER_01: into core business strategy, not when somebody buys another stack of licenses.
15:30
SPEAKER_00: That makes sense.
15:31
SPEAKER_00: If the boss isn't bought in, nobody else is going to take it seriously.
15:34
SPEAKER_00: It'll be another piece of software gathering dust.
15:36
SPEAKER_01: Right.
15:37
SPEAKER_01: And then there's the humanoid robots.
15:39
SPEAKER_01: Dr. Peter Fankhauser from Anybotics, they do four-legged robots.
15:42
SPEAKER_01: But Bernd Burnish,
15:43
SPEAKER_01: CEO of One X Technologies,
15:45
SPEAKER_01: says their NEO humanoid robot is planned to ship by 2026 as an open platform.
15:50
SPEAKER_01: Think about that.
15:51
SPEAKER_01: A general-purpose robot that can do human-like tasks, available as an open platform.
15:56
SPEAKER_00: A humanoid robot?
15:57
SPEAKER_00: So like C-3PO, but for the oil field?
15:59
SPEAKER_00: I don't know, Pac.
16:00
SPEAKER_00: I've seen some of those videos.
16:01
SPEAKER_00: They mostly fall over.
16:02
SPEAKER_01: They're getting much, much better.
16:04
SPEAKER_01: And the vision isn't for the oil field,
16:05
SPEAKER_01: though imagine one doing inspections in hazardous environments.
16:08
SPEAKER_01: It's for manufacturing, logistics, even home assistance.
16:12
SPEAKER_01: An open platform means developers can build applications for it, like for an iPhone.
16:16
SPEAKER_00: Okay, so that's a one on the clipboard to fully autonomous scale, right?
16:20
SPEAKER_00: Maybe a two.
16:21
SPEAKER_00: Falling over is not fully autonomous.
16:23
SPEAKER_01: No, no.
16:24
SPEAKER_01: I'd give the vision of a humanoid robot shipping as an open platform a solid eight.
16:28
SPEAKER_01: The current reality, maybe a four.
16:30
SPEAKER_01: But the potential for that kind of general-purpose physical automation...
16:34
SPEAKER_01: That's huge.
16:35
SPEAKER_01: It's the physical manifestation of those AI agents we were talking about.
16:38
SPEAKER_00: Eight, huh?
16:39
SPEAKER_00: I'm telling you I'll believe it when I see a robot change out a pump jack motor in
16:42
SPEAKER_00: a blizzard in North Dakota.
16:43
SPEAKER_00: Then we can talk about an eight.
16:45
SPEAKER_01: Added to the midnight build list, Derek.
16:47
SPEAKER_01: Robot pump jack mechanic.
16:48
SPEAKER_01: You heard it here first.
16:50
SPEAKER_00: You keep adding to that list pack.
16:51
SPEAKER_00: I'm one bad PDF scan away from retirement and you're out here building robot mechanics.
16:55
SPEAKER_01: Hey, you never know when you'll need one.
16:58
SPEAKER_01: Yeah, I'm seeing that.
16:58
SPEAKER_01: This sounds less like simple automation and more like rebuilding the work.
17:01
SPEAKER_01: That's a bigger bite than I expected.
17:02
SPEAKER_01: Way bigger.
17:19
SPEAKER_00: So let's talk about where silicon meets shale in a very real way.
17:23
SPEAKER_00: You mentioned all that M&A activity, Magnolia, Matador, all these companies buying up assets.
17:28
SPEAKER_00: And you mentioned SLB building out this digital infrastructure and NavVis with digital twins.
17:33
SPEAKER_01: Right.
17:34
SPEAKER_01: Massive capital changing hands and an increasing capability to manage
17:36
SPEAKER_01: and optimize those assets digitally.
17:38
SPEAKER_00: Okay, so here's the problem.
17:40
SPEAKER_00: When an operator acquires a new set of assets,
17:41
SPEAKER_00: whether it's a whole field or a few leases, they're not buying wells.
17:44
SPEAKER_00: They're buying a pile of data.
17:46
SPEAKER_00: And usually that pile is a mess.
17:48
SPEAKER_00: It's different formats,
17:49
SPEAKER_00: different systems,
17:50
SPEAKER_00: different levels of detail, some of its papers, some of its old digital files.
17:54
SPEAKER_00: And integrating that data, making it useful, that's a huge bottleneck.
17:57
SPEAKER_00: It costs a ton of money,
17:58
SPEAKER_00: takes forever,
17:59
SPEAKER_00: and often leads to missed opportunities because they can't get a clear picture
18:01
SPEAKER_00: of what they actually own.
18:03
SPEAKER_01: Oh, I've seen that mess.
18:04
SPEAKER_01: Production data is only the beginning.
18:05
SPEAKER_01: You also have permits,
18:06
SPEAKER_01: land records, contracts, environmental reports, and everything tying them together.
18:10
SPEAKER_00: True.
18:10
SPEAKER_00: Until that's sorted,
18:11
SPEAKER_00: you can't tell which wells to re-complete, shut in, or evaluate for infill drilling.
18:15
SPEAKER_00: You're flying blind, or at least looking through a very smudged windshield.
18:18
SPEAKER_01: So, the vision is, what if we could automate that entire data integration process?
18:23
SPEAKER_01: From the moment a deal is announced,
18:25
SPEAKER_01: or even before,
18:25
SPEAKER_01: using AI to ingest, normalize, and reconcile all that disparate asset data.
18:29
SPEAKER_00: That's the dream pack.
18:30
SPEAKER_00: Imagine if you could acquire a new field,
18:32
SPEAKER_00: and within days,
18:33
SPEAKER_00: not months,
18:34
SPEAKER_00: have a fully integrated,
18:35
SPEAKER_00: cleaned-up dataset that plugs right into your existing operational systems.
18:39
SPEAKER_00: No more armies of analysts manually transcribing production reports from old spreadsheets.
18:43
SPEAKER_00: No more landmen digging through scanned images of deeds to confirm ownership.
18:47
SPEAKER_01: Yes.
18:48
SPEAKER_01: Okay, so you've got this massive influx of documents, right?
18:52
SPEAKER_01: Structured data, unstructured text, images, maps.
18:56
SPEAKER_01: You throw it all into an AI-powered ingestion engine.
18:58
SPEAKER_01: The AI identifies what each document is,
19:01
SPEAKER_01: extracts key entities,
19:02
SPEAKER_01: well names, lease numbers, operators, effective dates, production volumes.
19:07
SPEAKER_01: It normalizes units,
19:08
SPEAKER_01: flags discrepancies,
19:09
SPEAKER_01: and even cross-references data points to build a holistic picture.
19:12
SPEAKER_00: So the system has to understand the material, not merely read it.
19:16
SPEAKER_00: It needs to connect a lease with the right well, royalty owner, and production volume.
19:20
SPEAKER_01: Right.
19:20
SPEAKER_01: It builds a connected operating picture.
19:22
SPEAKER_01: A living database where information about the acquired asset is linked.
19:26
SPEAKER_01: Ask for a wells net revenue interest and the system can pull the lease,
19:29
SPEAKER_01: division order, and production statement together.
19:31
SPEAKER_00: That would be a game changer.
19:32
SPEAKER_00: Seriously.
19:33
SPEAKER_00: Because right now that's days of work, weeks sometimes, for one asset.
19:38
SPEAKER_00: Multiply that by hundreds of wells in an acquisition and you're talking about
19:41
SPEAKER_00: a significant drag on getting that value.
19:43
SPEAKER_01: And not for the initial integration.
19:44
SPEAKER_01: Imagine that connected operating picture constantly updating.
19:47
SPEAKER_01: As new production data comes in,
19:49
SPEAKER_01: as new permits are filed,
19:50
SPEAKER_01: as regulatory changes occur,
19:51
SPEAKER_01: The AI automatically updates the digital twin of that acquired asset.
19:55
SPEAKER_01: So you always have the most current, accurate picture.
19:57
SPEAKER_00: So it's like the NavVis digital twin,
19:59
SPEAKER_00: but for all the paperwork and data behind the physical asset,
20:02
SPEAKER_00: a digital twin of the data itself.
20:04
SPEAKER_01: A data twin.
20:05
SPEAKER_01: I love that.
20:06
SPEAKER_01: And then, once you have that, you can deploy other AI agents on top of it.
20:10
SPEAKER_01: Agents that can automatically identify underperforming wells based on historical data.
20:14
SPEAKER_01: Agents that can flag potential title issues by comparing deed language
20:16
SPEAKER_01: to current state statutes.
20:18
SPEAKER_01: Agents that can predict maintenance needs for equipment based on operating parameters.
20:22
SPEAKER_00: So the system goes beyond cleanup.
20:24
SPEAKER_00: It helps run the business, spots opportunities and flags risks.
20:27
SPEAKER_00: That's much more useful than a fancy spreadsheet.
20:29
SPEAKER_01: Exactly.
20:30
SPEAKER_01: It frees landmen,
20:31
SPEAKER_01: engineers and analysts to make decisions,
20:33
SPEAKER_01: negotiate deals and design completions instead of spending half
20:35
SPEAKER_01: the week copying data between systems.
20:38
SPEAKER_00: I've seen guys spend days trying to reconcile production numbers between two different systems.
20:42
SPEAKER_00: A turbocharger that never sleeps and never makes a typo.
20:44
SPEAKER_00: And it learns.
21:08
SPEAKER_01: So, the next acquisition, it's even faster, even more accurate.
21:12
SPEAKER_01: It's a continuous improvement loop.
21:13
SPEAKER_00: That's pretty powerful, Pac.
21:15
SPEAKER_00: I'll give you that.
21:16
SPEAKER_00: That's a collision zone I could get behind.
21:18
SPEAKER_00: It solves a real painful problem that I've seen operators struggle with for decades.
21:22
SPEAKER_01: For M&A integration, that would remove one of the ugliest bottlenecks in the business.
21:26
SPEAKER_01: And the tech is getting there.
21:28
SPEAKER_01: We're seeing the pieces now with the latest practical AI systems.
21:31
SPEAKER_01: This has moved beyond science fiction.
21:33
SPEAKER_00: So what does all this mean for the folks out in the field,
21:35
SPEAKER_00: the landmen, the completion engineers, the pumpers?
21:37
SPEAKER_00: Is this more corporate talk or is it actually changing their day to day?
21:40
SPEAKER_01: It's actually changing it, Derek.
21:42
SPEAKER_01: Think about it.
21:43
SPEAKER_01: If you're a completion engineer and the data on your wells is cleaner,
21:46
SPEAKER_01: more accurate and integrated in real time, your simulations are better.
21:50
SPEAKER_01: Your frack designs are more precise.
21:51
SPEAKER_01: You're not guessing as much.
21:53
SPEAKER_01: That means fewer missteps, better production.
21:55
SPEAKER_00: Yeah, I can see that.
21:56
SPEAKER_00: Less time chasing down bad data, more time actually engineering.
21:59
SPEAKER_00: That's a good deal.
22:00
SPEAKER_01: And for the landman,
22:01
SPEAKER_01: if those title chains and lease agreements are being processed and cross-referenced
22:04
SPEAKER_01: by AI, they're not spending hours on the basic grunt work.
22:07
SPEAKER_01: They're focusing on the complex negotiations,
22:09
SPEAKER_01: the tricky curative issues,
22:10
SPEAKER_01: the stuff that really requires human judgment and experience.
22:13
SPEAKER_01: It elevates their role.
22:14
SPEAKER_00: That's a big one.
22:15
SPEAKER_00: Because a lot of what a landman does,
22:17
SPEAKER_00: especially in due diligence for an acquisition,
22:19
SPEAKER_00: is sifting through mountains of paperwork.
22:22
SPEAKER_00: If an AI can do the sifting, that frees them up to do the thinking.
22:25
SPEAKER_00: That's a huge shift.
22:26
SPEAKER_00: I've burned whole afternoons reconciling inherited lease files
22:29
SPEAKER_00: where every spreadsheet used a different name for the same well.
22:31
SPEAKER_00: If AI could clean that up in an hour, I'd spend the saved time on the judgment calls.
22:35
SPEAKER_00: Or fishing.
22:36
SPEAKER_00: Depends on the weather.
22:37
SPEAKER_01: Now you're talking.
22:38
SPEAKER_01: For pumpers, imagine predictive maintenance on every major piece of equipment.
22:42
SPEAKER_01: An alert arrives a week before failure and schedules maintenance for Tuesday.
22:45
SPEAKER_01: Fewer emergency calls, less downtime, and safer operations.
22:49
SPEAKER_00: Less 2 a.m. calls in a blizzard, you mean.
22:51
SPEAKER_00: That would be a godsend.
22:52
SPEAKER_00: Seriously, that's tangible.
22:53
SPEAKER_01: Right?
22:54
SPEAKER_01: Better tools can make those jobs more strategic, efficient, and less stressful.
22:58
SPEAKER_01: The overhyped claim says AI replaces everybody.
23:01
SPEAKER_01: The practical version helps people do their jobs better.
23:04
SPEAKER_00: All right, let me make you earn the brisket.
23:06
SPEAKER_00: We keep saying somebody could build this acquisition app.
23:08
SPEAKER_00: What would the first useful version actually do on Monday morning?
23:11
SPEAKER_01: Start with one acquired field, not the whole company.
23:14
SPEAKER_01: Give the app a read-only copy of the lease files,
23:16
SPEAKER_01: well list,
23:17
SPEAKER_01: production exports, division orders, permits, and the buyer's naming conventions.
23:21
SPEAKER_01: Its first job is matching records that appear to describe the same well,
23:24
SPEAKER_01: tract, owner, or agreement.
23:26
SPEAKER_00: Appear to describe?
23:27
SPEAKER_00: I like that wording.
23:29
SPEAKER_00: No robot gets to declare that Bill Smith and William J.
23:31
SPEAKER_00: Smith are the same man because both of them once lived near Odessa.
23:34
SPEAKER_01: Correct.
23:35
SPEAKER_01: Every match gets a confidence level and a link back to the source.
23:38
SPEAKER_01: High-confidence clerical matches can be grouped for review.
23:41
SPEAKER_01: Conflicts and weak matches go into an exception queue for a landman or analyst.
23:45
SPEAKER_00: An exception queue sounds much better than my old system.
23:47
SPEAKER_01: Which was?
23:48
SPEAKER_00: A yellow legal pad with three columns.
23:50
SPEAKER_00: Fix today, fix later, and ask somebody who retired in 2019.
23:53
SPEAKER_01: That third column has remarkable staying power across the industry.
23:57
SPEAKER_00: So does the spreadsheet named Final, then Final 2.
24:00
SPEAKER_00: Then Final 2 used this one.
24:01
SPEAKER_00: That file usually has a red tab nobody is allowed to open.
24:04
SPEAKER_01: The pilot should surface that chaos without pretending it knows more than it does.
24:08
SPEAKER_01: The home screen could show unresolved owner names,
24:10
SPEAKER_01: well-identifiers that disagree across systems,
24:12
SPEAKER_01: missing documents,
24:13
SPEAKER_01: acreage conflicts, and records that lack a reliable effective date.
24:16
SPEAKER_00: And when I click a conflict?
24:18
SPEAKER_01: You see both values side by side,
24:19
SPEAKER_01: the documents each value came from, and the date of every source.
24:23
SPEAKER_01: The app can suggest a resolution, but a named person approves it.
24:26
SPEAKER_01: Nothing writes back to the operating system during the pilot.
24:29
SPEAKER_00: Good.
24:29
SPEAKER_00: Read only keeps a software demo from becoming a title opinion with a loading spinner.
24:33
SPEAKER_01: Put that on a coffee mug.
24:35
SPEAKER_01: The second screen is the operating view.
24:37
SPEAKER_01: Once records are linked, the team can ask practical questions.
24:40
SPEAKER_01: Which wells have conflicting interest data?
24:42
SPEAKER_01: Which leases are tied to production but missing from the current system?
24:46
SPEAKER_01: Which equipment records disagree with field status?
24:49
SPEAKER_01: Which files need human review before an economic model uses them?
24:52
SPEAKER_00: That last one matters.
24:53
SPEAKER_00: A bad number travels fast once it lands in a pretty dashboard.
24:56
SPEAKER_01: Faster than a rumor at an investor conference.
24:58
SPEAKER_01: So every answer needs a visible source trail.
25:01
SPEAKER_01: If the app cannot show where an answer came from, the answer stays out of the decision report.
25:05
SPEAKER_00: Now we are talking.
25:06
SPEAKER_00: What would we measure after 30 days?
25:07
SPEAKER_01: Time to answer a diligence question,
25:09
SPEAKER_01: number of unresolved conflicts,
25:11
SPEAKER_01: hours spent reconciling duplicate records,
25:13
SPEAKER_01: and how often reviewers accept or reject a suggested match.
25:16
SPEAKER_01: You also track the age of the exception queue.
25:19
SPEAKER_01: If it keeps growing, the app has created a new inbox rather than solving a problem.
25:22
SPEAKER_00: I have enough inboxes.
25:24
SPEAKER_00: One of them sends me a security alert every time I successfully sign in.
25:27
SPEAKER_01: The pilot also needs a clean handoff.
25:29
SPEAKER_01: When a reviewer approves a match,
25:30
SPEAKER_01: the system records who decided, what evidence they saw, and what changed.
25:34
SPEAKER_01: If somebody questions it six months later,
25:36
SPEAKER_01: the team can reconstruct the decision without calling three departments.
25:39
SPEAKER_00: Or the fellow who retired in 2019.
25:41
SPEAKER_01: He can finally enjoy retirement.
25:43
SPEAKER_01: After the pilot proves useful, you expand one dataset at a time.
25:46
SPEAKER_01: Add maintenance history for equipment,
25:48
SPEAKER_01: then operating alerts, then the pieces that support field planning.
25:51
SPEAKER_01: Each expansion gets its own accuracy check and human gate.
25:54
SPEAKER_00: What about the pumper getting that two-in-the-morning alert?
25:56
SPEAKER_01: The app should rank alerts by consequence and confidence.
25:59
SPEAKER_01: A mild trend can wait for the morning report.
26:01
SPEAKER_01: A serious change can wake somebody up, but it should explain why.
26:05
SPEAKER_01: Which sensor changed?
26:06
SPEAKER_01: How long it persisted?
26:07
SPEAKER_01: And what historical pattern made it unusual?
26:09
SPEAKER_00: If a machine wakes me up at two, it better bring evidence and coffee.
26:13
SPEAKER_01: Coffee integration is phase two.
26:15
SPEAKER_01: The first build is the acquisition data reconciler.
26:17
SPEAKER_01: Read-only inputs,
26:18
SPEAKER_01: source-backed matches, an exception queue, human approval, and an audit record.
26:22
SPEAKER_01: Small enough to test, valuable enough to matter.
26:25
SPEAKER_00: And honest enough to say, I do not know.
26:27
SPEAKER_00: That may be the most advanced feature in the whole thing.
26:29
SPEAKER_01: It is certainly the feature I want in every vendor presentation.
26:32
SPEAKER_00: Especially the ones where the demo data has 12 perfect wells,
26:35
SPEAKER_00: no missing documents, and every owner is named John Sample.
26:38
SPEAKER_01: John Sample has immaculate title and has never changed his mailing address.
26:41
SPEAKER_00: Put him in the brochure, keep him out of my pilot.
26:44
SPEAKER_00: So the takeaway is, don't ignore it.
26:45
SPEAKER_00: Don't think it's for the big guys.
26:47
SPEAKER_00: Because if you're not looking at how this tech can make your operations smarter,
26:50
SPEAKER_00: faster, and more efficient, somebody else is.
26:52
SPEAKER_01: Agreed.
26:53
SPEAKER_01: Start small,
26:54
SPEAKER_01: identify your biggest pain points,
26:55
SPEAKER_01: and then look for ways AI can chip away at those manual, repetitive tasks.
26:59
SPEAKER_01: This is the point where you'll find the real ROI.
27:01
SPEAKER_00: Yeah.
27:01
SPEAKER_00: And if any of this applies to your operation, reach out, siliconandshale at gmail.com.
27:06
SPEAKER_00: We're talking to operators every week about this stuff.
27:08
SPEAKER_01: We are.
27:09
SPEAKER_01: And we love hearing from you.
27:11
SPEAKER_00: Though if your software demo needs six browser tabs,
27:13
SPEAKER_00: three passwords,
27:13
SPEAKER_00: and the presenter's personal hotspot, I'm deducting two points before we start.
27:17
SPEAKER_01: That's unfair.
27:18
SPEAKER_01: Sometimes the hotspot is the most reliable part of the product.
27:21
SPEAKER_00: Last demo I watched,
27:22
SPEAKER_00: the one-click setup had 14 clicks and a calendar invite with a solutions engineer.
27:26
SPEAKER_01: 14 is technically a number, and the engineer was very nice.
27:29
SPEAKER_00: Nice fellow.
27:30
SPEAKER_00: Great headset.
27:31
SPEAKER_00: The software spent 20 minutes asking whether I was an administrator.
27:34
SPEAKER_01: Were you?
27:35
SPEAKER_00: Pac, I wasn't even the administrator of the conference room thermostat.
27:38
SPEAKER_01: Alright, that earns a 7 on the clipboard to autonomous scale.
27:41
SPEAKER_01: The extra point is for surviving the thermostat.
27:43
SPEAKER_00: So, you think those robots are really going to be shipping by 2026?
27:47
SPEAKER_00: I'm still skeptical about that one.
27:49
SPEAKER_00: I've got a buddy who's been trying to get a new part for his Ford for six months.
27:52
SPEAKER_00: I don't know how they're going to ship a robot.
27:54
SPEAKER_01: Shipping doesn't mean mass market adoption, Derek.
27:56
SPEAKER_01: It means the first ones are rolling off the line.
27:58
SPEAKER_01: We're talking prototypes.
28:00
SPEAKER_00: Still, I'll believe it when I see it.
28:02
SPEAKER_00: I've got a meeting with a landowner tomorrow who still uses a flip phone.
28:05
SPEAKER_00: I don't think he's going to be too impressed by a robot.
28:07
SPEAKER_01: One last thing.
28:08
SPEAKER_01: If your company solves a real problem for operators or land teams and wants
28:11
SPEAKER_01: to sponsor the show, write us at siliconandshale at gmail.com.
28:15
SPEAKER_01: We test products before we talk about them, so bring something sturdier than a perfect demo.
28:19
SPEAKER_00: That's actually true.
28:20
SPEAKER_00: We will.
28:21
SPEAKER_00: I'm gonna go try to explain what a PDF is to a flip phone user.
28:24
SPEAKER_00: Wish me luck.
28:25
SPEAKER_01: Good luck, Derek.