Episode 16: "Interview with Steve Wasick, Founder and CEO of infoSentience"
Listen to this episode:
Episode description:
In this episode I have a chat with Steve Wasick, founder and CEO of infoSentience, a company that uses AI to provide millions of automated reports to customers like CBS Sports, the Chicago Mercantile Exchange, and others.
infoSentience: https://infosentience.com
AI Quick Bits on YouTube: https://www.youtube.com/@ScotPansing
Episode transcript:
SCOT: Hello, everyone, and welcome back to AI QuickBits: Snackable Artificial Intelligence Content for Everyone. My name is Scot Pansing, and in this episode I’m speaking with Steve Wasick, founder and CEO of infoSentience, a company that uses AI to provide millions of automated reports to customers like CBS Sports, the Chicago Mercantile Exchange, and others.
You’ve probably seen automated and templatized content on the web, and in my opinion it’s pretty obvious when I come across it. Steve has an interesting approach to this space and I think you’ll enjoy hearing about his company and work.
If you like this podcast please consider leaving a review on Apple, Spotify, Audible or from wherever you may be listening! I also post audio-only videos on YouTube and episode transcripts on Substack in case reading is more your thing! I’ll put these links in the episode notes.
So without any further delay, here’s my conversation with Steve Wasick of infoSentience!
SCOT: All right, Steve. Well, thank you for talking to me today. I really appreciate it.
STEVE: Hey, thanks so much for having me.
SCOT: Yeah, I'd love to get right into it and hear more about your company infoSentience. But as a setup, I would like to discuss basically how even before this generative AI explosion that we're going through right now that started with ChatGPT and Midjourney, that we've all seen these articles on the Internet that are clearly templatized and generated. In my experience, it was something like on, say, financial news or investing sites, after something like a corporate earnings report is released, you could clearly see that the balance sheet and other data were consumed and converted to, like, an article format. So how does infoSentience play in this space? What are some examples of your work? How do you fit into that? I've looked at some of the examples that you sent me and they clearly don't read like those things that I'm talking about. It really feels like there's like a writing style going on.
STEVE: Yeah, I think what you're picking up on is sort of the uniqueness of the information that's being sent to you. And I think that that contrasts strongly with what you talked about, which is when you're reading something that's clearly just sort of templated, or what I call like mad Libs styled writing, where there's just some basic structure where they wedge in the most obvious pieces of information, we pick up on that really quickly, right? Because when you're writing as a human being on any topic, you're going to be drawn towards the unique, the interesting, the difference, the wild swings, or the really interesting context.
As a writer, that's what you're going to put into your content. Whereas if you have a template, you can't include any of that information because all of those things are things that are dynamic, that may happen, but probably will not happen, right? And so if you have a template, you have to just stick with the tried and true right, so if you're talking about a report, like a quarterly report, it's like, all right, well, there was an earnings value and there was a revenue value, and here's where it is for the year and here's what the stock did. It's like all the things that are guaranteed to happen, right? That's what's going to be in there.
And it's really easy to pick up how unnatural that is when compared to human writing, where we want to sprinkle in the things that are not guaranteed to happen, the things that are specific to this particular piece of information or this particular data. So that's, I think, what you sort of responded to as a reader in the past, and certainly when I was looking at the space when I was starting, I saw those same things and felt like we could do better with a different approach. And so the approach that we take is to allow our system to go through the data, figure out everything that's interesting that happened, right, no matter how related it is to each other.
Just find first as a first step, find everything that's interesting and sort of rank order what is the most interesting and most compelling, and then give those pieces the intelligence to self assemble into a narrative that actually has, like a clear beginning, middle and end and sort of clear main points and clear context on those main points. And it's definitely a more difficult technical challenge than a lot of other people in this space who have, again, taken much more of a Mad libs approach. But it's pretty clear when you're reading it, the difference.
SCOT: Now you have some financial experience with this. I mean, Chicago Mercantile Exchange, you've done some work for them, but also fantasy sports. Do you want to talk a little bit about that? That seems like a really cool space to be in because there are so many, I don't even know how many fantasy leagues there must be. It could be millions, right?
STEVE: Definitely millions, at least.
SCOT: So talk about how you help fantasy leagues, even the small ones, with having fun with the product.
STEVE: Yeah, fantasy sports is actually our first product, which we did for CBS Sports, and we're still doing it. And basically this was the genesis of the entire company because I had this idea when I was finishing up law school where I was in a fantasy league, I was really busy. I barely even knew what was going on with my team, let alone the rest of the league. And I was like, man, there's a lot of sports that I don't even follow the games. I just like the writing. I just like the drama and all the stories.
And I was like, boy, if we could do that for a fantasy league, it would be so much more compelling to play in the league if you had all this insight sort of brought to your attention. And so that's what actually started me along the path to creating this think. Thankfully, my intuition was correct because people on CBS Sports really, really love being able to get coverage that speaks to the exact things that are happening in their league, right?
Like the coaching decisions that they made and whether they took revenge on their current week opponent for when they beat them in the playoffs last year, or what their series record with this person is over time, or how this win is going to affect their chances to move on or to make the playoffs, or just everything that you would talk about in regular sports, like which players have been just disastrous, or which player finally had a good game after all these years, or finally you put them on the bench and now they have a great game, right?
Like all these things that make fantasy sports so interesting, our system can actually speak to that and show them that and sort of let them be seen in some ways, right? Like all the things that they're doing are being picked up on by this system. That's how we started. But the basic analysis engine is pretty similar regardless of what we're looking at.
I mean, fantasy sports is very similar to having a stock portfolio where you have these individual players that you think are going to do well and then they either do well or they don't. And you can compare them versus sort of averages and other portfolios or other segments of the market or what have you. So after I built this out and was fortunate enough to get CBS Sports as a client early on, I realized pretty quickly that the intelligence that we needed for fantasy could very easily be applied to other know.
SCOT: When I think of fantasy sports, in addition to, of course, there's all of the data that's happening in the real sports world, right? All of the stats, all of the things that are going on with the athletes and the teams, all of that. And obviously that's the engine that drives fantasy sports.
But when I've done fantasy sports, what kind of keeps me involved or what kept me kind of really being entertained was if you really have a passionate I don't know what you call it, like the group leader or the manager, the person that has organized that fantasy sports league, a really good one, will send out that weekly recap, right? And it's not about the athletes and what really happened. It's about, “Oh, Jimmy just absolutely destroyed Steve or Janine!” or “This person was number seven in our league last week and they've come up to number two,” — they're actually talking about the social dynamics within that small group of people that are playing in that league, small or large.
Like you get a really good weekly recap from your league manager that's full of the drama and everything. Not just about the sports, the athletes but the actual members of the fantasy league, that is something that's really enjoyable. And I feel like that's something that you're also tapping into, right? Like your system is doing that as well. It's not just about the athletes data.
STEVE: Yeah. Thank you. Yeah, I think when we started, CBS was making use of an internal function that they had where players could post articles to the league, right. Just like you said. A lot of league managers had previously been posting these weekly articles using that same service.
So when we started doing it and posting under the name Fantasy Journalists, a lot of people were very confused as to who was doing it because it read so naturally that they were sure that somebody in their league was doing it. And they told me that they monitored a lot of arguments on the forums in these leagues where the people be like, I know you're the fantasy journalist.
SCOT: Why can't you just admit it?
STEVE: I know you're writing these. And they'd be like, It's not me. I think it's so and so. People really they thought that it was somebody in the league because just like you said, it's such a great feature when somebody does that and really pulls these things out so that people can quickly sort of digest the most interesting things.
SCOT: I can imagine a customer like CBS is very appreciative because that must keep people involved with fantasy. The time spent, the engagement, that's huge. And that's a lot of time for a league manager to sit and write something up like that every week. There's probably both sides of it. Some are probably like, oh, I like doing that, and this is going in.. But I would say it's just my feeling that a lot of them are probably like, this is great. I don't have to spend hours every weekend writing up, like a really compelling recap. It's just sort of like that and then everyone's happy and everyone's spending more time on it.
So that's a really interesting use case. Aside from fantasy sports and finance, are there any other sectors or sort of use cases that you're playing in?
STEVE: Yeah, we're in a few other verticals. I mean, essentially our system is content agnostic, so if there's a data set that you could give to a human analyst and have them pull insights from and write up, then our system can do the same thing. We have other finance products. We work with a company called Scientists.com to actually do some recaps of biostocks. Like, they have some proprietary information that they incorporate into the reports as well. So it's like a mix of their own data and public stock data.
And we have a marketing product that deals with Internet marketing campaigns, B2C stuff. We also work with Max Preps to do high school sports. We have multiple sort of sports clients. We have a medical client right now. We work with IU Health to automate doctor biographies, which is something that they really like, and we're hoping to expand to some other hospital groups here soon and then we've worked with some other companies in the past in other verticals.
We feel pretty confident that our system is able to deal with any complex data set.
SCOT: Yeah, I think you mentioned science. I think published scientific papers would be an interesting one as far as maybe in a certain subset of research, collecting a bunch of papers in a certain field, that would be interesting.
One thing I'd like to sort of just taking a step back and to get your thoughts on this. I feel like there's definitely examples and there's times when I feel like using a large language model and summarizing or generating text is super helpful. And then there are times where I see it and it just personally turns me off.
I'll give an example: on LinkedIn if I post an article that I find really interesting, and I find LinkedIn to be a pretty.. I really enjoy the platform these days because I feel like I really do get a lot of engaging discussion on LinkedIn. If I post something, if it's not just posting the same thing that everyone else is posting that day, but if I find something I find interesting and my network of connections finds interesting, sometimes the comments underneath are really valuable discussion, but then there'll be.. I'll call it “CommentGPT.” Some person that literally posts a comment on LinkedIn, that's like a two paragraph summary of the article I posted and it feels just like those mad libs like you were saying. And I'm like, “What value are you adding?” It actually is, I feel like diluting this platform, you're not bringing your authentic self to the conversation. I feel like it's almost like doing damage.
I don't like that and I don't know if it's because if that's just like, I need to get with the program and LinkedIn is just going to be a bunch of CommentGPT everywhere, but for me, that's not the right venue for it. Do you have any thoughts on that?
STEVE: Yeah, I think that it's not the exact same problem as the templated approach because those are fixed, right? Whereas LLMs in theory, they could write up about anything, but they certainly have a sort of tendency to the generic, right? And so that's what I think people pick up on when they read ChatGPT. I would say that both of these issues are testaments to just how crazy complex language is.
Because if you talk about a template, there's nothing wrong with it, right? It's like, oh, this was written by a robot because something is spelled wrong or it repeats the same word ten times in a row or anything. It's like you're just picking up on this, like, hey, this is all just the most basic information with no context or no uniqueness. Whereas the same thing with the LLMs it's like it's well written. There's nothing wrong with it. It's just that you can sort of pick up on this fact that this is so down the middle of the road that it doesn't have the sort of quirks and peculiarities that tend to pop out when something's written by a person that has a sort of unique perspective.
And so it's amazing how we can pick up on these things. And it's actually important, right? It's not just like an aesthetic preference. The reason we don't want something that's totally generic is that we've probably heard it all before, right? So just like you're saying, it's not really adding anything to say. Here's the compiled right down the middle of the road thoughts on any topic, right? That's not why we read things. We want to read something different, something unique, something that has a perspective, right? And we just don't get that from a lot of stuff with ChatGPT.
SCOT: Can we talk a little bit about what is kind of under the hood with the technology at infoSentience? What can you publicly share about the models that you use? Do you use APIs from OpenAI? Or do you have sort of homebrewed sauce models? What's under the hood?
STEVE: Yeah, we could definitely talk about it. We use what we call conceptual automata, which is basically breaking down every story into the smallest possible subcomponents and then putting the intelligence on those subcomponents. And I'll give you an example first before talking about the benefits of that.
So if you think about the world of sports, you could have the concept of a team, right, and you could put different types of intelligence on that, and then you could have a concept of a team winning a game, right, that's an event, and that might be a different event in different sports, right? You might have more points in one sport, you might have more goals in another sport. You might have more runs in baseball. So there could be subtleties of even what that means to win.
Or maybe like in soccer, sometimes you play two games back to back and you take the aggregate scores. There's lots of ways you can win something, right? But then you can also have the concept of like a streak, which is a sequence of events happening, the same event happening over time, and you can put all those three together and say, oh, this team's on a winning streak. And now you have like an actual story that you might be able to use as a component for an article. But by putting the intelligence on each individual subcomponent as they combine together, you can view that the thing that's now been combined, you can understand sort of everything that's going on with it and how it relates to other things.
So, for instance, you might have a streak of a team losing a game, right? And, well, two of those components are the same, right? Two of those three components. They both have to do with a team, they both have to do with a streak, but it's just a different event, right? And these types of complexities just get bigger and bigger and bigger and more and more complicated as you add more and more things, right? Like a winning streak can itself be a subcomponent of a story. You could say, hey, this team's coming into the game on a three game winning streak while their opponent is coming into the game on a three game losing streak.
And even as I said that to you, I put in that transition. This team's coming in on a winning streak while this other team is coming in on a losing streak. Because those two things are conceptually related, right? And as human beings, we understand all these conceptual similarities really easily and can adapt our language and adapt how we structure what we're talking about to take into account similarities or differences or understand how do I avoid talking about this one thing that's clearly just a subcomponent of what I'd said before. All these things come really naturally to human beings, but are very difficult for computers.
And so how we've tried to get around that again is by putting the intelligence on each subcomponent so that as we start to build the stories up, right, like as we start to build up the individual events and then build those up into sentences and then sentences into paragraphs. We can understand how everything is supposed to relate to each other and kind of have each of these subcomponents communicate with each other in such a way that we can end the whole process with a narrative that actually reads just like it would be if a human being wrote it.
SCOT: Yeah, I think what you're kind of getting at is especially with streaks, it's like the drama, right? We want to read conflict and drama and a story and streaks can also be on the individual level. You have athletes that, oh, they've scored X amount of points or had a double or triple double four games in a row, or even this player has been scoreless for the last seven games or something like that. So I think that does speak to the drama. It's very interesting. I think sports kind of play perfectly into this. That's great. I'm kind of curious, also, if you have chatbots on the roadmap, it seems like your product is about delivering actual reports. Do you have something that you could speak about down the road? Can you talk about your roadmap at all? I was thinking like a chatbot that you could talk to about how things have gone so far in the fantasy league. If it's near the end of the season and the playoffs are about to start, that might be interesting.
STEVE: Yeah, I think it'd be difficult for us to do that because the chatbot just adds like a whole other layer of dynamism to the language generation that we can't really handle. And it's something that's where these LLMs really shine, right, is that they can basically talk on any topic, right? They have an incredible range. The problem with these LLMs is that you have this issue of them sounding really generic. They're also not debuggable in the sense that if it says something that you don't want it to say it's really hard to understand why or get it to change. And they don't necessarily work really well with large data sets.
There's like a lot of limiting factors with LLMs but where they really shine is being able to have these open ended text based conversations and so I don't think that's going to be a place that we're really competing. I mean, I think in terms of our just we're really just trying to execute on the contracts that we have right now we have a big new thing that we're rolling out for the Chicago Mercantile Exchange.
We have a lot of new sports content which is really public facing and that we really want to show how far along the technology has progressed because we've made a lot of improvements that we haven't been able to show off because we were working with sort of an older code stack. And so now that we have everything transferred to our new code stack, a lot of these public facing things that we're doing right now, we want to turn it up to eleven and get people really excited about just how insightful and how in depth some of the things that we can do are in either sports or finance or other spaces.
SCOT: Very cool. Yeah, I'm excited to see what is down the line for you guys. I'm kind of curious also, do you use generative AI tools in the daily operations of your company? Do the employees use generative AI tools or do you have automated processes that are using any of these latest tools that have hit in the last year?
STEVE: That's a great question. I would say a qualified “no.” I'll say first off, that one of the things that our system really needs is scale because it takes a while. Since we are kind of hand modeling the analysis that we do, it takes us a while to do it and so we typically need some sort of report that's going to be going out a lot for it to make economic sense.
However, I will say one of the things that we've done is not just do narratives but we actually can take the conceptual understanding that we built into the system and use it to create visualizations and pivot tables and charts and all sorts of things like that. And we definitely use those systems when we are creating the content because a lot of times we're working with companies that they have data that's sort of inconsistent or we're trying to sort of balance what we're putting into the narratives, which can be tricky.
And so we essentially use our own system to sort of visualize the data or look at what we've produced and whether they're sort of like hotspots in terms of things we're talking about, too much, things like that. So we do actually use this essentially because the system is all about synthesis and telling you what you need to know. It's like it can actually do that for us in terms of when we're creating the content just by virtue of reading it and visualizing it. We get a really good sense of what's going on with the underlying data and whether there are issues with the data that we need to talk to with the client about that's interesting.
SCOT: Do you have charts and graphs that are starting to come along with these reports and summaries? Like even for fantasy sports, is that something that is provided or is on the roadmap as far as these reports?
STEVE: Yeah, honestly we can provide those but it's just sort of limited by CVS. Like it'd just be work for them to kind of set aside space for visualizations. But in terms of our system, yeah, we have some really cool capabilities that I think the best place to see it in the near future is going to be with the Chicago Mercantile Exchange. We're going to be ingesting all of their commodity data and basically creating a live up to the minute website covering every single topic within the system, like all the commodities, but a lot of individual metrics and asset classes and all sorts of things. And you can actually customize the website to your preferences so that when you log on, it's more likely to show the headlines and stories that you're more interested in. And actually you can even customize the types of things that it focuses on in terms of futures or options like the things that you're trading.
But one of the other things that's going to be really cool about this website is that you can drag and drop any sentence into where we have the charts and tables on this website and it will visualize that sentence or show you the data behind that sentence so that you can very quickly go from narrative to graphics and then you can actually go back as well. So if there's a data point in a chart where, let's say there's like a big drop on that particular day in this data point, you can click on that data point in that chart and boom, you're taken right to the narrative that focuses on what's going on for that data point. So it's this really cool fluid mix between stories and data and visualizations that intuitively we can do as human beings, right?
We can't actually instantly create charts and graphs, but because human beings think conceptually, it's not really any more difficult to sort of write a story about a metric going up or down or create a chart for it, right? We're able to do both of those things because they're both just based off of the same concept. And so when you have conceptual thinking embedded in your software, which is what we've done, then it's capable of doing that exact same thing.
SCOT: Very cool. Let's take another quick step back. You mentioned law school, so I'd love to hear a little bit more about your background. Did you pivot from being like, were you ever a lawyer or how did you end up being a founder of an AI company after law school?
STEVE: Yeah, that's a great question. I don't really know. I was at Northwestern Law School in Chicago, and I was in my third year, which is like your last year of law school. And I had this idea for a fantasy product, as I was telling you, that that was the impetus for the company. And I started working on it, and pretty soon I realized that the techniques that I was doing in order to make the narratives actually compelling would be able to be applied to a lot of other use cases.
And so I did like a really small friends and family round just to get off the ground, essentially, and started this company. I did graduate, but I never took the bar. And other than summer internships, I never practiced as an attorney. But I started this company and that was that. I certainly took some lumps by virtue of the fact that I didn't have a lot of training in programming. I did take it in high school, and I kind of kept up on it, so I wasn't a total novice, but I definitely had a skill deficit that I needed to correct. My first employee was a really good programmer, so that definitely helped. But there's no substitute for just sort of taking your lumps and figuring things out the hard way. But thankfully, because this field in general is so new, it's not like I missed a class on this topic. There are no classes here on this particular field, so everything was brand new anyway. In some ways, like having the background that I did. I was actually an English undergrad and then law school. A lot of it is just about logical analysis, logical relationships, particularly within language. So it gave me a pretty good background, to be honest, to attack some of the problems with creating language.
SCOT: And you must be getting ready for your busy season with sports. I mean, I'm just thinking, like, we're about to start right now. Here we are, we're almost in September in 2023 when this is being recorded. So you've got basketball coming up, NBA, you have the NFL is going to start pretty soon. And then I'm kind of curious also because gambling and online gambling be sort of coming on state by state.
It just feels like just from my little bit of paying attention to it, it feels like the Supreme Court a few years ago kind of sort of said like, we're going to kind of leave this to the states and then one by one it sort of feels like different states are coming online and permitting gambling in some form or fashion. And it just feels like soon there's a near future where people are going to be on their phones doing like little prop bets, like $5 on this person's basket or whatever all over. It seems like it's going to really balloon as an industry. And so are you looking at that to get in there? Is there a space for you in that ecosystem?
STEVE: I sure hope so. I mean, we're definitely in there right now because we do live sports for CBS where we cover a whole bunch of sports. We cover college and pro football, a lot of European soccer leagues, college and pro basketball, some other ones that we're adding. And so basically we provide previews for those games and also some gambling focused content that kind of like previews the game from a gambling perspective. And those are really valuable articles, and we're certainly looking to add to that, because I think you're right. And everybody sees where this is going, where now that gambling has been legalized, it's becoming a huge industry. And it's really just at the beginning. Because even though the Supreme Court removed the national prohibition, there were still a lot of prohibitions within individual states. So it's definitely a process as far as the rollout of gambling across the country. And it's proceeding apace, right, with more states being added all the time.
So everybody knows that this is like a huge competitive landscape where there's going to be a lot of first mover advantages in terms of locking people in to a particular betting platform. So there's a pretty big scramble there and content is king when it comes to that. Like how do we get people to go to our site versus somebody else's? Well, it's all about just having better content, more content, timely content and that's all things that we can help out with.
SCOT: Yeah, I would think that the fantasy sports is like you would have a natural evolution into that gambling arena and that I could totally see a partnership with a DraftKings or a Penn Gaming now that they're doing something with ESPN. And it feels like the summaries or that the generated text could come from that perspective, like the lines. And from the gambling perspective, you have also horse racing. It just feels like there are so many elements of sports that are now that the gambling is really becoming super popular. So that's why I asked that. This is fascinating.
Is there anything else on the roadmap that you want to talk to me about or is there any like I said, it sounds like you're gearing up for a big fall autumn and so what else is going on? And what types of customers are you targeting to expand out beyond sports and finance?
STEVE: Well, we don't know. That's the short answer because it's a double edged sword. Of course anybody's talked about VC stuff or entrepreneur stuff you say, oh, well, we can sell to anybody. It's kind of like, well, you kind of sell it to nobody then because you're too broad and that's been a challenge for us because I think we really are really capable in terms of we can analyze any sort of data set but one of the key limiting factors that I mentioned is scale, right?
Where we need somebody who has a lot of reports that need to be built. Because if it's just a quarterly report that somebody's doing it's like, great, we can save them that time. But maybe it's one employee spending 10 hours and it's just not worth it for us to do the report, let alone the fact that if they're spending 10 hours on it, they're going to do a better job than our system is. Because at the end of the day, like, a human reporter is still the best possible is going to write the best possible report.
So we need to find somebody that has a lot of data where there's a lot of analytics going on with it, right? It can't just be synthesized into one or two numbers. I always give the example if all you care about is who won between the Colts and the Titans you don't need our system, right? You can just look at the scoreboard but if you want to know, quote unquote what happened? Well, thousands of things happened that have implications left and right of the past and the future. So if you want just a synthesis of the 15 most interesting things that's where you need our system. So we need to find those use cases where you have a lot of data that needs synthesis, but you need it done at scale. And it's hard a little bit to know what those reports are going to be from outside companies.
And so generally, we just try to get intros with typically larger organizations to just talk to them about what their data needs are, because usually they have this problem, right, where they have something that they have data scattered all over the place. They don't really have a good handle of it. They've got all sorts of people who want to access it and some of them need a report on this and some of them need a report on that. Well, it's like our system is the perfect solution to that, right, where we can come in and do that at scale and instantaneously and 100% accurate, all these different things. So I know that's not a great marketing answer to say I don't know but hopefully anybody who's listening to this, if that's you then please reach out because we'd love to just have an intro call.
We can show you the demo that we put together for the CME, which is I think everybody who's seen it has been pretty blown away by it because it really is showing awesome capability that I don't think anybody else has. So we can kind of show you the demo, talk about the data issues that you're having and hopefully come up with a solution.
SCOT: Yeah. The website is infosentience.com, right?
STEVE: Yes.
SCOT: And so I think also another thing I was thinking is that an interesting audience for this might even be like journalists that need to prep for going on air a game just finished or something and they need to quickly prepare something to go on air and speak about or even let's take this a step further. You could launch your own sports journalist network with AI avatars speaking your summaries to compete with Fox Sports, right?
I mean, if you really wanted, that would be kind of an interesting bizarre play, but that's something that or people could use your product to launch something like that, like another company. Like, “These summaries are wonderful. Why don't I just create like a little avatar with did and use the voice from Eleven Labs and have like a little sports newscast reading your content?” I think there's going to be a lot of these virtual entertainers in the so yeah, it's wide open. And I really appreciate you speaking with me today, Steve. I've learned a lot and I really enjoyed learning about your company. I think this is a fascinating use of AI. I think it's great to see it being used to do something additive know, like you said, it's not just a summary. Like you said, if you just want to know who won the game or you just want to know it's not a Mad Libs. It's like this feels like it's something that actually is creating a story, the drama, like I said, I really enjoyed hearing about it and best of luck and I'd love to stay in touch and thank you for coming on AI Quick Bits. I really appreciate it.
STEVE: Scot, thanks again for having me.
SCOT: Absolutely, take care.
STEVE: Thank you.