I respective school's code of conduct. The president of Harvard University, Claudine Gay, has apologized for her response. I'm Scott Carr. Damage assessments and cleanup efforts are underway after a deadly tornado outbreak in Tennessee. At least six people were killed dozens more injured when multiple twisters tour through several communities, including the Nashville area Saturday evening. I'm Chris Caragio, NBC News Radio or on board
kcaa's Inland Express KCAA Home Linda NTIDAM the station that needs know. This year behind, the information economy has a ride. The world is teeming with innovation as new business models reinvent every industry industry. Inside Analysis is your source of information and insight about how to make the most of this exciting new era. Learn more at inside analysis dot com, insideanalysis dot com. And now here's your host, Eric Kavanaugh. Oh yes, folks, welcome to the future.
Indeed, your host Eric Cavanaugh here on the only coast to coast radio show in the US of ATA. It's all about the information economy. Inside Analysis doing another one of our pre records for the holiday season, and I'm really excited to have jig Nash Patel with me. He's with a company called data Chat. He's at a couple of companies, acquired one by Twitter apparently, which is pretty impressive. It's done a lot of interesting things, and
of course he's also at Carnegie Millon University. But we're going to talk about really the trajectory of answers because that's what we all really want. Right We've done data warehousing, we do data science. All point is to get to answers that can help us solve business problems. And I get the feeling that he's managed to find a way to short circuit some of that stuff, but we'll find out. So jig Nash, Welcome to dm Radio. Thanks so
much for your time today. Tell us data Chats. Where did the idea come from and what are you working on? Yeah, thanks Eric for having me on and hello to your audience. A real pleasure to be here. Data Chats started with an idea for my students and I and that's what we do in research. We start to look at problems that are a few years out. So back in twenty seventeen, when data science was really starting to
become popular, we noticed a trend that data science. The way was practiced then and to a large extent, how it's practiced today, is to provide the tools and mechanisms for data science programmers to write code from which basically allows
them to do their analysis. And what we realized at that time is often the way people wrote their data science workflows in the form of what's called as notebooks, was to fill one cell of the notebook at a time, and each cell might do a specific task, like I might have gotten a bunch of data and I want to extract the area code from a column that has
phone numbers information in it. And the way people would do that is they would go and type in that question saying, how do I extract area code information from a string that's a pull number, and to google our sack overflow, find pieces of code, try to put that into the cell, and then try to adapt that. That's how they would do solve each piece of the data science puzzle one step at a time, said can we automate that? Can we do something bigger with it? Yeah? And that's a pretty
interesting observation because what you're looking at is the actual de facto workflow. What are people doing on these machines. How does that process look and you saw into the future a bit here with the ability to leverage what we now call large language models. It's just a variation of artificial intelligence. And as we
say on the show, the lms are really just pre divengines. They're predicting what text they think you want to get based upon your prompt and based upon what they were trained on, which by and large was this massive corpus of text out there in the real world. So you kind of saw I think into the future a bit, maybe explain that and what you've built. Yeah, that's great. So what we notice is there's a lot of repetition in what people tay to do. People are not asking totally random questions all the
time. They're often asking similar types of questions, if not exactly the same. Similarly, when stackover, for our Google is returning answers back to them in the form of yes, code, I've seen that, similar it's doing it based on a retrieval mechanism. And Eric, as you pointed out, the lms effectively learn a ton of information from everything they've seen, including code and their synthesis machine that complement what searched us and they'll synthesize new code.
So we started to think about the automation that is potentially possible back in twenty seventeen by having the human just ask the question and synthesize the code in one form or the other. And of course at that time LLMS weren't available, so we were using what by today's standard, or even by standard by the standard just a year ago, might look very rudimentary, but basically using a lot of traditional language of technology, passing sentences, trying to understand the grammar,
understanding the sentence, and then trying to produce the code. But with the limbs, all of that changes in a dramatic way, and all of that that we were dreaming at that point is now started to become real in a very big way. Yeah, and we are at a very strange time right now. Let me just throw out some context for our listeners here, because you get accustomed to a certain standard or a certain paradigm, and we're all very much accustomed to the Google paradigm. You Google for something, you
get a result, you go on about your day. Well, Google, of course indexes the web. Yahoo started indexing the web with hadoop, right, and the HDFS had doop distributed file system, which was then tried as an analytic engine by cloud Eira and map R and hoar Works and all these other folks, and that kind of went sideways after a while because there was too much reverse engineering that had to be done. But there was a whole
ecosystem that built up around this stuff. So I mentioned this because I'm watching as Google is straining, and Google has all this pressure from AdWords, and now when you search for something, you're going to get ten to thirty different AdWords listings before you get an organic search result, which is great for their business. They make a lot of money on all that stuff, but it's skewing what you find, and so it's making it harder to find the stuff
that you wanted. And then of course we've had all this SEO search engine optimization stuff where people are trying to understand the system, gain the system if they can. I played around with tactics to do all that stuff. But my point is that what we have accepted as a standard, which is Google
and how Google retrieves information is really in flux right now. And it's in part because of the ads, it's in part because the web is a fast moving entity and lots of things are changing, and it's in part because of these large language models that are severely disrupting how people go about ascertaining information. So what do they do? They hallucinate sometimes, right, they make up
stuff. As my buddy Eugene Burke, he's working on a stealth project with us right now, mentions the problem with the llms that they don't have an epistemological barrier, meaning they don't know what they don't know, and that's when they just make stuff up. And I've actually had a thread going with the go to market guy, Adam Something from open Ai who came out and said, point blank, look, remember the hallucinations are more of a feature,
not a bug. We've designed these engines to fuse information together and then deliver generate new content. Okay, So for certain use cases, that's fine. For a creative copy, for short stories, for articles that don't have to be really precise in terms of what content is in them, that's fine. But then you get into the enterprise space where it becomes very important to be
correct and to know exactly what you're doing. And it seems to me that you with this data chat you are kind of filling in one of the missing pieces that will allow for this next generation of technologies AI to really come to fruition and get us to a place where we can trust what it's saying,
just like we've historically trusted what an enterprise data warehouse has told us. But just if you would comment on where I'm going with that, what do you think about that, and what do you think about how data chat could be one of the pieces to rebuild this puzzle. Yeah, and you know, pick a company your thread. I think of the technologies as two distinct collaborative technologies. One is search. Search allows you to say I'm going to specify
something and give me a precise retrieval of information from the original database. So that's Google Search. You see that in warehouses too. When you fire up a sequel query, you don't get that precise result back. And you can think of when we go to Google Search, probably what we want most of the time is search. The second part what lms do is synthesis, which
is different. Synthesis says I'm going to blend together information that I know, sometimes making stuff up to give you a coherent answer, which may or may not be based on facts in the database over which I've learned. Search and synthesis are two separate components, and synthesis is what has now become possible. Right, so you allude very well to what Google's trying to do and do search and synthesis and how that blends together in the search paradigm is tbd.
But that's kind of where a lot of the struggle is how do you blend ELMS with search. We still want search for many aspects, as we might all potentially agree with it, right, sometimes you just want to search. Coming back to data chat and how that's connected is what we do is we'll say point us to a database, a structured database, even a CSB file, whatever is your data, and then ask a question. We will convert
that into an answer. Now, remember is it totally a little while back, previously we were trying to figure out how to write the piece of code that could produce the answer that was twenty seventeen. We've moved well beyond that in data chat, where what we'll do now is you give us a question and will produce you an answer. That question might be a search question, it might be find me this thing in the database, find me the customer
that produced the highest return, highest gross value, and sale. And that's a retrieval question. There might be a synthesis question or a prediction question that might say, predict for me, which is the customer that's like to be the most profitable over the next one or year. To the user, we've simplified the interface with the UX is just typing in that question. You'll synthesize that into appropriate code that might be part CQ, part machine learning code that
might be written in Python, bias automatically, part of it. Visualization in JavaScript will translate across three different languages. The user doesn't see any of that, but gets an answer now. And now the question is how do I, as a user take that answer and do something with it. To do
something with it, I have to develop some level of trust. And this is where the unique aspect of data chat comes into play, where because we are academics and rooted in academia, we took the science part of data science seriously from day one. Science requires at least two pillars. One is reproducibility
and the other one is transparency. What data chat does today is it will give you back your answer, often in the form of a picture, but along with that, it will give you back a step by step of the components that went into producing that answer. Think about a cookbook, right, you look at a recipe and you'd say, I've got something on my table. If you're given a recipe for that, you now have an idea,
a very precise way as to how that was produced. If that recipe is correct to the food you see on your table, that is transparent and reproducible. The same philosophy in data chat. We won't give you back code. We've invented a new language called Guided English Language, which is in English, but it's a subset of English. And you can get into that whole theory
next if we need to. There's a branch of linguistics that says, if I want to communicate with you precisely in natural language English, for example, then if I just use grammar in the full richness of the language, there's a high chance that I may say something that may be misinterpreted because language is
not precise. It's not wrong, sure, but there are ways of doing taking portions of that language, reducing its structure so that you get precision in that and be able to communicate and that's what we've done in data chat, where you'll get back a step by step recipe in English, so it might say I did X at this point, then why, And just like a recipe that you see in a cookbook which uses very few verbs to be precise
and to avoid this miscommunication. For example, recipes will nearly always say saute for that action. They won't do different synonyms of that, and the sentence structure will be very simple on purpose by designed to get that precision. That's exactly what we do in data chat. We'll give you back the step by step recipe in this language uses all these properties of being precise and allows a user to induitively understand what happened at each of those steps. We'll go further.
You can treat that step by step recipe itself as a program. Right, I gave that to you. That's transparency. Second part was reproducibility. We'll give that step by step recipe to you as a program, even though it doesn't feel like a program. Imagine getting your cookbook in pressing the play button on step number one to sort a garlic, and that just happens. And that's what we'll do. We'll take the recipe and allow you to play step by step see how it's been cooked, so that you can reproduce that.
So we get transparency to make it easy for anyone to understand using the same paradigm of simple English and recipe. And then you sability by saying you can actually play it step by step to see how it was cooked to produce that final answer. Yeah, this is interesting. So what you're talking about doing is enabling the long coveted conversation with your data. That's the mission. So you've got a database, you can use SQL to query the database.
That's one way. Then you need to know how to write SQ code unless you have a tool that does it for you. I mean, I'm old enough to remember a company called Progress Software, it's still around these days. It had a tool called Progress Easy Ask. This is like sixteen years ago or something. And what they would do is very interesting is you would type in a natural language query and it would create the syntax of SQL for you
so you can see what it was doing. You're doing something similar to that where you're using a natural language query and then you're underneath the covers using a variety of different things. This guided English language for example, to spin up bits of code to solve for that particular question. I'm going down this road because it's it's very interesting. It's similar to what we're seeing with the large
language models and enterprise versions of them. And here's my personal theory what we're going to see here, Every serious company of any significant size will have their own AI model. They're going to pick their poison. They're going to choose Google or Microsoft with open AI, or perhaps Anthropic or someone else is going to come along and do AI models, and then they're going to start to
train them. And what I've learned in this recent journey of trying to research this stuff is you can either embed data into the model, which is probably not the best idea, or you embedded into a vector database and then use that as your anchors of truth. Is how someone described it to me. And that's sort of enabling in this, of course concept of rag of regenerative like augmented what is it retrieval augmented generation right where you're trying to it's like,
Okay, I got an idea what I want. I want to go to the world and on them see what it is. They'm going to be all right, make sure that is what I really wanted based upon my priorities or my data which is in my probably vector database or something like that. Long way of saying, what I'm really interested to see how this How what happens here is that these llms or even this technology, you've got data chat if curated properly with the right embeddings, the right embedding strategy, the right
rag strategy. Why do I need to do all that heavy lifting on the data warehousing front anymore? Isn't this gonna get to the answers for me? Well, the answer right now is maybe not. We don't know because it still has these accuracy problems. But if we solve those accuracy problems, that's a serious deal. And it seems to me that you are embarking on a very similar journey in allowing for this interaction with the data, and you've even
got a dynamic app building component to it as well. But what do you think about all that? Yeah? I think this is these are there's a whole bunch of questions in there. Let's break it apart. First, is we've been looking for this holy grail of how do we get programmers not to be the gatekeepers to getting answers from data is one example. Microsoft used to have something called Microsoft English that they disbanded. Problem with all of these tools
is twofold. One is they never quite worked even in a single task, say generating sequel. The bigger problem, however, is today in the modern world post data sciences creation, which is like now were about a decade into this data science thing. Data science is a lot more than writing sequel.
Squel gives you a way of getting a report out from the data, but data science then requires and you could use SQL for feature engineering portion, but to really get deeper insights, you have to feed it into machine learning models, which is a different language. You know you're going to do that in Python and stuff like that. Then to visualize that you're probably going to put that into a different language too. So we've never really solved the data science
in an automated fashion, which is what data chat does. As this said, on behalf of the users that same recipe. Part of that might have been sequel, Part of that might be some Python code that we wrote. Is JavaScript. We go across three languages and you don't have to know either of those languages. The other big aspect is who is the user progress and
all of these other tools. Even today, a large amount of effort is being spent by companies like data Breaks and Snowflake and everyone else to write seql code. But the user then is a programmer, right right. We'll tell you what. Let's pick this up after the break. We got a break coming up here, folks. Don't touch that down. You were listening to Inside Analysis. Welcome back to Inside Analysis. Here's your host, Eric Tavanaugh.
All right, folks, but having a fascinating conversation with jig Nesh Patel with data chat. He's done many other things too. And I'm getting all excited because you know, I've watched the world of data warehousing, I've watched the world of data science. I've been amazed that those are a very different world. These people often don't even talk to each other, which makes exactly
zero sense to me. But I think that the economic pressures bearing down on us right now, plus the power of these new AI models is going to force is going to catalyze, basically, and it's force organizations to get a lot smarter about what they're doing and how they're doing it, and paying attention to the value that they get from all this stuff because what's happening. And
I have an old buddy, Dave Wells. He was the education director at the Data Warehousing Institute years ago, and he used to always refer to something he called under the hood technology, and his point was that you don't have to know how your car engine is working. To know that it works and to have faith in it and to drive it to work in drive it places. You don't have to know all that stuff. The same is becoming true
in this data space. And it seems to me, Jignest that what you're doing is really pushing that all forward because you're saying, look, we're to handle the assembly of code where you're going to choose which language is necessary for this particular use case, and just take care of that stuff for you so you get the result that you want. Because all anyone wants are the answers.
That's the all point, the all point of doing data warehousing, and the whole point of doing data science is to get to answers to more and more interesting and compelling questions. And that's really where you're focusing your attention. So continue where you were. I know why, I'll cut you off dealing the last segment. Go ahead. Oh that's great. You exactly got it. Is the whole point of collecting data is to get value from it.
You get value from it when the data actress in your organization, including an especially the non programmers who can make business decisions, can directly ask the questions in a way that is trust that they can trust. And that's really what data chat does right. And there's a whole bunch of technologies that now makes it possible. But so far, even when we've gone into warehousing, we've said, okay, we'll all this money into getting an enterprise data into a
warehouse. Then what we'll do is to allow every data actor to get access on it. We'll put a dashboard up for you, or we'll tell you, hey, why don't you become a sequel programmer. Oh, by the way, if you're a SEQL programmer, good luck, you can do data science. You also have to become a Python programmer. And oh by the way, if you're a Python program if you want to do some visualization stuff, you have to become a programmer and some visual language maybe Tableau, maybe
JavaScript or something else. So it's become impossible for data actors that are not necessarily programmers to do that. And that's kind of what we are trying to do with data chat is saying, forget about the language will derive code for you in different languages. We just focus on the question and now that the programming language is English, and it's just not English to seqel, it's English to plus Python plus JavaScript for us really more powerful than what you hear at
the industry talk about saying will generate sequel for you. That's just a small part of the puzzle for the data science. We've lost all of these three and allow anyone in your organization, not just programmers, to really directly derive the value of data without saying I need to find a ticket with the Tableau team to create a dashboard for me to use it right right, right,
right right. So, and the other interesting thing here is to your point, a SEQL query is going to access bits and pieces of data that are important for telling your story. But they're just component parts. They are not the mosaic, they are not the picture of what's really happening. It's a part of what's happening. But you know, I even joke about this on the show that you know, data warehousing was born forty odd years ago. When processors were slow, pipes were thin, storage was expensive. All of
those factors have completely changed. Now processors are incredibly fast, we can parallelize them, the pipes are fat, storage is cheap, so all those dynamics are inverted quite frankly. And what we did back then is we had to strip out all this context just to get the stuff through the thin little lines into a model that we can then analyze an understanding query, like an O
lap cube for example. And then you have to reconstruct all that context, either in your report about it or a dashboard, or in some way you describe the situation to your board or your stakeholders or whatever. The point is, we had to strip out all this context, and it seems to me that now you really don't have to do that as much. And what you're enabling is this conversation with a data set. Now, I'm curious to know what kinds of data sets can you tackle. Can I connect you to an
ERP for example, or a clickstream analysis system? What can I connect data chat too? Specifically? Yeah, so data chat works with tabular data, so you can connect us to a warehouse, you can connect it to CSV files, s park files, so that sort. And sometimes people might take data that might come from ritual data sets like images, convert that into features and that becomes tabular data that you could point to and build models on. So have you focused on the tabular data at the point in a journey.
At some point in the future, that may change, but that's the big portion of the market that we are addressing right now. Yeah, that's that's interesting, that's good. And then just to help our audience understand these large language models, they use vectors, so you get embeddings and they vectorize,
which basically means they convert text to numeric values. And it's my understanding the reason you do that is because it's very easy to use existing technologies to compare and contrast and find similar numeric values across this expanse of embeddings, and that's how they kind of operate. So your approach is that we're to focus on the tabular data because it's very well known, it's very common. Everyone understands
Excel and spreadsheets and all that kind of stuff. So you're really tackling what is a major component of the data world, right Yeah, And that's what the business analyst. Ultimately you're going to need to connect that to some tabula data on which you make decisions. Right, more than ninety percent of decisions, believe it or not, even today get made in Excel. Yeah. When people can't make a decision and Excel because the data is too large or
something else, they'll start to make it fit within Excel. And so, in spite of decades of us working on data warehouses, faster, sequel, paralyzation, all of that stuff, sadly a lot of decision making still doesn't happen with all of these tools because ultimately, the decision makers are not necessarily programmers, and they are not necessarily even if you point them to a data breaks or a snow if they can give them an access to the database,
what are they going to do? Systems are complex, right, And that is the challenge. It's like the bottomneck to making fast, agile decision in a correct fashion based on the truth and the data. Today is not how fast your database engine goes, it's how fast the human goes, right, And it's all about human empowerment in a safe manner that allows you to move
forward faster. Yeah, that's interesting, and so let's talk about again, someone uses your technology, they use an English language question of a particular data set, and under the covers. What you're then doing is parsing the meaning of the words of the syntax, then mapping that to the data, and then using a combination of languages. And this is very interesting to generate your visual answer. So it'll do SQL, but it'll also do Python on some
other various languages too, like JavaScript or something. And what's interesting there is that I try to explain to folks, languages and programming are a lot like languages in human languages, meaning they have different syntax. They are not just a different set of words for the same set of objects. It's really a different way of viewing the world. And they each have their own strengths and weaknesses. You know, SQL is very good at getting discrete bits of data
from a database, but Python is very good at weaving things together. I remember learning about ten years ago that in the risk management world, where I spent many years doing webinars for the Global Association of Risk Professionals. In risk management, Python is very popular because the risk managers have five, seven, twelve different systems that they're supposed to be monitoring well, the signal from any given one of them might be hard to understand or what it might be such
that it's difficult to know and there's a problem. But when you aggregate them, when you look at the big picture, that's when you can start understanding, oh, wait a minute, if this is going up and that's going down, we've got a problem. And so Python is very good at that capacity to kind of weave stuff together. Is that one reason why you chose to use these different languages under the hood to be able to tackle just about
any natural language query you come across. Yeah, great question, going back, and this is a perfect way to connect to the earlier half of we talked about sequels. Very good at retrieval stuff and putting together reports from that retrieved stuff. That's what it excels at. It's good at ETL where you're transforming the data a very structured way. Python which is a whole ecosystem, and the machine learning part of that is what's super exciting, including in the
example you mentioned, it can build models on the data. The models can then explain patterns in the data. It can synthesize beyond that nlm M and that is a model. So you need data, you need to be able to extract it and search it, but you also need to synthesize models. And that's NOTQL. Sql can't build models. They can't build machine learning models. It can barely build a regression model. Forget about the more complex models that you need to build. You say, what's interesting in my data?
Right? Right? It can do predictions. So all of that comes from modeling data plus model has been like oil and water. And that's why people have to learn both SQL and Python vs say you have to learn neither. Let's take care of all of that stuff for you and you and you will
reveal what the recipe was. Right. That's the transparency side. And this, you know, one of the big challenges that the lms are going to have in terms of getting enterprise penetration for very serious decision makee is the lack of transparency. And if you don't have transparency, then you don't know how it got the answer. And if you don't know how I got the answer, you know and you got the right answer. It's a very difficult thing
to do. And I remember watching as the data bricks folks were demonstrating how they were curating these models and just thinking to myself, man, you know, we always say, don't boil the ocean. So if you're going to go out there and try to like throw all your data into a large language model and they, oh, we'll just curate it, will you, and it'll be done by the end of the thirty third century or something, it's going to take forever to do that. So you need bite sized chunks to
be able to handle. And that's why folks are talking about frequently asked questions or customer service for example at chatbots. There are some really good use cases for these kinds of technologies, but they are discreet and they are not the sort of bigger, broader, what's really happening in my business kinds of questions.
And that's what you get when you synthesize, to your point, the sequel the data that pulled out, but the model that I can build on top of the data, and just to explain to our audience, maybe models they're also recipes, right, they're like, okay, look for this, once you find this, do that, once you see this, do this, once you see that, do that, and it's sort of a process then shows you what the world looks like Okay, it's not really what I
wanted to try. Again, you train the model, you play around with the model. But that's kind of what the models are, right. There are complex recipes for how to interpret data and then generate an answer. Right, yeah, perfect. And it's even more exciting now because it's imagine I'm working on a problem and maybe it is predicting sales. Now, maybe that we say there's a model for it. Right. It might be a regression model, or it might be some of the more complex model. But that's
not enough. What you really want is to say, on my data, which probably looks different than someone else's data trying to solve the same problem, Right, how do I build a model on the fly that fits to my data to answer the question? And that's what we do. We'll build a model on your data on the fly, use your specifically created model for you to answer the question that is really relevant to you. So interesting that requires all kinds of rocket science, data science, see wizardry to build models.
We're saying, don't worry about it, we got that covered for you. Wow, data models for on your data on the fly, and explain that to you intros you can understand and that's that's the magic, and that's that's it's all centered around human productivity is the goal. You know, I'm gonna throw a bit of a curve ball question at you and folks, Jig did not get any of these questions in advanced just so you know, we don't. We don't roll that way. But is there some calculus or some math
underneath all of this? The reason I asked that is I remember from calculus class. The key is that you had to learn to identify the pattern of the problem and then know which formula to apply, and then of course you have to do the work to apply it and to see if it reasons out. But there are X number of formula basic formula that you're using calculus to unravel problems, and it sounds to me like there's something like that underneath the
hood with your technology. Is that correct? Absolutely? And this is well known in literature. So we try a bunch of different models, and the number of models you can try and your data is not just like what model classes like what it sets of equations to try, but also parameters for each
one in that space is extremely large. All kinds of intelligence built into a platform that will look on the fly at your data, look at what you're trying to do, and say, you know, I don't mean to try the billion options because to your point, that will come back in the twenty fourth century, right if we just we will then use intelligence built into the platform to understand the context from the data from your questions, to build a
small number of models, not just one small number of models on the fly, raise them against each other to see which one there's better an answering your question, and then give you the answer from that, and all of that will be fully automate for you. Yeah, So that reminds me of what data robot would doing a few years ago before they ran into their very strange problems. But what I loved about data robots approach was this auto mL.
So you would load your data in like let's say it's mortgage loan data, and you want to just see a regression mode you understand, okay, where did loans go bad versus where did loans stay good? And what they would do is automatically run that data against like seven or eight different algorithms and figure
out what one is best. So you're doing the same thing that they're doing in that sense, but you're actually dynamically spinning these things up because again, when you consider the permutations, when you consider all the different ways you can
get covariance between functions, right, there are so many possible answers. So if you can automate that early process and get to a handful of models, that's a tremendous advance in being able to get to that specific model that's going to answer your question, right, Yeah, absolutely, And you bring up
an extra excellent point. Industry has been focused on very small things. So data robot was trying to solve part of the data science problem, but still targeted towards largely programmers, but not SEQL, not visualization, not that. And really the point is you want to do all of that and do that in a way that is accessible to everyone. Yeah, you know, we're going to pick up after the break. This is a fantastic conversation. Don't
touch that doubt you are listening to Inside Analysis. Welcome back to Inside Analysis. Here's your host, Eric Tabanaugh. All right, folks, back here Inside Analysis. What's fun is we're moving through the whole evolution of data management and data warehousing into data science and AI and capturing all this stuff, and I promise you folks, it's all turning around. Right now. We are
at the ultimate inflection point in this industry, which is very exciting. It's this is very very good news because all the plumbing and the hammering away. Let me tell you, it's not exciting, okay. Like the poor data engineers who are building these pipelines and like waiting for the data to get there, they're pulling their hair out if they have any hair left. It's a very difficult job to do. But a lot of that is going to be
solved with this new era of tools and technologies. It I'll bring jig nsh Ptel back in from data chat. You know that I completely understand the value the synergy. In fact, there's a German concept of gestalt, which says that the sum is greater than the whole of its parts. And when you get these things together, really cool stuff can happen. I'll just throw one
last thought at you. I came up with this idea a couple of years ago when folks were saying, oh, it's AI versus data warehousing, Like, it's not AI versus data warehousing, and a way, data warehousing is like your left eye and AI is like your right eye. And when you get those two, then you get depth perception because if you only have one eye, you don't have depth perception because you have no way to triangulate where
things are right. And so what Jignesh has been working on here with data chat really, I think is the next step in that process, and it really is that synthesizing component that will take your natural language query, figure out what the data is telling you, then build a series of models that they compete with each other to figure out the answer, and then you get the
answer and you open the commander to show them the whole process. Right, go ahead, Yeah, and that's exactly right, you know, coming back to the different the journey that the industry has gone through. First of it's all about can we even afford to keep all of this data together and make it That was the data warehousing the cloud are doupe world that then trusted it into big data stuff, which is all about moving that often to the cloud
so that you didn't you could make that economies of scale even better. And then came the data science world, which is about saying, hey, sqel's pretty dumb, it doesn't It only knows data, not models. It's two models and all this really clugy gluewar and companies that just did one versus the other? And yet now and that's all twenty ten right solution, and as
we look forward, it's about neither of those. These are all just foundational blocks on top of which the goal is to enable the humans to make decisions. Whether it is using data warehousing, sequel technology or data science technology. No one cares. What you care about is did you get the answer that could move your business in a different direction That would have been very hot for
you to determine without the data being present. So of course you need the warehouse, but you don't necessarily need to deploy an army of programmers now that has to do an el pipeline and a melt uning pipeline and then presentation stuff. It's like, can you bring it up to the front where the humans are in charge and the humans don't have to delegate to other humans to get
even simple stuff done. It's the true empowerment of the data actors, the human data actors in an organization is to free them from having to go to school for six, eight, ten years to become programmers, because that's just a mean to an end. The end is to get insights. Yeah. No, that's exactly right. And it's almost like you're delivering what a lot of people back in the day thought Tableau would, right. I remember the
first I actually saw. I think I saw the first ever public demo of Tableaux because I worked at the Data Warehousing Institute and five and there was a conference in Seattle where they were just part of one event after the show, like one of these parties that you have and give way free beer. And I watched, and I saw the first ever. I was the first person in the room because I worked there, and I watched, I was like, oh, that's going to be good, you know, because no one
has done this yet. That was just one tiny little piece of the puzzle. But when you can synthesize all these various engines to generate again what we want, which is the answer, right, And it's like there is such a tremendous opportunity cost to doing things the old fashioned way because, as you suggest, if you have to ask for permission from this person to do this, wait for that person to do that, Oh, that's going to be a week. He's on vacation, like all of these pauses just rush creativity,
crush innovation, and just kill the spirit. Whereas if I can just answer or get a question answered in this sort of visual format and then understand how it got there, now you're teasing the brain. You're getting the creativity rolling, you're getting those juices flowing in your mind. That's when good things happen. That's when you figure things out and go, oh my goodness, this is I know what we have to do. Now. It's not this
product, it's that product. Okay, Bob, let's go do X y Z. That's the so called aha moment that only happens if you're in this fluid conversation with the data. And that's what you're enabling right exactly, and the whole vision behind data chat. Take those words data chat first. That chat portion always meant had two meanings to it. One is, I, as a human, should be able to talk to my data and get trusted
answers that I can verify and reproduce. But the other part of that is, Eric, I could chat with you about an insight that I've found, and this is a human to human chat that it enables give you that visualization that came to along with the recipe. And now, because we have a common grounds for saying how did I arrive at this insight in a fact driven way, you have a better human to human conversation. And you might say, you know what, step number three in this recipe looked at all the
data. Oh, but you know what we should take away twenty twenty and twenty one because there were aberrations and let's go read this stuff. So let me go change that step in the recipe. So it's just as powerful in terms of what we do is enabling an individual to talk to the data directly without any intermediateary. But it's also about what we as humans can do because now we are coming at it in a factual, data driven fashion, in a truly data driven, factual fashion. Right, we are not guessing what
are we talking about? You can look at a Tableau dashboard and what does it mean? You're just gonna assume it. This means what it means? Can you interrogate it? Can you say did it miss a step? No? But that human to human chat portion is a big part of a vision, and it all connects if you have this way of explaining how you came to your answer in the first reproducible way. It enables both these modalities of communication, and that's what is super exciting, right And you know, I'll
end with a bit of an analogy here. So online and social media and via email, people get to these big fights. It appens all the times. Where all this hate is online is because people are kind of going at each other. But if you just sit down and talk to a person, you'll always be amazed at how much you actually agree on. And a lot of that friction results from misunderstanding, misinterpretation, and really from superimposing your beliefs
onto someone else when you don't really know what they think. You just know what they say or what they've posted on a chat, and you can't read sarcasm in chats and things of this nature. It's hard to So these are very serious problems in communication that as solved if you just talk to someone and what you're doing is enabling kind of conversation with the data, but then also with your counterparts and the organization, so you can collaborate on working through what
we just saw here. That's a whole different story from debating whether or not it's true. And then Okay, next month's meeting, we'll try to figure it out again. Well, you just lost another month, and we can't afford to lose months these days, folks. I'm telling you like that the pressures from the economy, from global forces, from artificial intelligence, all this stuff is really bearing down and forcing change, forcing companies to get real about
what they're doing. And I think you've got a pretty powerful technology here to enable those data driven, data focused conversations where you have the transparency, you have the recipe so you can see under the hood what it was doing. That's very powerful. What do you think? Yeah, absolutely, And I think it's the confluence of all these technologies that are coming together. We have the processing that was simply not available ten years ago to run any of these
modeling things in an economical fashion. And I know that some of these things need. We've got the data because we've been collecting it now for twenty years. Nobody throws up a data and that's awesome. And now we've got the business drive like every CEO has heard of LLLM. Every CEO is investing in that as one of the top priorities in the company. So the opportunity exists so super exciting time for all of us who are in this data world,
which is taking different forms databases, big data. Now everything is AI. But at the end of the day, data plus algorithms plus processing power that becomes available the synthesis of that is what we are living and it's fascinating. Well, and you know, we've got about one minute left here. The other interesting thing is I want more and more business users, including myself, to better understand the different algorithms in what they do, Like a K means,
for example, what exactly does that do? I think the more business people can wrap their heads around how those things operate, the easier they'll be able to map their own thoughts to their own world of data. What do you think? Real quick? Yeah, I agree. I think it's going back to your car example is that you don't need to be to know how to build an engine or fix an engine to understand a car to drive a
car, and that's totally fine. But if you're a race car driver, even though you're not fixed engines, you probably need to at des understand a little bit of how the engine works because guess what, you'll get a little bit more out of that now, a lot of automation is making that easier. So I think there's always value in learning something even though you're not going to be using that in a direct fashion, because it allows you to appreciate
at a deeper level. So fan of learning things absolutely, Yeah, well, folks, that's what we're doing. That was a great quote I heard from the CTO of Boom Media the day. I asked him, you know what's the most important thing in this era of AI, and he said learning and he meant human learning, Like we need to learn about these things and what they do and how they operate, and be very crystal clear about what
we're absorbing in terms of information and insight from these systems. And folks with a fantastic conversation with jig Nash Patel of data chatook them up data chat dot AI. We'll talk to you next time. You've been listening to Inside Analysis. The legacy Southern California's KCAA, the number one talk radio station in the Inlet Empire. KCAA Radio has openings for one hour talk shows. If you want to host a radio show, now is the time. Make kca your
flangship station. Our rates are affordable and our services are second to none. We broadcast to a population of five million people plus. We stream and podcast on all major online audio and video systems. If you've been thinking about broadcasting a weekly radio program on real radio plus the Internet, contact our CEO at two two eight one five nine nine ninety eight hundred two eight one five nine
nine ninety eight hundred. You can skype your show from your home to our Redlands, California studio, where our live producers and engineers are ready to work with you personally. A radio program on KCAA is the perfect work from home advocation in these stressful times. Just time kcaradio dot com into your browser to learn more about hosting a show on the best station in the nation, or call our CEO for details too. Eight one five nine nine ninety eight hundred.
Open for takeout and delivery, El Tapiac, Mexican food restaurant in the tri city center of Redlands is back. Their entire family is on hand to serve up their delicious burritos, but chaka terizo Waivosmanchero's steak and eggs just part
of their mouth watering great food. Since fifteen thirty one, people have marveled at the miracle of El Tapiac and now you can marvel at the great food the Lugo family has been serving up for over two decades, Nestled quietly in the corner of the Tri City Center shopping mall next to Burlington Code Factory. Support them. They can't wait to serve you some of their delectable, authentic South of the Border Mexican Fair at great prices, served up with love.
Support the area's best loved Mexican food restaurant in these tough times. Order up a tasty meal on the phone for delivery or takeout for breakfast, lunch, or dinner from ten am to six pm called nine to nine threeoh seven zero zero one seven. That's nine to nine three oh seven zero zero seventeen. Or google Altapiac Redlands and treat yourself to Hebo Tea Club's original pure power Drgo Super Tea helps build red corpusoles in the blood, which carry oxygen to organs
and cells. Our organs and cells need oxygen to regenerate themselves. The immune system needs oxygen to develop, and cancer dies in oxygen. So the tea is great for healthy people because it helps build the immune system and it can surely be miraculous for some inviting a potentially life threatening disease due to an infection, diabetes, or cancer. The tea is also organic and naturally caffeine free. A one pound package of tea is forty nine ninety five, which includes
shipping. To order, please visit to Hebotea Club dot com. T hebow is spelled T like tom a h ee b like boyoh. They continue with the word T and then the word club. The complete website is to Hebotea Club dot com or call us at eight one eight six one zero eight zero eight eight Monday through Saturday nine am to five pm California time. That's eight one eight six one zero eight zero eight eight to hebot Club dot com.
It's that time of year again, No, not the holidays. Medicare open enrollment and if you have questions about Medicare you should talk to the local experts. Paul Barrett and Associates. All of his agents are certified with plans that are accepted by most of the medical groups in our area. Call nine oh nine seven nine three oh three eight five. Their service is free and after forty two years of the business, their agents are trained to help you pick
the plan that's right for you. With sixty years of fascinating facts, this is the Man from Yesterday and back in time to this time in nineteen sixty eight, Merrily Rush and the Turnabouts will appear at Disneyland at New Year's Eve. They have a huge hit here in nineteen sixty eight with Angel of the Morning Just call Me jos times. And from this time in nineteen ninety one, actor Richard Gear, who is forty two, marries his longtime girlfriend Cindy
Crawford, who's a model. It's the first marriage for both. And the thing is is, I don't think it's that uncommon to one of you married for a year or two before all of a sudden you make another change and have a baby. And I mean, your life keeps changing and changing and changing. And from about this time in December of nineteen seventy eight, this Tuesday Night on ABC TV, The Carpenter's Karen and Richard in a Smith's Portrait
features Gene Kelly and Jimmy and Christy McNichol. But I and in my Dream, I'm Christmas with more at Man from Yesterday dot Com. I'm Rick Smith, host of the Rick Smith show, inviting you to listen to my show during the noon hour every weekday right here on KCIA. My show is sponsored locally by Teamsters nineteen thirty two, a strong union with fourteen thousand members in the IE. Our message is clear, unions improve the lives of working people.
You have a right to form and join a union, So go to Teamsters nineteen thirty two, dot org and get started now. America exists today as a bizarre anomaly. We profess to be an electoral democracy, yet we are ruled by a governmental plutocracy. One especially gross example of this incongruity is the overwhelming power of big money over the people's will by a wide margin. Americans of all political stripes want to ban the distorting force of huge electoral campaign
donations by favor seeking corporations and ultra rich elites. Yet nothing. National and state lawmakers take the plutocratic money and promptly bury democratic reforms that would stop the gusher of corrupt cash. An analysis of donations in this year's congressional races shows
that this plutocratic perversion of our politics and government has reached absurd levels. Historian Nancy Maclin and public interest advocate Frank Clemente have documented that the two main super PACs trying to put Republicans in Congress got about half of their one hundred and eighty eight million dollar election fund from just twenty seven billionaires. Also, a corporate front that backs gop CAN candidates, Club for Growth, got nearly thirty
five million dollars from only three billionaires. Thanks to years of congressional stonewalling and the steady partisan stacking of our top courts with corporate ideologus, there is essentially no limits on this purchase of lawmakers. Indeed, the biggest donors are even
allowed to undermine democracy in secret, not revealing their identity. Lawmakers elected with this money not only support the billionaire's special interests agenda, which the people don't want, but also are fierce opponents of any reforms to increase voter participation in
our governing process, which people do want. This is Jim Hitier saying that's why, for example, efforts to guarantee every eligible American a constitutional right to vote have not moved forward in Congress, even though the great majority of people favorite Lord of the Rooms and tire shop. Sam Bernino says it's hard being away from home during the holidays even arder. If you're soldier overseas securing our freedom, send lets and packages can save prayers for our troops overseas. Remember
they're fighting for our freedom. That's Lord of the Rims and Tire Shop at one point five East Baseline Street in Samborandido, serving the community with great customer service and low prices. Call it nine to nine eight to eight ninety seven seventy five. That's nine O nine eight A eight ninety seven seventy five. Lord of the Rims and Tire Shop is on the air because they care. Do you like to safely leverage bank money to earn double digit returns income tax
free, with guarantees and no downside market risk. How can you do this? This is Farence, host of the Your Personal Bank Show. One. You fund a high cash value policy one time to earn dividends and interest to establish a bank line of credit using the cash in your policy as collateral. When you earn more in dividends from your policy than the interest the bank charges, you keep the difference, and the difference is average two to five percent
annually in your favor for the past forty plus years. Three, the bank funds contributions years two to twenty plus. Each year the bank adds funds, your rate of return increases. Your average rate of return can grow to strong double digits annually within a few years. Contact meat your Personal Bank dot com, Your Personal Bank dot com or eight sixty six two six eight four four two two eight sixty six two six eight four four two two for more info,
or tune in to the Your Personal Bank Show. Your Personal Bank Show airs Tuesdays at four pm right here on CASEAA ten fifty am and one oh six point five m this station that leaves no listeners behind. For several years, KSEAA has been marketing the Youngevity brand of nutritional and personal care products. Our experience with Youngevity has been one hundred percent positive, so we are pleased to recommend them to you. Regarding nutritional supplements, we recommend pollen Burst in
the berry flavor and tangy tangerine two point zero in the tablet form. For regularity issues, we wreck amend three day cleanse, and for personal care we recommend morning hydration cream. You can shop online for Youngevity at www dot KCAA team dot com, or you can order by phone by calling eight hundred ninet eighty two three one ninety seven and tell Customer Support that you are part of
the KCAA team. Youngevity is an American company based in San Diego. Call Youngevity at eight hundred ninety eight two three one ninety seven and ask about monthly autoship that allows you to buy Youngevity products at wholesale prices. That number again eight hundred nine eight two three one nine seven. I always hear from our clients who hired another firm that they wish they'd hire DNA Financial. First, don't have regrets about your IRS tax case. Just hire the best in the
first place. One owed one hundred and fifty thousand to the IRS and it's spent thousands on another firm. We stopped the levis, negotiated a payment plan, and had their penalties forgiven. And while every case is different, we guarantee that we'll find your perfect resolution and get it done right. For a free consultation, call us at eight six six two zero one zero one. That's eight six six to zero one zero one five six. Then you can
say DNA from the Bureau of Economic Geology. This is Earth Day. When did our ancestors leave the trees and begin to lead a life on solid ground. Some real life CSI has given us a big clue. Lucy, the three million year old skeleton of one of our oldest known human relatives, was recently on a museum tour of the United States. During her visit, the University of Texas, scientists examined her skeleton with geological CT scanners, similar to
medical cat scans, but with higher resolution. Fossil bones often break as they're buried, but Lucy's upper arm showed something unusual. It was compressed with sharp fracture lines and tiny bone fragments intact. The team called in an orthopedic surgeon, who confirmed that this injury in modern human only occurs when they extend their arms to try to break a fall from considerable height. So they began to look for other fracture evidence of a fall and found it in her ankle,
knee, pelvis, and ribs. They then look to modern chimpanzees, which are about the same size as Lucy. Chimps nest in trees at heights of up to thirty five feet high, enough that a fall could result in the same type and degree of fracturing. Found in Lucy. Further studies using the ct data showed that her ratio of arm strength the leg strength was more like chimps than humans. Both finding suggests that three million years ago, our ancestors
may have lived a significant part of their lives in trees. I'm Scott Tinker and this Spaniard date with Earth. Earth Date is produced by the Bureau of Economic Geology at the University of Texas at Austin. Earth Date is researched by Julie Hennings, written by Harry Lynch, and distributed by Mark Blunt and Casey Walker Stories. Follow us on Facebook or visit Earthdate dot oorg. Are you tired of the same old conversations that everyone keeps talking about? Want to hear
something new, fun and exciting. Listen to Whatever Works here on KCAA every Sunday at seven pm. We have amazing guests with interesting and unique topics. We do giveaways, and we have sponsorship opportunities. It's more than just a show, it's a lifestyle. You can also find us on KCA radio dot com or our website KCA Whatever Works dot Com to find upcoming and previous episodes.
We like to thank all the KCAA listeners for supporting our show. Did you know here at KCAA ten fifty am that we developed an app for all your Android devices. We're talking about your smartphone, your tablets, you name it. You have an Android format. You can take KCAA with you everywhere you go. We're talking about our audio stream, our video stream, and even our podcast. Go to KCAA express dot com. That's KSECAA express dot
com. KCAA express dot com. Listina KCAA Lomolinda at one O six point five FM, K two ninety three CF, Brino Valley, NBC News Radio. I'm Chris Gragio. Former President Trump now says he won't testify in his fraud trial in New York tomorrow. He posted his decision on truth Social today, adding that he's already previously testified. Trump was set to make his second appearance on the witness stand to be questioned by his own attorney, says the
final witness for the defense. The office of New York Attorney General Letitia James, filed the two hundred and fifty million dollar lawsuit against the Trump family and the Trump Organization for allegedly inflating financial statements by billions of dollars in an effort to receive more favorable loans. Secretary of State Anthony Blincoln believes Israel does not want to harm Palestinian civilians, but also should do more to avoid civilian casualties
and allow humanitarian aid. Speaking on CNN stead to the Union, Blincoln said Israel needs to close the gap between intent and results.