Episode 2 - Welcome to AI FYI - What is AI? 

Transcription

0:00:21 - Joe C

Welcome to AI FYI and hello, Andy and Kiran. We're going to talk in a minute a little bit more about ourselves and what we're going to be talking about today, but first it's Halloween weekend here. What's everyone's Halloween plans? 

0:00:35 - Kiran V

My wife teaches at a yoga studio and so it was all the people from the yoga studio at the party and I was the only one there that didn't work in fitness and everyone's like, oh, that's so cool, you're an engineer and I'm like I feel like it's a pretty standard thing to be an engineer in the Bay Area, so it doesn't feel special. So that was interesting. 

0:00:59 - Joe C

That actually reminds me I one time, side note, I had an idea for a haunted house yoga class, but anyways, topic for another day. Andy, how about you? 

0:01:08 - Andy B

I know a large group of people who would go to the haunted house yoga class. I'm not like a big Halloween person and it's always surprising to me like who's really Halloweeny and who's not. I went to a Hozier concert last night. I don't usually go to a lot of concerts but a friend of mine got me a ticket and it was really fun and it was an interesting mix of like girls dressed as mermaids or sirens with full rubber green ears and then just like people in like jogging clothes who just wanted to come see a show and just comfortably stand all night. But the show was really good and it got kind of like weirdly heated. Multiple times a lot of people collapsed in the crowd which I thought was interesting. 

And Hozier kept like turning lights on and pausing and apologizing, but like he could see the crowd right and so he was pointing out where people were not doing very well. And then at the end between the show and the mandatory encore that now every show has, there was a whole talk to your representative about Palestine situation and he's Irish right, so he's really familiar with occupation and things like that and US involvement. So that was just a mix of stuff that was kind of Halloween themed. 

0:02:28 - Joe C

Well, for me, I'm going to a party tonight and I'm going as corn, and that's all I'm going to say on the topic. 

0:02:36 - Andy B

This text me a photo. I need to see this. 

0:02:38 - Kiran V

Yeah, I need to see this picture. 

0:02:41 - Andy B

Was it you, a couple of years ago, who you came to the office, dressed as AI or robot?

0:02:51 - Joe C

Are you thinking when I was a Ferrero Rocher?

0:02:53 - Andy B

Well, maybe that's why I was thinking it was metallic. 

0:02:55 - Joe C

Metallic kind of looks like a robot. 

0:02:58 - Andy B

Yeah, that's great, All right. What are we talking about today? 

0:03:03 - Kiran V

Okay, well, today this is our episode one of AI FYI, and so I think really the way that we want to kick off this series is to just do an intro on AI and kind of talk about you know - what is AI? Who are we? Why are we talking about this topic? And start to you know, maybe touch on some of the topics that we want to get into. Just to start off, we can talk to you a little about who we are. So I'll start. My name is Kiran. 

I am an engineer. I've been building AI applications for close to a decade now, which is just kind of crazy to think about that. I've been doing this for so long. When I first started as an engineer, you would hear people that were like been doing this for like 10 years, 12 years, and I'm like, oh my God, those guys are so experienced and now I've been through some things. So it's it's cool to kind of look back and reflect on my career journey in AI. What else about myself? I love cars, I love driving and I live in the suburbs, so I, you know, get out there, go hiking, have a garden, pretty typical things. 

0:04:25 - Joe C

Great, I'll go next. So my name is Joe and I'm a product designer. I've been in tech a little less than that, I suppose, maybe like eight years. I worked in nonprofits for a while. I worked making tech solutions for schools and then have been in the business of AI and developer experiences for a few years now. I think one interesting thing about being a designer is that we're tasked with thinking about the human using AI, using tech and what they're going through and their experience, and that, paired with my strong interest in tech in general and sort of where technology is going to take us, I'm always kind of thinking about the human aspect and how it affects us on a small and large scale. I live in San Francisco. I like Japanese food, I like illustration and talking about AI and tech. 

0:05:17 - Andy B

Yeah, my name is Andy and I'm a product manager. I think I have 11 years of experience in tech and AI now, which, again, very weird to think about. It's like 11 or 12 years. I've been a product manager for almost six of those. 

At this point, and if you're not familiar with tech, the product manager job is kind of a weird one to explain. If you think about a team full of incredible designers and engineers and a bunch of customers, there's often like a lot more ideas of what to build than you'll ever have time to finish, and so my responsibilities are figuring out like, what makes sense to build now and what is it going to take to actually realize our vision for it. So many things that you start just never pan out because you didn't have the right resources to begin with. Or you build something incredible but there's no customer for it and your company tanks. So my job is to basically do that connection between what we want to build and what we need to build and making sure that's all going to pan out. Yeah, so I think I have the coolest job. 

0:06:31 - Kiran V

Yes, product management is really cool. So, for those of you listening, Andy, Joe and I actually met at a company called Figure Eight and the company actually helped other companies build AI. So it was really cool and unique because we got to see a whole bunch of different types of applications of AI and many, many different ways that it can be used. Generally, if you're working in a technology company, there's a particular mission and specific types of thing that the company is trying to build. Figure Eight was actually helping other companies create their AI, so we got to see, you know, a lot of different cases. 

0:07:11 - Andy B

Yeah, and because the three of us were doing what's called user research, so we were actually the ones like building prototypes and connecting them with our customers. I don't think it would be an exaggeration to say that between the three of us we've talked to 300 machine learning teams in all kinds of industries. So we'll explain what the term machine learning means in the context of AI in a second, but long story short, three of us have had pretty unique and wide exposure to a wide world of what's happening out there in AI. 

0:07:44 - Kiran V

Not to mention we created a bunch of patents. 

0:07:46 - Joe C

Aall right, so let's get into the topic today, and we're going to start with the big question what is AI? I think this is probably the broadest spot we can start in. AI is applied machine learning and we're going to talk about machine learning. But there are a lot of ideas and even emotions attached to this and it's somewhat theoretical. It takes many forms, but what it is is the idea of using computing power or machines to replicate or simulate human behavior. And that can happen on many scales. There's sort of two terms we can talk about at this point - Narrow AI, meaning we're developing systems that are very focused on one task and maybe designed to do something very small. You might say that a simple math problem or even the functions of a calculator is AI, all the way up to general artificial intelligence, which is the more scarier idea, or intense idea that we could have AI that meets or exceeds human intelligence and behavior. So AI sort of is everything in between and this idea of using machines to do what humans do. 

0:09:06 - Kiran V

Yeah, and I think something that comes up frequently when talking about AI is this concept of machine learning, and machine learning is one implementation or one way that you can build AI. So if you think about, you know layers of this, artificial intelligence is the broadest umbrella of enabling machines to do human-like tasks. Machine learning is a subset of AI, which is a particular way that AI is built. There's also many other ways to build quote-unquote AI. You know you have regression models, you have random forests, you have all sorts of stochastic or deterministic methods of creating AI. 

0:09:56 - Andy B

Yeah, I never cease to learn from these guys. But one thing to be aware of in the world of AI is there's something that's known as domains in AI, or regions, and essentially in the practice of machine learning, which is what a lot of people who work in AI do in order to make machine learning models (we'll explain those in a second) appear as AI to people. There's different areas. We're gonna have to do different episodes on all these different areas because it's way too broad to cover, but we'll introduce you guys to some of the major ones today. So one big area of machine learning is a domain called computer vision, and this is exactly what it sounds like. Can you make a computer see? So this is the tools and the technology and techniques and math behind making a computer behave like a pair of eyes. 

So simple examples of this would be like the US post office uses something called optical character recognition, OCR, to read the address. They don't have a human being available to read every address of every piece of junk mail we all get. So they can do things where they can scan and read and computerize the reading. So that's like a really simple way of doing computer vision. That was one of the first large scale applications. But this goes all the way to doing generative computer vision, which we'll talk about more in a second, to like the object detection, so you can type into your phone app on photos what you're looking for and it finds you all the pictures of your dog. That is an example of computer vision and it's a very common term and people getting PhDs might just specialize in the subset of computer vision which again interacts with photos and videos. 

0:11:59 - Kiran V

Yeah, and then another very common area in AI that you'll hear is NLP, or natural language processing. Sometimes you'll hear NLU, which is natural language understanding, and the general idea here is that we, as humans, use language to communicate, both oral, written, and how can we enable a machine to listen or understand what humans say or put out through their communication, through their language, and again start to make human type decisions? So if we have a piece of text, can a machine look at that text and understand, actually understand what it's talking about? And then this starts to get into, you know, conversational models, chatbots. 

0:12:55 - Andy B

Interestingly, audio, like recordings of people speaking, is often bundled underneath NLP because, ultimately, even though it's an audio file, not a piece of text, you can do a process to turn it into written texts like a transcription and then treat it like anything else that's been written down. 

0:13:15 - Joe C

Absolutely, and this would also include, Kiran, you said conversational AI, so this would be interacting with Siri,  Google Voice, any voice assistance. Now I wanna talk about generative models, which isn't exactly a domain, but more an approach to how certain models are trained, and this is very topical. We've heard a lot about generative models this past year, and chat GPT is one. So the idea of a generative model is that you take a very large amount of data and you use it to train the model, and then what's happening is essentially pattern recognition. So if it has a ton of examples to go off of, then it can replicate what is the most likely outcome, given all that data. So chat GPT is one very common, very powerful example. That's an NLP use case. There's also examples in computer vision. Maybe you've heard of, like Dolly 3 or Stable Diffusion. These produce images based off of large amounts of image data. 

0:14:20 - Kiran V

Yeah, and the concept with a generative model is that the model, is generating something that looks like a human made it so Chat GPT is producing text or paragraphs that look like a human wrote it. Dolly-3 is producing images that looks like a human might have created those. 

0:14:40 - Andy B

And we're in a really exciting time right now where these generative models have like really made the news and a lot more people are interacting with AI knowingly and with intention than ever have before. So I think that's something that's really exciting about these generative models is they're like really easy to use and to find something helpful they can do for you. And some other areas that we just wanna touch on briefly is we're talking about computer vision, NLP and generative models because that's very understandable to a lot of people and it's kind of the hot new thing, especially with generative AI. But there's other areas of machine learning that actually have just as much impact on your daily life but you maybe don't even think about them as AI. For example, there's a domain of machine learning called search relevance. It's the reason why you can go into a Google search bar type in the first three lyrics of a song you kind of forgot, and it comes back with the name of the song, the author of the song, the year it was released, the full lyrics, right. It's about understanding the query that you made, the search and giving you the most relevant results. I'm sure every website you've ever used has a search bar and those are all powered by a type of machine learning, and also behind the scenes is a type of machine learning, like domain, called regressions, and regressions is basically these predictive models that do the same thing that you were forced to do in high school algebra - so predict where this line is gonna go next. But they do this at really large scales and you basically like plot a bunch of points and then try to predict where the next one's going to go. This is really common in banking and in finance industries. So, for example, if you've ever applied for a loan or a credit card, there's almost certainly a regression model that's trying to predict will this person be likely or not likely to pay back their loan. And they might do that by just predicting what is the most likely income you will have next year based on your previous income. But what's really important to understand about all these domains is that, just like a person has many senses you can speak, you can hear, you can smell, you can see, you can feel temperature -  A lot of AI is powered by something we're calling multimodal solutions. 

So think about if you've ever written a description on a website and it spits out like a generated image for you. That's multimodal, right, because you put in text and you got out an image and its multimodal. Basically, you could have a single machine learning model that is multimodal. So there's a model that can take in text and spit out images, but just as often, or probably even more often, you would chain different types of machine learning applications together to solve one AI problem. So if you Imagine I do this all the time, which is I use chat GPT or Google - something where I actually use voice to text because I'm too lazy to type which is a model that Google has to turn my voice into text and then that text gets submitted to to open AI. 

0:18:10 - Joe C

I think there's now a release from open AI where Chat GPT which up until recently has worked with text, can now produce images. So it's possible that you know your first steps could ultimately result in an image getting revealed. 

0:18:28 - Kiran V

Yeah, yeah. And I think another example of you know multimodal AI at work is like self driving cars, right, self driving cars have a whole bunch of different quote unquote senses that it uses to actually perform that act of driving by itself, right. You need to look at the road around you, so that's going to be computer vision. You need to listen for things like sirens or horns, so that's going to, you know, use audio models to process and understand, like, hey, someone is honking at me from behind, right? So you need a lot of different types of models as input to be able to effectively execute on the task at hand, which, in this case, is driving a car. 

0:19:14 - Joe C

That's right. All right, we've talked about a couple of domains and approaches and ML, and we probably used a lot of language that maybe you're unfamiliar with, so we want to spend a little time talking about some common terms and some additional bigger ideas in AI. So let's jump into a few of those. 

0:19:32 - Kiran V

Yeah, so I think one to start with is a 'model' right. We keep talking about AI models, machine learning models. It's a model. A model in its basic sense is a algorithm right. It is a piece of technology that is used to perform AI tasks right, and so when we think about a model, that is the actual algorithm or the actual, you know smart stuff right, when we're talking about AI and you might use a single model, so you can have a model that is performing computer vision tasks, or you can have a model that's performing natural language processing tasks, or, in the case of the self-driving car, you might have many models working in parallel to execute on that task. 

0:20:26 - Andy B

And it's important to add to this. Each model basically takes some data in and spits some data out. There's variations on this, like how much data in, how much data out, but that's essentially what they're doing and if you think of it as this like little box that can make these changes, the size of that box varies. Inside that box is basically linear algebra that people can't understand and wouldn't write. That the model wrote for itself. And when you hear terms like neural network and large language model, all that people are discussing is what's in that box and how big is the box. So some of these generative models you need a lot of linear algebra to power creating an image. So there's a really big box with a lot of math inside and some models when we talk about they're small or they're light. It's just - imagine a really small box that has everything it needs to take data in and out. 

0:21:27 - Kiran V

Yeah, and for me as an engineer who is not a data scientist, I will generally ignore the things inside that box. I'll just understand, what can that box do for me, right? If I give it an image, it will tell me where all the people are in the image. Great, I don't need to know how it works, and I will now take that box and go put it into an application so that now I can search for images of people on my phone. And so you know a lot of cases you'll hear people refer to that as a black box, because they don't care what's inside it. The data scientists did all the smart things and now they can take that and use it in a specific application, whatever that might be. 

0:22:09 - Andy B

And to jump off what Kiran was saying like, as a product manager (and I work with the entire team that comes to bring a solution of AI to you, the user) I do sometimes care what's in the box, but I care a lot about the shape of the box in some sense. So, for example, Joe could tell me that the model is really solving the user's problem and Kiran is like, yup, I'm able to put data in and take data out and it's behaving like we expect. And then I realized that, like, the bill for the size of the box is 10 times what we'll ever be able to charge somebody to use the box. So then I'm like, hmm, we have to try again, but with a smaller box. And so when we talk about models, we're talking about these discrete little processing boxes. Let's call them that. We're trying to piece them together to solve problems for AI. 

0:23:05 - Kiran V

Yeah, so I think the next piece that follows from here is how do we actually create that box? How is AI created? We'll have a whole episode on more of the details of that, but in the most basic sense, what we do is we go through a process called training. So you have a model or you wanna create a model. Well, we need to teach the machine how to do whatever task we want, and that process is called training. We have a whole episode planned for training data and how it's used and what you can do with it. But at the most basic sense, we train models. So, just like you might train a person to do a task, we're training a model, or we're training a machine to do a particular task, and this can be done in many different ways. 

What's most often used is you take a whole set of data that represents the type of task you're trying to do. So, for example, if I'm trying to recognize humans in an image, I'll take a bunch of images with humans in it. I will label that and again we'll get into exactly what does that mean. But I'm indicating to the machine this is where all the people are in those images. Now the machine will look at those images. It'll say, okay, cool, you're telling me that this is a person. Great, I'm gonna just understand that however I can. And again, for me as an engineer, I'll consider that a black box. The data scientists are the ones that are really gonna get into the weeds of how is that model learning from those images? But once you go through this process of training now, you're gonna have a model that is able to perform that particular task of, in this case, recognizing where humans are in images. 

0:25:00 - Joe C

All right, and we can't really talk about data and training data without talking about bias. So bias really is our entry point into the underlying ethics of AI, and it is how well a model performs and that's according to does it meet the expectations of those who have created it and those who are using it. The expectations of humanity? Is it safe, is it trustworthy? And a model's bias depends on the data that goes into it, and the underlying idea is really bad data in is going to be bad data out, so garbage training data will result in a garbage model. And this is really why it's so important to use data that has been looked at very closely to remove any bias, to make sure it's not racist or sexist, for example, and to make sure that once it goes into the model and the model's released out into the world, it is going to be something that is safe and fair to use. 

0:26:05 - Andy B

And this can get kind of complicated because Joe mentioned it's important to remove bias, but sometimes to remove bias you have to inject bias. So something that's been kicking around in the world a lot is how come, when you type show me a picture of a CEO into Stability AI, it shows you most often a white man. Well, if you think about it, most CEOs are white men, and so the model is mimicking what it sees in the world. And then it comes this question of do you want the model to mimic reality or do you want it to mimic our ideal, which is probably that CEOs look a lot more like everyone else too. And so what's? The responsibility of the team that is developing these AI solutions is to decide when do I remove bias and when do I inject bias so that I'm correcting for the right thing for this solution that I'm building.

So on that note, you've probably heard of two really common AI applications I want to define and provide some context on. Odds are, almost everyone listening to this has heard the word chatbot. Chatbot refers to the sort of AI that you interact with, either typing or sometimes by voice, that talks back and forth with you, and chatbots are not new technology. Odds are you have been on a chatbot via a customer service tool as early as the mid 2000s, but what's interesting is there's been a lot of innovation in this space. I don't think anybody's gonna be surprised when I say that the chatbot you interacted with and customer service and banking in 2012 is absolutely annoying and bad compared to the chatbot you could interact with on Chat GPT today. That will be much more delightful and helpful. So this technology has been around for kind of a while, but the underlying method of how we make that AI work, that use case of a chatbot, that's changed and we're getting more sophisticated in trying different techniques to produce better chatbots. 

Another term that gets brought up a lot is something called facial recognition and again, not new technology, but something that's getting brought up a lot these days. Facial recognition is simply - can the computer vision model look at a frame of an image it's going to be a frame from a video or a photo and recognize the face and then maybe even who that person is? And odds are you have done facial recognition model training. If you've ever uploaded a picture to Facebook in the mid 2010s and tagged somebody's face in it, congratulations. You were part of training a facial recognition algorithm and you know the applications of facial recognition. Some of them are really cool, like you can get into your phone without having to type in your passcode and it's a lot less annoying. And some of them are really sketchy, like maybe the government should not be doing mass surveillance of innocent people and tracking where everyone goes all of the time. It's a broad range of topics that fall under facial recognition. That's in the domain of computer vision that we talked about earlier. 

0:29:30 - Kiran V

Yeah, and the last common term I wanted to talk about is something a little more, you know, like space age, futuristic, sci-fi. We hear often about the singularity. The singularity is coming. Well, what does that mean? You know, in pop culture, sci-fi, when we talk about or we hear about the singularity, this is when humans and machines have become one, and what does that mean? Well, really we don't know, because it's not happened and it's probably not going to happen for a very, very long time. 

But this concept is that human consciousness can now live inside machines, and that is if you talk to a data scientist who's like one of those mad scientists that you meet in the lab and you're like what do you really want to do with AI? These like really esoteric individuals are like I want to create the singularity, I want to build a machine that does everything a human can do. And, you know, eventually, take my human consciousness, put it into a machine and now we can live forever because machines don't die the same way that humans do. So that's kind of, you know, one of the far-fetched terms in AI, but that's the singularity when you hear it. 

0:30:57 - Andy B

And if you're thinking this sounds like a good TV show, you are correct. There are like two TV shows I've watched that kind of talk about the singularity or show it. One on Netflix called Altered Carbon. I only watched a few episodes. It was too violent for me. But people like consciousness gets loaded onto these like discs in their head and they can switch bodies, and bodies are grown in labs. That's one example of the singularity. There's a much cuter show on Amazon video called Upload, I think, and it's about people who experience medical disasters so they have their consciousness uploaded to this, like server essentially, and they're living in an imaginary hotel in this place and it's really interesting. And how they interact with people who are still in the meat space, the flesh space, versus digital and is a digital person even the same person. All topics that are discussed in that. So if this sounds like interesting to you, you are, along with Hollywood execs, enjoying that. Go check those shows out. 

0:32:04 - Kiran V

The singularity is coming. 

0:32:06 - Andy B

No, it's not. 

0:32:08 - Joe C

I think it's coming. All right. So we've covered, we've scraped the surface of what AI is. I think we've thrown out a lot of facts. But we also want to let you know that this podcast is also going to be talking about and sort of debating opinions and theories in AI as well. So I'm sure we'll be talking and debating more about the singularity in the future. But, as you have heard, there is a lot to talk about here and we're going to be covering it in future episodes. Let's maybe talk sort of blanket about our thoughts on AI and sort of, you know, our elevator pitch on what we think of AI. 

0:32:52 - Andy B

Yeah. 

So I want to echo AI is just, it's a big thing. 

There's a lot of depth, there's a lot of breadth and on top of that, because we're in this inflection point right now, where a lot more people are interacting knowingly with AI than ever before, there's a lot of confusion in this world. Since I've been in this industry for a long time, we know that you have actually been interacting with models way more than you think you have been, but now everyone's more aware of it and suddenly everybody's like an armchair expert and there's a lot of content out there that is, at best, misleading and, at worst, outright and harmful lies. So that was some of my like motivation for this podcast, and we want to, you know, peel back the onion and make it more transparent to people so that you have information to form your own opinions about things, facts and as well as some informed opinions from us. But yeah, I personally think that AI and machine learning, it's just a tool and, like any tool, what determines its value and its ethics is the hand that wields it. That's my take. 

0:34:05 - Joe C

I agree with that and I think there's a lot to learn from history here about how new technologies get used in the world. AI, in many ways, is something new, something groundbreaking. It's something different, but in a lot of ways it could follow the path of other technologies in the past that have been used for good and used for bad. So I agree, it's sort of what we make of it and how we choose to use it, and I know the three of us are optimistic about how it can be used and we'll be highlighting some of the promising ways we see AI in the world. 

0:34:43 - Kiran V

Yeah, and I think the thing for me about AI is it falls into a category that very few things fall into, in that it is applicable to such a wide variety of things. So if we think of tools, any tool garden tools, house tools, kitchen tools, whatever tools generally have a purpose and are able to do a particular thing. If I have a shovel, it's going to help me dig a hole in the ground to put my plants in, but I can't use the same shovel to go and build a house or repair my car. AI as a tool is something that can be applied to so many different things, and I think a couple other examples of this are like the internet. There's no particular domain that the internet is used for. You can use it for health, you can use it for social, you can use it for news. There's so many things that you can use the internet for Same thing with a computer. 

Back in the day we had a Walkman, which, for those of you that don't know it, was a portable music player that either took a cassette tape and you can put a cassette, and then the cooler versions, you had a CD player. That was a tool to allow people to listen to music, but that's all it could do. You can talk to someone on the phone, you can do calculations on it, but now we have these generalized computers like cell phones and laptops and desktops where you can do a whole wide array of applications like play a game or talk to someone or do your math homework. 

And AI is one of these tools where you can apply it to so many different verticals, so many different domains, whether that's facial recognition with computer vision or chat bots with natural language processing. It can make art, it can make music, it can help you identify when people are talking about threats in a news article. And we still haven't figured out, I think, the limit to what we can do with AI, which is why it's so exciting and such a large topic and why we wanted to create this podcast. So, I think, even for us just to really understand what can you do with AI, where is this going and how are we going to get there? 

0:37:10 - Andy B

Yeah, absolutely Echoing everything that Joe and Kiran said. Right now we're just in this place where there's the people who have high familiarity and high access to AI and machine learning. It's a very small portion of people and it's a very small portion of organizations, and I think information is power, knowledge is power, and it's really going to be interesting for us to hear from you all. What do you want to know? So we're going to probably start off with some very high level introductory concepts, because it's good to just clear the air and make sure we're all using the same words for things. You should feel comfortable being able to describe what AI is and what machine learning is and what a model is, and from there we can go really into depth. As to very particular niche spaces in AI that we have expertise in, for example, Kiran and I work together in natural language processing and some really sophisticated technology. We can go really deep on how some of this stuff works and, like right now, I'm elbows deep in large language models and we can discuss cutting edge research and technology, but to me it doesn't help anybody to talk about. Look at the 8% lift you can get da, da, da da If you don't understand why this is important and how it impacts everything around it. So with that, this is your call to action. Let us know what you want to hear. 

Obviously, do the podcast thing. If you don't mind. You can find us on Spotify, Apple Podcasts, Google Podcasts, all those places. Rate us, tell us what you think, subscribe if you like what we're doing. Share this podcast. Thats it friends. Our website is aifyipod.com. Our email is aifyipod.com@gmail.com. You can hopefully find it on our website soon. Reach out, don't be shy. We would love to make sure that we're tuning this content to what you want to hear. Have a great rest of the week. 

0:39:23 - Joe C

Thanks for listening. 

0:39:25 - Kiran V

See ya.