The following transcript has been edited for length and readability. Listen to the entire discussion here on The Broadband Bunch. The Broadband Bunch is sponsored by ETI Software.
Joe Coldebella:
Hello and welcome to another episode of The Broadband Bunch. I’m Joe Coldebella, and we are in the Harrison Edwards Media Center at Mountain Connect in Denver, Colorado. Joining me is Rhyan Neble, VP of Product Innovation at ETI Software Solutions. Rhyan, welcome to The Broadband Bunch.
Rhyan Neble:
Thanks for having me, Joe. How are you doing?
Joe Coldebella:
I’m doing pretty good. Just before we begin, ETI is the sponsor of the podcast. I can’t thank you guys enough for supporting us and sharing so much information with the world of broadband. Really appreciate it.
Rhyan Neble:
This is going to sound a little corny, but I am a longtime fan. I listen in on a regular basis, so I’m happy we can support the program.
Joe Coldebella:
Truly appreciate that. Rhyan, listen, would love it if you could share with our listeners your origin story. It’s always awesome to listen to the folks in terms of how they found their way into the world of broadband.
Rhyan Neble:
I stumbled headfirst into it. I was asked back in 2001 if I would support IT at a conference that was happening up in McCall, Idaho. Some folks were in the process of planning a new ski area or new ski resort, and I got along great with the real estate development team and the sales team. Somewhere along the line, somebody said, “Well, we’re going to create a closed ecosystem for telecommunications. It’s going to be triple play: voice, video, data. The entire resort is going to be next-generation technology. Do you want to participate?” I was like, “Yeah, absolutely. That sounds fun.” So I worked there for five years and then the Credit Suisse debacle happened and resorts in the States dried up.
So somebody reached out. They were in the process of building out a planned community, another resort this time on the island of Curaçao in the Netherlands Antilles. That was fun too. Somewhere along the line, somebody got this wild idea that we were going to pass 50,000 people with fiber from the landing station on the island as we were getting fiber into the resort. Let’s just become a national carrier. I’d been in ILEC before up in Idaho, but now we were going to go international.
We bought a Class-5 switch, and off to the races. And the entire platform was ETI triad for OSS, Cogsdale, and Microsoft Dynamics for the CRM and ERP solutions. It was an ecosystem that made sense to me. All the pieces fit together nicely. We could deliver all kinds of crazy services to people. I ended up bouncing between islands for about a decade, supporting telecommunications on St. Martin and telecommunications in Curaçao.
Joe Coldebella:
How did you get to ETI? I mean, it seems like you’re in the Caribbean. I would hope that there’s that at one point, right?
Rhyan Neble:
I have been a fan of ETI since the first time I met Pete Pifer. He came down to the island. I was a customer, and I’ve been running around advocating to anybody who would listen that if you needed an OSS solution, they were the guys to talk to. Then about five years ago, the opportunity opened up that I could join the team.
Joe Coldebella:
Awesome. They pulled you in. From what I’ve been told, you’re an absolute wizard. So now you are the VP of Product Innovation. Now, this amazing thing of AI jumped into the world in general last November. It’s funny because it’s one of those things where all throughout time in the early 1900s, I’m sure you’ve heard the quote from the American Patent Office, everything that will be invented has been invented.
Then you’ve got Paul Krugman, the New York Times columnist and economist. He said that the internet is probably going to be about equal to the fax machine, right? Then you’ve got Mackenzie and Co. When AT&T asked them, “Is the cell phone market going to be a big deal?” They said, “Oh, you’re going to sell probably about 900,000 phones.” So I think what it shows is that technology always wins out.
Rhyan Neble:
Well, a great example of that is what we saw with ChatGPT. I mean, when we think about success with the internet, Twitter was, I want to say, three months, four months to get to a million subscribers. That was the big target. I mean, that was the reason Twitter was the greatest internet company ever. Then TikTok blew that away. Then ChatGPT comes along and gets to a million subscribers in five days.
Joe Coldebella:
Which it’s incredible.
Rhyan Neble:
Order of magnitude difference.
Joe Coldebella:
So obviously, ChatGPT is on everyone’s mind. You’ve got things like Midjourney and all these different-
Rhyan Neble:
Generative AI.
Joe Coldebella:
They are almost public-facing, but there’s a lot more behind-the-scenes that are cooking. I was wondering if you could touch on that.
Rhyan Neble:
Well, there’s a misunderstanding about AI in general. I mean, first off, the concept of artificial intelligence is 200 years old. We’ve been on this steady journey to adopt these technologies for a long, long time. When we talk about pattern recognition and some of the machine learning tool sets, we’ve been using the machine learning tool set. One of the first things that we did at ETI when I started the development of the service management platform with the team was, we were using RPA, robotic process automation to script and automate a lot of the customer service sales and provisioning interactions between the BSS and the OSS systems. So introducing conditional logic and saying, “When we see this, this is how we’re going to automate these flows.” But that’s been around.
When we look at a self-organizing network, a SON, a self-organizing network that is capable of automatic fault healing and reconstructing itself based on demand in real-time, those are tools that are baked into technologies like ONAP or that ONAP was designed to be able to leverage. So software-defined networking and network function virtualization, these tool sets were built to harness AI and vice versa. AI was built around being able to leverage these self-organizing network features. That stuff is stuff that the big carriers, AT&T in particular, Verizon, T-Systems, Telefonica, have been deploying those kinds of tools for a while.
Joe Coldebella:
So was ChatGPT the tipping point?
Rhyan Neble:
I think it’s the public perception. Generative AI jumped into the public. We all saw Terminator movies and were afraid of what the AI is. And there’s been this stuff off in the ether, but it wasn’t mainstream. ChatGPT was the first time, even though it was a three-year-old dataset and wasn’t connected to the internet, and yada, yada, yada, it set everybody’s minds on fire. This is real. This is something that I can use today. Well, it might be a couple of weeks old now. There was a thing that came out from MIT where we now have the documentation on how powerful it really is.
Both quantitatively and qualitatively, using GPT tools, just text generation, and generative AI for creating text, the people that implemented it saw an 80% increase in the amount of work that they were able to do, but we also saw a 17% qualitative improvement in the work that they were doing. So not just quantitatively, we nearly saw a doubling of capability. These are untuned, immature, very, very new tools with no formal training programs, just people intuiting their way into using them. And we’re seeing an 80% uptake in total workforce productivity and a 17% and potentially much higher improvement in quality.
Joe Coldebella:
Those are really good stats. I wonder, is it because it gives the user an opportunity to use the tool to make their job easier in the sense that they’re doing the work, but they’re more an editor than actually a content creator? So then they build off as opposed to starting from.
Rhyan Neble:
I think that’s part of it, and that’s actually one of the most important things with this first generation of these generative tools. If you feed AI AI-generated content, it will actually go insane. There’s an acronym, MAD, which is this long complicated thing dysmorphia, but the concept of AI madness is largely exacerbated by feeding an AI its own product. So if it gets into a loop, feeding on itself, it just goes insane.
Joe Coldebella:
I could just see it. There’s a sci-fi movie, right? The AIs have taken over the world, and the way the humans take back the world is they turn the AI in itself.
Rhyan Neble:
The AI turns on itself regardless. The only thing that saves the AI right now is when a human being comes in and reintroduces something new into the model. So the neural networks, if you train them against themselves, they self-destruct. To the best of my knowledge, I don’t think anybody’s seen an AI that was successfully able to train another AI. What we’re seeing is that you have to have people train the AIs, and that’s been actually one of the dirty little secrets of ChatGPT. A lot of the prompts are actually going through human hands for the pre-processing.
So a lot of the stuff that you’re typing is actually getting written or getting read by somebody else. That long pause right after you type something, you’re actually having a conversation with a person at the other end of that a lot of the time.
Joe Coldebella:
Really?
Rhyan Neble:
Yeah. Now a lot of the lookup and a lot of the validation’s happening. Microsoft’s Copilot is actually pretty interesting in that they’ve automated pre-processing and post-processing logic. So the standard GPT tool sets are designed to be engaging, and they will actively lie to you because they’re trying to create. We’ve trained generative AI to be creative. Because it’s creative, it doesn’t know the difference necessarily about when it’s making something up just for the sake of making it up or when it’s making something up that’s deliberately deceptive. It can’t make that distinguishment.
Joe Coldebella:
So if I could do a little bit of shifting in terms of bringing it to the broadband world, is this one of those things where every organization should at least dedicate someone to, if not actually implementing it, at least educating themselves on this, like cybersecurity? Is it one of those things where you ignore it at your own peril?
Rhyan Neble:
Yeah, absolutely. If somebody isn’t already looking at it in your organization, you’re already behind the curve. The adoption rates that we saw with ChatGPT tell a tremendous story of how fast this is spreading through the collective human zeitgeist. We’re going to be augmented by artificial intelligence on one level or another. We are doing it today.
Joe Coldebella:
Okay. So is that also because of the fact that it’s going to be integral because there’s going to be a shortage of employees, or is it one of those things that we need because we’re coming on fiber? There’s so much data being produced that it’s not nice to have; it’s a must-have.
Rhyan Neble:
Well, there are a couple of things going on there. The first thing that you have to recognize is that with the general accounting office saying that we’re short 36,000 or 38,000 people now just to be able to close the digital divide to build everything out. Those same people aren’t going to be needed in the workforce five years after the infrastructure is built. So what are they going to do? We need a bunch of manual labor and hands to go build that infrastructure, but we need to be able to support it. The support tools that are required after the infrastructure is in place are radically different. They’re software-based. They’re AI-based. We can solve a lot of problems in this early generation of AI. Even without the tuning, we’re seeing that 80% improvement in manpower capability.
That’s a huge uplift, and that hasn’t been considered in the general accounting office’s numbers. So we’re going to get more out of people initially as we adopt AI, but we’re also seeing that the skillset radically shifts as we move toward hyper scaler and large cloud-based hosting solutions as data centers go away. With AT&T, we saw the wire centers drying up and going away, and then there was a shift back with the hyper scalers needing services out at the edge. But that’s all hyper-scaler infrastructure at that point. So that’s AT&T, AWS, and Google owning these large data center footprints that are offloading from all the traditional telcos.
Joe Coldebella:
So I’m a little confused. Is AI going to replace folks, or are they going to work with folks? Is it going to be a combination? I’m sure some jobs are going to go away, but I would assume also some jobs are going to be created from this as well. Correct?
Rhyan Neble:
I think you’re going to see three waves. The first wave right now is this curiosity, but you’re going to see a huge uptick people in using AI to be more effective. And we’re going to see productivity explode for the next three to five years. What you’re going to start to see after that is you’re going to see that we’re going to have a lot of specialization leveraging the AI toolsets. It takes some skill to actually properly write a prompt and get the most out of it.
Joe Coldebella:
A prompt engineer, I had never heard that term until a few months ago, and apparently-
Rhyan Neble:
Whole new concept.
Joe Coldebella:
… if you have kids and they’re entering school, point them in the direction of a prompt engineer because it seems as though it’s a field that’s growing and it’s a field that’s very lucrative.
Rhyan Neble:
I saw a report a couple of months ago, and it was like, we’re going to see thousands of billionaires, an enormous number of billionaires. We’re on track to there being the first trillionaire, and we can talk about all the social implications of that.
Joe Coldebella:
I can rest assured it’s not going to be me.
Rhyan Neble:
Well, me either, but the other side of it though is that we’re seeing an entire generation that’s going to grow up with AI the way the Gen Z generation grew up with a mobile phone, the way you and I grew up with computers. And we saw cell phones, right?
Joe Coldebella:
Now, you’re scaring me with this.
Rhyan Neble:
So you’re going to see a generation that is augmented. I mean, we have a sense of what it’s like to have the internet at our fingertips. But there are some people who get more out of the internet because they know how to properly search it, or they can drill in and research it. Imagine if all of the data was immediately at your fingertips, and it was relevant. Imagine how much more powerful that is. That’s one of a myriad of use cases that you’ve got now that the AI is embedded directly into our browsers. You’re going to see a generation of people now grow up where AI was commonplace.
Joe Coldebella:
Please correct me if I’m wrong, because I’m definitely neophyte, but Moore’s law, right? Everything in terms of technology expands, I forget, five to seven years.
Rhyan Neble:
The doubling was based on transistor count and Moore’s law is broken at this point.
Joe Coldebella:
Because I have heard that. So is it broken because it’s going to be even quicker, or is it going to slow down a little bit eventually?
Rhyan Neble:
Moore’s law is irrelevant because it deals with something that we’ve now reached the limits of subatomic physics. We can’t make transistors denser. As we get into the six and four-nanometer processes, that’s the limit. But forget all of that because quantum computing is right around the corner, and quantum computing is going to leave Moore’s law in the dust. When we get AI plugged into quantum computing, that’s when, I mean, Michio Kaku, and I just mispronounced that name terribly.
Joe Coldebella:
I don’t think he’ll mind.
Rhyan Neble:
I hope not. I have nothing but respect for that man. The genius of AI loaded onto a quantum computer where you have an almost infinite number of bits to play with is going to be incredible. You’re going to see a processing capability hockey stick. I mean, you’re going to have a conversation and not know that you’re not talking to a person. Right now, what’s that term, the chasm? It’s like with movies, the suspension of disbelief. You have a conversation with an IVR bot or a chatbot. Now you know I’m not talking to a person yet. Siri just doesn’t quite feel like you’re having a conversation with a person. That’s going to evaporate over the next three to five years.
Joe Coldebella:
Right. I think it’s interesting though, because it’s slowly getting there. I’ve had recent interactions with Siri where it has taken on a little bit of a humanistic approach in terms of the interaction. Obviously, it’s still a little clunky.
Rhyan Neble:
Well, I’m sure somewhere at the back end of things, there’s Bard and on the Microsoft side, we’re actively engaging Copilot tools directly into IVR prompting so that we can have conversations with GPT-enabled AI.
Joe Coldebella:
Okay. Well, that’s a good segue. We’re talking really high level here, and I would love it if you could unpack that. So if AI is going to permeate the world and the world of broadband, how is it that some companies, and let’s use ETI as an example, how are they using AI to benefit the customers, to benefit the subscribers, to benefit everyone in general?
Rhyan Neble:
Six general places where we’ve been leveraging AI and telecommunications, but the generative AI stuff is a complete shift. We started out on the telecommunications side, and we were really focusing on network automation. We transitioned from network automation to being able to do customer service interaction, introduced robotic process automation into those tool sets, and started to look at solutions for new products that were AI-enabled or AI-driven. Siri’s a great example of how you use an AI to productize an entire class of other products, but there’s an internet dependency on that. So all those devices became IOT devices.
All those devices could then feedback telemetry. Being able to do real-time telemetry and network monitoring became a whole other section where we started to leverage AI because we had billions or trillions of data points. It’s on par with everyday companies like AT&T, Verizon, and Frontier. They’re producing trillions of pieces of data, and it’s more than any human being can deal with. It requires an AI to be able to look for the patterns in these huge data sets. So big data processing is one of the things that’s been feeding AI.
Joe Coldebella:
So let’s say I’m an ISP, and I’ve got 35,000 subscribers. How can I use AI, or how can I employ a software company that uses AI to make the processes easier? The phrase that I hear often is the swivel chair. It seems like there’s all these different things that people are doing.
Rhyan Neble:
I’m not a big advocate of swivel chair. I mean, one of the things that we built the service management platform for was a single pane of glass. I don’t need to own your data, but I want to tap into all of it and pull all of that back. Because once I have that data within my single pane of glass, I can also feed that data into the AI. Things like predictive maintenance, recognizing faults, and proactively dispatching field service technicians before the customer even recognizes that there’s a problem or complains about anything. If your vision is to bring joy to the customer, it’s certainly just not to have conflict with a customer and proactively solve problems.
Joe Coldebella:
Then AI allows for all the back-of-the-house stuff to be cleaned up prior? That way, it gives the CRS or the folks the chance to concentrate on the problems that always come up. So there’s more effort put towards problems and everything else gets pushed into the background.
Rhyan Neble:
Partly, but there are also some things, I mean proactive logic, but also being able to recover from when things are falling apart like somebody’s late for work or they’re not going to show up because they’re sick. We saw a lot of this with COVID, and you have to reshuffle all the field service tickets for the day. All the installs that are going to happen this week have now been changed because there was a traffic accident and one of your vehicles is now offline. And you’re not going to be able to make those installs. So that used to require going through the tape charts and rebuilding all the field service activities, all the field service scheduling by hand.
Now we have AIs that look at it and take into account weighted averages for how much load you want on individual people, minimizing the number of left-hand turns so that you get better insurance premiums, minimal amount of traffic so you have less wear and tear on the vehicles and better fuel efficiency, taking streets that have fewer stop signs or traffic lights so that you’re not breaking as often and yada, yada, yada. It just goes on and on and on where you can have these AIs self-optimize the entire workflow. That’s not because the data was necessarily even there in the first place. It’s forming that information based on looking at the patterns of how you use the system and how it’s evolved.
One of the coolest things that we’ve been playing with Microsoft and it just GA’d. It just became generally available last week and was a new tool around power automation that allows us to monitor the way people are using the system and make suggestions about “We see these patterns in what you do. Would it be okay if we automated that for you? So that instead of it taking 15 clicks to do that or instead of it being two minutes to get your desktop all set up first thing in the morning, how about I just script that and it sets itself up from now on the way it’s supposed to save you two, three minutes here, five or six minutes there throughout your day?”
Joe Coldebella:
Is that what it’s all about? It’s these subtle changes that people don’t really see happening, but it’s happening.
Rhyan Neble:
Good AI will be subtle. Good AI will be so… I don’t want to say insidious, but it’s going to just be there. And you’re going to find yourself taking advantage of it more and more. You’re going to realize at some point it’s not going to be there one afternoon, and you’re going to be, “Wow, I really am not as productive as –”
Joe Coldebella:
Right, because it’s almost like a course correction. Course correct, and then ultimately you get to the right answer. So you’ve got this AI. Then are you also engaged in using augmented reality?
Rhyan Neble:
Yeah. So Microsoft did all of that work with HoloLens, the next generation of HoloLens. They had that big military contract around HoloLens. While we support the HoloLens, augmented reality on an Android or tablet is a fantastic tool for field service technicians.
Joe Coldebella:
Could you give an example?
Rhyan Neble:
Well, yeah, I mean, and this plays really well into the whole problem that we have with the shortage of people. We need to very quickly be able to get people out into the field that can be productive. And there’s going to be minimal time to train them, and they’re not going to have the expertise. There’s also a huge number of people that are going to be leaving the telecommunications industry. They’re going to retire out here in the not-too-distant future. You have all of that expertise that’s basically senior management now. They’re all sitting in offices somewhere.
With augmented reality, if I’m in the field and I’ve got my tablet that I’m able to do my onboarding with, I’m checking my field service work orders off. I’m going through my tasks, and I run into something, some troubleshooting steps or something. I’m not familiar with how I’m going to solve this problem. I press a button to be able to do the remote assist. It pulls up all of my team’s contacts. I can pick anybody in the company. And I can prioritize people in that work order who have the skill set to be able to help me with the problem or the task that I’m currently on.
I press that button. And I reach out and get a lifeline from a senior guy who maybe he’s already in the field working on a different project or at a different site, or maybe he’s sitting in the back at the knock or at the warehouse. Now, he’s looking through the camera, and he’s able to draw on his screen with his mouse or just from his phone. He’s able to say, “You see this thing over here, and you’re able to mark it up,” and say, “You take that, you plug it in here, grab your yellow clip off the potentiometer, and you plug it into that.”
Joe Coldebella:
There’s the potential of that. It’s almost like a play-by-play where you’ve got the goggles on. Hey, listen, I need you to do A, B, and C.
Rhyan Neble:
So if you want to go hands-free, that’s what the HoloLens is for. I mean HoloLens is like $3,000, but at the end of the day, if you save two truck rolls, I think the industry average is 1,500 bucks for a truck roll now. So if over the life of that HoloLens, you avoid having to send somebody out on another ticket, those will pay for themselves.
If you don’t want to use a HoloLens, any tablet, or any Android-enabled device, and I prefer Android, I understand that they work on iPhones. I have not personally tested it, but you can run around with a tablet, or you can run around with a smartphone. We’ve extended that toolset so that we can provision from it so that we can manage the ONTs. And we’ve “telcoized”, to borrow from Jeff Fraleigh, the augmented reality tool set. But it’s a fundamental tool.
Then the next generation of it, when the AI creeps into it, there’s a toolset from Microsoft called Guides, which is a training tool. Boeing uses it for teaching people how to do wiring harnesses as you’re wiring up an engine or wiring up an airplane on the assembly line. The next generation of it is going to be the AI optimizing the processes and taking you through it without… It’s going to make suggestions.
Joe Coldebella:
So is the idea that you’re allowing someone to learn from the very best? So whatever you’re doing, you’re getting the optimum performance, because you’re drawing from whichever is the peak.
Rhyan Neble:
Right. You’re pulling in the AI resources; you’re pulling in the seasoned expertise resource to provide in the field the best possible. Basically, these neophytes are able to be brought in, and they can become a set of smart hands. They’re guided to perform the tasks that they need to do until they’re comfortable doing the task on their own. Then they don’t have to make those phone calls anymore. It’s an early crutch. You can always phone a friend. You can always get a second opinion if you’re not happy with what you’re looking at.
Joe Coldebella:
That almost follows the doctor ethos or surgeon ethos of seeing one, teach one, do one. Then obviously the person who learns from the senior also can pass that information on to someone else. It’s a great way to also just exchange information throughout the company.
Rhyan Neble:
And throughout the organization and throughout the industry. I mean, there will be a time in the not-too-distant future where you’re going to see things from organizations like the TM Forum where they’re going to have a prescription similar to what they do with eTOM. We’re going to start to see these examples of processes that are going to come in. I wouldn’t be surprised if people like Calix and Nokia started doing a similar type of solution for installing their hardware maintaining their hardware and supporting their hardware.
Joe Coldebella:
This has been a phenomenal visit. Rhyan, before we wrap up, I got one more question, because I know that you’ve got to hit the road. I don’t want you to miss your flight. In terms of the breadth of these AI systems, I was at the press conference yesterday. And you said that you had 160 tools. So if a company were to engage with AI, is it one of those things where they would’ve to use 160?
Rhyan Neble:
No, no, no. It is fully modular. Like I said earlier, if we’re doing our job, there will be absolutely things that you’re going to overtly interact with.
Joe Coldebella:
That everyone will use. Okay.
Rhyan Neble:
But if I’m doing my job, there’s a lot of stuff that’s baked in that’s very subtle. If I do it correctly, it’s subtle, and it’s just a really cool tool. It’s a widget built into the UI, and you’re just able to take advantage of it. Then there will be other things that will be much bigger ticket items. The thing that you have to appreciate is all of this AI stuff, there’s a lot of overhead. There are a lot of processing requirements. There’s a lot of power necessary to drive some of these things. So there are going to be certain gates for licensing and for the commoditized resource on the hyperscaler side from Azure, AWS, and Google that feed into the system.
But with the things that are baked in out of the box, we’re going to have a lot of things available immediately, subtly. Then if you want to flip the script and have fully autonomous field service dispatching, for instance, that’s a separate licensing piece. You can flip the switch on that, and we can turn it all on.
Joe Coldebella:
So then ultimately what AI becomes is an a la carte that in terms of a software company or an ISP, they work together to find out what fits best for them. So then they get the optimal out of the software itself.
Rhyan Neble:
Yeah, what’s going to be the biggest bang for your buck on day one? Then if you want to evolve into the total tool set long-term, you can. If you want to take it all at once, it’s going to take more time to deploy your solution, but you’re going to start with everything out of the gate and give yourself the best possible advantage. So whatever strategically works for you.
Joe Coldebella:
Then I lied. I have one other question, and then we’ll wrap it up here. Then there’s the two things, Brownfield versus Greenfield. So is Brownfield more complicated because you’re dealing with a legacy system, and then it’s shifting over? Whereas a Greenfield allows you to start from scratch, but it gives you a little bit more of an open field?
Rhyan Neble:
Yeah, I mean, specifically about AI, the AI tools aren’t going to do a whole lot for either organization until they’ve had a chance to learn how your processes work and how you’re using them. They need to monitor your data for a little while. So some things can get trained up in as little as three or four days of observing you. Some things take six months or a year to fully actualize.
Joe Coldebella:
That’s what software in general is that unfortunately, it’s not always a flip of a switch. There are definitely some growing pains.
Rhyan Neble:
Right. That’s true.
Joe Coldebella:
Rhyan, I’m sure we could talk for hours and hours about this. It really is a fascinating time just in terms of how the world is evolving. I really appreciate you spending the time unpacking this super complicated, at least to me, subject. I would love it later down the line as things continue to develop, bring you back on, and just drill you for your expertise. Thank you so very, very much.
Rhyan Neble:
Anytime, Joe.
Joe Coldebella:
All right. That’s going to wrap up this episode of The Broadband Bunch. Until next time, see you guys later.
