Speaker Jeffries said on NBC's Meet the Press that they have had informal conversations about a bipartisan governing coalition. He said the House needs to vote on bills that have substantial support from both sides of the aisle. Jeffries blamed a small group of Republican extremists for trying to dictate the agenda. I'm Chris Caragio, NBC News Radio, kc AA Radio, Lo Melinda, Where no listener is ever left behind. The information economy has a rid. 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 Comsideanalysis dot com. And now here's your host, Eric Kavanaugh. Ladies and gentlemen, Hello, and welcome back once again to the only coast to coast radio show in the US of A that's all about the information economy. Inside Analysis, your host Eric Kavanaugh here,
and folks. I am tickled pink to be talking to a legend in our industry, mister Andreas weigand he has been in this business for thirty odd years. He has an incredible history in our industry. He advised Jack Ma and Angela Merkel, their data coach. He worked with Jeff Bezos as chief scientist over at Amazon, and he got his PhD some well, he started it some thirty years ago. He knows a couple of things about artificial intelligence AI, its applications, how to get it right, how maybe we won't get
it right. So we're so pleased to learn from him about what is going on in this crazy world of ours. And you also want to book data for the people. But let's talk about AI for the people and what this stuff really is. And I'll just give you some of my thoughts. I mean, what I find fascinating about these new foundational models and the transformers is that what we're really witnessing here is a reflection of our culture fed through a
very high compute power engine. So these CHATGPT barred technologies, they're basically predictive engines for language. And what the folks who built these things did, to
my understanding, is they really vectorized information language. So letters, words, sentences, paragraphs, headings, et cetera, all vectorized, in other words, turned into mathematical quotients, if you will, and that allows for traditional statistical analysis, and that is the engine behind being able to predict language as you would predict in other use cases whether or not someone's going to buy something.
So these are predictive engines. They do reflect our culture and there are some shortcomings, and the shortcomings I think could be solved with embeddings and with some guard rails, but we'll find out from the expert. So Andreas, thank you so much for your time today. Tell us about your take on AI and how can we ensure that AI becomes AI for the people already thank
you for having me. So let's reflect over the last year, in the minds of the people, AI has greatly taken more importance than it was the case prior to that. Why is that the case? I think the are two reasons. One, the amount of computation, the amount of data has been grown exponentially for the last thirty years or something like that, so there is no clear moment where there was like a face transition or something like that.
But it is the perception of the people that something amazing has happened with GPTs three for instance. And what is that? Why is that the case. I think it's the case because suddenly we have a way where the computer interacts with us through language, through our normal language, and everybody on the planet can get a few thing about how far things some path come along.
So for me, the sudden popularity of GPT three or GPT four now is because journalists realized, Wow, this is amazing, and then quickly the world realized just how amazing it is. And I have to confess I've spent many hours of my life in an earlier version Sam Altman had given me access to maybe two years ago, and I had promised not to talk to anybody about the results, so I still want do that now. But then, of course with tatt and I have to also say, I'm amazed that it knows
more than any friend I have now lendth amount go ahead one. Google also knows more than any friend I have. So we shouldn't be that surprised because we have seen this in the late nineties that while you suddenly have a box that you enter something in, and no single friend of yours can give as many answers as cool who were good at the time. So we have had this and we forgot it. We got so used to it. That's number one. Number two is, however, that it has much more capability then
just returning snippets of web pages what googles and studus. And for example, I'm doing a workshop in a week, and I wrote an abstract what I wanted to do in those three hours, and then I thought, I run it through GPT whatever the current version was, and writing the abstract probably was an hour. Carefully comparing sentence by sentence what I wrote with what GPT suggested
took probably two hours because it really this engagement. This you know, I really engaged with those alternatives as if a friend has suggested, oh, andreez, you should make me write it this way. So I maybe reverted to my own in eighty or ninety percent of the changes, but that is just because I have my style of writing, and GPTs it's most that is the
right way of writing. So I think at the end of the day it was definitely not worth the time for me, but it was another interesting experiment and maybe I should take data for the people my book and put it into GPT and see what's coming out. Particularly so it's coming out in a few languages, in German, Korean, Japanese, Russia quite good. The Chinese translation is very bad because the publisher took one page for a person or one
person page for translating it, so it's extremely inconsistent. It's basically non comprehensive Chinese. So maybe maybe I should just use GVT and produce an English translation of it. It's at least it would be consistent, right anyways, So that is my impression. It is absolutely shocking. It remains shocking even for me. And the question that is coming now is what will be the roll of people in a world that we have GPT as much around as we have
running water coming out of faces? Right right? This is the big question. And it seems to me that this is another technology. I mean, you look around the world. You don't see too many farmers using yaks to till the soil because we now have big machines to do that. And of course the world of agriculture around this planet has become heavily industrialized, and that's
because big machines get the job done faster and more efficiently. And I've always believed the machine can do the job better and faster and more efficiently than the human. Let the machine do the job as long as it's economically feasible. But there is something to be said for boundaries. So a good friend of mine, Eugene Burke, came up with this great concept. He says that right now, which at GIPTA doesn't have an epistemological boundary, in other words,
it doesn't know what it doesn't know. And I think the magic here is going to come in the embeddings and and you're let's call it first class citizen embeddings. What do you want to do from a grounding perspective? And all the folks I talk to are saying the same thing, that they are focusing on particular industry verticals because that way they can get to a working model pretty quickly, as opposed to using a large language model as a general purpose
tool and then trying to train it on your corporate data. That's going to take probably a long time and not get you where you need to go. But think about, just think about this idea. For example, think about the history of media. And I'm a media guy, okay, so I understand how media works, and it's largely been a push model for decades,
for centuries. Quite frankly, it turned into a pull model for kind of a short period of time with Google, and we had all these bloggers who kind of flourished, and then that there was sort of a race to the bottom with content around things like big data and AI, and so it gets harder and harder to get to really good content unless you know where to look,
and that's hard to do. But what I think is interesting is that most media companies take a very let's call it cherry picking approach to coverage. Someone decides, I would say, sometimes arbitrarily, Oh, let's focus on this story. Let's focus on that story. That's the old way of doing media. I think the new way is going to leverage AI both for generation and consumption and what I want to do. And you might like this because
you talked about data for the people and AI for the people. Right now, AI largely guides or governs what you see in places like Facebook and LinkedIn and Instagram and all these different tools, but the model is behind the scenes, so you don't get to see the model. And I and some colleagues of Mind are working on something where I want to make that model transparent. I want you, Andreaswigan, to own your feed and to understand your own
model. I want to make that model transparent to you so you can adjust it and fine tune it, and then create this new conduit to information around the world that will be dynamically generated by systems of record. So think like the Federal Register here in the United States, that is a system of record. When something goes out there, that means something changed and people want to
know about that. And I think people are going to be able to subscribe to dynamic newsletters where these AI engines will capture and process and deliver highly targeted, highly personalized information for me based upon the things I'm trying to research on, and I think that is going to be tremendously disruptive to the traditional media models. And you talked about fake news. There's that, and there's also
just biased news. So I think run the beginning of an absolutely massive transformation in how people research information, how information is generated dynamically from like I say, systems of record, and then how it's channeled to us. And I am all in favor of putting individual people all around the world in control of their feed to be able to see what they want to see, especially in the government like government spending, where does all the money actually go, who
gets it? What do we get for that? All that kind of fun stuff I think is coming. What are your thoughts about all that? Do you want to narrow it down and go back to the interview? Yes, okay, sorry I threw a lot at you. I know. Yeah, so let's look at I think the role of people. I think what's very important is that you learn to ask good questions. So curiosity is more important
than it has ever been. I used to say that, you know, years ago with Schogle, but I'm saving it even with more emphasis now with Tragical Team. Second one is the idea of modeling. So you talked about modeling, and what's the essence of modeling. It's two things. One that you have an objective function which you're trying to optimize, and two that you are very clear what's within them model and what's outside the model, which that
decision we call the error right. And I can, for example, quote a Ted talk I heard many years ago in Long Beach where a person in the first half of your talk make the point that you really really should use paper bags as opposed to plastic bags. And then you know, you see the images of the fish in the ocean and we've all seen sure. And then exactly halfway your first talk, they switched over and they told you why you should show use plastic bags rather than paperbags, and the images were that
of deforestation. And okay, So it is the question about what are the externalities, what do we consider outside the model, and what do we consider inside the model. So those two remarks I think I would like to make what you said, one curiosity is king, and two understanding the objective function of Spitus said, understanding the boundaries of the models in the model and what's not in the model is key. Yeah, those are very very good points.
And so let's dive into transparency. We've got about two minutes left here. I'm a huge fan of transparency, and certainly for anything government related, because it's our government, and so why shouldn't we know what our government is up to, because I think they wind up going down wormholes for various reasons. But transparency in data. Of course, in GDPR, someone can request an EU, citizens can say, hey, show me all the data you
have about me. Years ago that would have been basically impossible. It's not impossible today. It can actually be effected today, which I think is a fantastic balance. And I do believe that you look at how all these big companies are training their models on our data through Facebook and Google and Twitter x whatever you want to call it, and I think that we should also have public facing, public accessible reposit stories of information about transactions, the kinds of
things that SAP and Amazon or others can learn by aggregating this data. I want a public service where citizens and businesses can subscribe to that and use that data to better understand which products they should buy, et cetera. But what do you think about all that and what do you think about transparency in particular
on dreams? Yeah, transparency. In my book, I talk about transparency rules right to see the data, to see what's happening to your data, and to have the right to see the reward ratio between what you and out of the data you give. So transparent is a good thing, but very difficult to implement. I think a more important question is to just talk about transparency, but really to talk about what can we do with it? The action I mean, what is the value of data? The value of data
is only the action difference different actions make based on the data. So I like to focus more on what do pee to do differently and how can we educate people so data literacy, is there any chance left? It's a very honest question for people to know whether it is fake image of somebody that you see that was created by image generator, or whether it's a trimp. So those things we took for granted all over our lives, we cannot take for
granted anymore, and it would be a very interesting world. What is next now? I'm a big fan of the Center for Humane Technology and their fears of what can go wrong and what can we do to try to stir things in the right direction. I don't subscribe as many of my colleagues do to the Moratorium, and I because having lived in China, I had a house to Shai, my husband is from China. I am of no illusion that even if in the US we you know, do everything to stop you know,
any research China, and if not China then Russia. Will we just laugh all the way to the bank to have the ad ah. But the thoughts the Center for Human Technology, what they have, it's basically education. It is having people learn what can go wrong, and successfully we managed to deal with this with nuclear wars. We haven't had nuclear war since World War Two. Thank god. We lived in a very good, very lucky period in the last fifty six years. No period in the history of universe has
been as good as the last fifty six years. And we can only whatever we say is probably not strong enough that we should take more time to think about it, to think about the actions we take, and to play with it, and to get used to that exponentially growing world and to the question about how can we ensure And I don't have an answer, but how can we shure that this is not an AI against the people, but that it
would be an AI for the people? Mm hm wow. What a great bit of wisdom there from Andreazwevegan. Thank you so much for your time today, sir, it's an honor to be talking to you. We'll be right back. You're listening to Inside Analysis to welcome back to Inside Analysis. Here's your host, Eric Tabanac. All right, ladies and gentlemen, welcome back
to Inside Analysis. I'm pleased to be sitting at Boomy World in Silicon Valley now talking to Ken Maglio of Worldwide Technology, and they're a big Boomy customer. In fact, you use this from what I can understand as a core component of your solution architectures. Right, tell us a bit about it. How do you use it? How has it changed your day to day life? Yeah, so worldwide technology. We utilize BOOMY to integrate our data with
our customers. So when we are integrating with them for all of our ticketing needs and providing managed services to our customers, we need to be able to bond our systems with theirs. Okay, so effectively we're working with one set of data, one Golden record as a BOOMY term as between both our systems.
So that's where we utilize BOOMY today to integrate those systems together so that we're all operating on the same ticket, the same data, the same comments, the same work being done, regardless of we're a separate company from our customers, right, we're doing managed services for them. We also utilize it in automation self service, you know, for onboarding purposes, reaching into OEM and other vendors where maybe in integration or alerting or another system can't produce something,
but we can if we get the wrong data. So we utilize BOOMY to do a lot of extractions and then also like I said, this e bonding process that we've built, that's interesting. So it's really a core composent of your solution delivery strategy, right, yes, yes, it is. It is core components included in every offering that we give out with our managed services to our customers because it's so core to how we work with them as a trusted advisor. Yeah, that's very interesting, and so you can cover
any number of solutions with this. You talked about networking, you talked about operations, efficiency gains, things of this nature. So any number of the things that an IT consulting firm would come in and do for a company, that's what you're doing. And a component sort of backbone of that is the
Boomy platform, right yep. So we build a system within Boomy to be able to synchronize all of that data between ours systems and instead of trying to build individual integrations just between one customer and us and then the next customer repeat down the road, we built an entire engine or factory if you want we call it, to be able to allow us to repeat this and almost have a product that we can offer to customers. Say okay, as a handshake,
we'll agree that we're sending this same data between us. But now this same process is repeatable. And so now we used to initially start our integrations when we first did our first one between just one technology and another without boomy, and it was six months and quite expensive of an effort. I'm now measuring what the team can do in about forty man hours of effort, not duration, but about I've taken that from what used to be six months down
to about forty hours. Wow. So what we're trying to do is provide that efficiency. Now, obviously there's some nuances, but that's the biggest gain for what we're looking at now is we can actually move at the speed of our customers. And this isn't a six month long project. After we're go alive. This is a great we're live here a month or two after we're
up and running with them. Yeah. No, what you've done is you've optimized the process by building an engine that is versatile enough to handle lots of different kinds of projects such that you don't have to go and build out bespoke solutions for every different customer because to your point, that takes a lot of
time. And there's all I mean in our industry, we've talked about the sort of reuse, reduced recycle, like remember the old de mantra of environmentalism, and you're practicing what you preach, right, Yeah, And that's our big thing is I didn't want to build a system, you know, and then the team to have that overhead and management of Well, now once we get past five or ten, we've got a rat's nest of integrations and we've
got problems because this integration fired. One, it didn't realize this integration but you know, circular logic, all these types of things. So instead we needed a system that was repeatable. Two, we needed to to scale because I don't know when the end of our customers are going to you know, be But there's another aspect that we didn't plan for that when we built the engine, we all of a sudden just said, well, if we tweak a things, the engine could handle it. And that was we started to
engage partners, and our partners wanted us to bond with their systems. So now I have the scenario where it's us, our system and a customer system. But now that customer can have one or five partners, and I have to know traffic cop where that ticket goes to, which partner, who has the ball in through all of the system, and we were able to build that simply using a lot of the out of the box features of Boomy's Master
Data Hub, and we didn't have to write that code. Now we had to build this integration, this factory to make it all work and tie it all together. But a lot of what was created was some of the core way that Boomy works. And I don't know if what we've done is completely unique, but when we've talked to others like HM, we didn't think of
having Mastered Data Hub as that type of a use case interesting. So it's been a very very interesting story to be able to tell to other companies and even even Boomy themselves to go that's kind of a unique use case that isn't always the way that it's used. So well, it's funny. I've found that one of the most interesting parts of the software world when you develop an enterprise application, when you some years later talk to your clients, you will
find them doing the most creative things that you never thought of. Yes, and that's why you have user conferences, that's why you talk to people, that's why you have strategic engagements in partners, right, is to learn Wow, so you did that as a sort of reset a stone of integration essentially, right, right, And that's some of the nice things about these conferences is sharing that feedback as well, right, and being able to sit down and say, hey, by the way, there's this other the other scenario
that you might not think of, and they're like, oh, that is a good idea. Let's we need to get on that, you know, type type of scenario. Or talking to other individuals where what you've solved they need and then vice versa, like oh, you did it that way, that's an interesting way to do that, right, and or even the validation of we all did it the same way because we all got to the same
you know, we all came to the same conclusion. So that's that's been a great thing about these conferences and being able to connect with booming as open as they are. Yeah. Well, and you know, every company is unique, every situation is unique, but still there are broader patterns that are
similar across different environments. And what you figured out is that you can carve out x number of patterns that will get you eighty percent of the way there with new solutions such that, yes, you do have to finish the twenty percent, but you're eighty percent there from the get go, and that's kind of chat gipt string right gause eighty percent of the box then you find too and that's exactly the right when we when we sat down the team and I
built this, it was we knew it's not going to cover one hundred percent of the scenario. We know we have to continue to modify and make a little bit unique each customer. What we didn't want was every time we sat down it to be a snowflake every time. So we needed to start with more of a product that we can offer that has the levers and the knobs that we can turn, or the availability that well, because we're in boomy, we have even more power than that. You know, we can swap
something out, but at least we're starting from a template. Plus, now that we've built this system, certain things like air handling or data validation is universal throughout all of this, and we don't have to go re implement its boiler plate. It's included from the get go, so we're able to track and do things that if we had to build this that would be something you'd
have to build over and over again potentially. So we've been able to gain efficiencies and be able to really build I believe in a very good robust system within Boomy, a system of systems within Boomy to be able to allow us to really integrate and hand a lot of different use cases of our customers. Because one system service now very prevalent in the industry, everybody implements service now
differently. There's not one company that just uses it straight out of the box, and so when you come into that, it's not necessarily the technical piece that's the hard part. It's the business processes and what they do and what they've configured above and beyond the out of box that, well, now we
have to bond with that. How do we make something that we don't have or our processes different work with their process So through mapping and other things we've created, we've been able to mitigate some of that, to be able to change the data as it's coming in into our global Golden record universal model, so that all of their uniqueness gets transformed into one standard model, which happens
to be ours. But and vice versa. If we had something very unique that we did as an MSP as a manager provider to make it work, or we would handle that in our integration back to the hub, back to that record, so that Golden Hub, that record becomes the universal standard of what represents our data in our different models across the board. It's one record for everyone. So it's an interesting challenge and working with customers to understand it's
not separate copies and you have your different cop we have our. It's one thing. That's we're now operating on one playing field. That's very interesting. Well, and as a consultancy, I'm sure you know this. I don't have to tell you that you need to deliver value to your clients on a regular basis and there cannot be big long pauses and value and months of time going by where people are scratching their heads and getting her it's like, all
right, what are these guys doing for us again? So you've kind of solved that by creating this engine, this multifaceted engine, and every time you learn something, you bake that into the stack. Right, Yes, And that's something we're actually on Version two just rewrote it this year, Version two of our e bonding engine that we've written because we did some organizational and system changes that we wanted to use Boomy as the center point for everything. Now.
We started as doing a traditional integration system to system and so that has existed for a long time that just got replaced with this new version two oh and in doing that, we also learned how to mature our own model.
Like you said, we learned from one point zero, and now that we have that, we have all this from two point zero, and we've even gone beyond that to really start utilizing a lot of Boomy features that we were touching on in the first one, because the first time around you're a little probably gun shy, so but this time around we've just jumped in with two
feet and we're already seeing the benefits from that. Because, for example, we always have to with a ticket tell what device, right, if we have an issue with an alert happened and we created a or an incident, we have to pass that along to you in the ticket itself. In the case, well, originally we were just passing the name of the device back in the day in the version one point zero. Well, with two point
zero, we now have an entire relationship. We have that actual device as an entity so eventually, if we want to make two point zero have a device synchronization with our customer, it's native. It's there. We just have to build the endpoint to push the data to the customer. But it's in our system already, so we have features like that, or being able to respond to data that's being changed. Hey, it went from a priority two to a priority one. Well, Boomy can see that now it couldn't before.
Well, now that I can see that, I can actually reach out to someplace, like if I want to hook it to Twilio and text people that, oh there's a critical alert text to get everybody on the horn. We can now do that because all of our data is now centralized and boomy. So we're getting a ton of benefit. Interesting. So one of the on our show years ago joke to actually the pre show that in the IT world, we're always saying that one more layer of abstraction can solve all our
problems. Right, But that's kind of what you've done, right. You've created a layer of abstraction. You're using Boomy as the technology that runs this engine, but it is an engine that you have built on top of that solution. So it is this it's a deep layer of abstraction, right, it's our as I always call it, the system of systems, right, But that's that's one of the things that I try to do and then work with a team to make sure they understand it. And like I said,
this huge team effort to get this done. Even though we're a small team, we're very mighty in getting this work done. So's we're doing a lot of things, even pulling some stuff out of boomy. And because we're trying to build this repeatability of scale, we actually pulled some of it out into sequel and tabilized and data like data drive driven drove the actual integration itself. So now when we templatize them, we could start from our stand well that
already has those knobs and levers. We can do but it's not a redeployment. It's just changed some data values and we're off to the races. So we're doing things like that at a basic level to really kind of drive what we do so that there's less overhead of It's just like a deploying code I
wouldn't want to change. If I had a business rule engine, I wouldn't want to redeploy that whole engine and the code that runs it, just because my business rule change now I want to have that it driven by data and a table or something. We've taken that same approach with high integration. Yeah, we got about a minute and a half left here. The pace of change has accelerated so much in the past really ten months, I would say, when chetchipt came out of these other things. And I think some people
are getting a bit over their skis. But what's your advice to clients about how to mitigate the potential risk of change and how to just stay on track and make sure you just maintain com and keep going. Yeah. So it's a great challenge because one of the things is you have to keep going with change. And that's the one constant is there will be changed. So building whatever you take and utilize with any of your system boomy, et cetera.
Is being able to handle that change in a way that you're not painting yourself into a corner. Because it will change. It will come down the pipe, right, you will have to make that process or you will have to make that integration do something different. So if you can take the time to I'm not saying make it perfect because that's analysis paralysis. But if you can take the time to build yourself just a little bit of room. You came
up with the perfect term. Don't paint yourself into a corner. It's very dangerous these days. You have to watch out for that because it'll be very painful if that happens. Oh, you're gonna be a very unhappy camper. Yes, probably looking for a new job or a new client is something. Yes, well, look this gentleman up online, folks. Ken Maglio from Worldwide Technology can congratulations. Thank you, sir. We'll be right back.
You're listening to Inside Analysis. Welcome back to Inside Analysis. Here's your host, Eric Tabanat all right, folks, back here on Inside Analysis, here in Boomy World. And I'm very excited to be with a Jay Natarajen. He is with United Techno and you are an enterprise architect. That's good. And we were just talking a moment ago about how cloud is really a new center of gravity, which is quite fascinating. But personally, I believe that
on prem data centers are not going to go away anytime soon. We're going to live in a hybrid world probably forever. That's just my personal theory. But how does Boomy play into the modern enterprise architects world? Yeah, thanks, thanks for the opportunity. The CIO of one of our major retail customers just said today to me that Boomy is the center of his universe. It
is the most impart and application in his enterprise application portfolio. If you do all ten years behind, the most important application in enterprise portfolio was the ERP. Now the ARP is getting smaller, the monolith is getting broken. It's become the best of breed SaaS application composable. And they're all desperate, they're all spread spread and they can be changed very quickly. What and you need a smart, intelligent platform to make sure all of this is connected and everything
is smooth, is growing smoothly and operationally ready for our customers. The operational importance is extremely important for our customers. The ability to connect these desperate systems and to change them right. If they don't like one, they can change the other to a strategy to be able to change them is extremely important to
our customer and Boomy provides that platform. Yeah, what's a beating heart now of the modern de facto enterprise architecture which is hybrid but cloud first, cloud native ideally, but reaching into all these endpoints, whether that's the traditional data center, whether that's the edge, whatever the case may be. You have this beating heart of intelligent automation and integration. And as you're discussing, that's
booming. Yep, that's true. Yeah, And what we do and boomy is a great platform for this, and it's not just an integration platform now and now it's become an intelligent integration platform with a new AI capability. Is all that is going to do is to make this more accessible, make this more easily deployable. It's so much more easily consumable. They have a good events stream event streams product that was just released makes the connectivity and scalability of
this platform a lot more usable for our customers. And that's why Big Enterprise wants. They want something that is obviously easy to use and quick to use, but they want something that can scale, that can moderate well, that can take not just like today's traffic, but three years down what the traffic is going to be that then they should be able to do that well. So you kind of alluded to something here, which is the rate of change in the world at large, right now, which is off the charts.
I mean, it's just crazy how fast things are changing. And if you're a CIO, you want to be as agile as possible, and so you want to be able to leverage new technologies. You want to be able to leverage new processes. You want to be able to bake insights into your existing processes, and you don't want a lot of lead time. When you come up with the idea to get something done, you want to go, Aha,
let's do this, and you get it done. And that's what this marshaling area for integration and automation and complex have been processing boils down to. Is your engine your de facto business integration engine? Is that about right? Yeah? That's good. I mean what I keep saying, I have a bl B that I'd like to plug in linked and I blog every two weeks. And it's not about what we do, it's about why we do,
why we do things. Why I don't wake up at two in the morning to fix something because we catch something ahead of time using develops and c s in automation. One of the blogs I say, integrate, don't complicate, and it's all about how do you simplify the process in paper and make sure that there's skill visibility between what you're trying to do and the end goal and
continue to iterate and build on that. Right now, we see Boomy's platform is a great platform for us to do that, right, whether it's simply simplified UI driven approach and the ability to connect different systems all in a single
screen gives us the ability to do that. You need, well, no, that's okay, So you need you need a command and control center basically, and that's kind of what Boomy is becoming to a certain extent, because again, you have the cloud, you have on POM data centers, you have all these edge places, and you have all these different ways of doing things. New startups are coming every single day for just about every kind of
functionality you could want. And so if you have a stable foundation for intelligent integration and automation, that becomes this this de facto center of your environment. Right, that's good. I mean the platform is you know, gives it has the ability for us to see everything together in a single place. Right. Hybrid integration is the norm. It's not it's not like an exception. It's the rule. Right, It's the rule. There's always going to be
on premise systems. There's always going to be cloud systems. There's going to be hybrid with the clouds. I have data sitting in a zoo, I have data sitting in a w S and GCP that I need to connect and those that's where the systems are and the ability for me to connect everything, see everything in a single platform. The command and control, as you said, it's pretty good, right, and that's exactly what the platform provides.
But it also provides the with control, which is important. And there is fine grain access control so that even though everybody can you can see something, only people who are allowed to see can see it. There is tight security and authentication. There's type security and what you can do and what you and everything else and you know you cannot. I think that's that's also very critical. And for most of the platform you're talking about, especially integrate, that
data is on PREND or not on PREND. The data is with the customer. It doesn't kind of leave wooly in most sense, it's sitting on your file shaft systems and it's it's displayed and rendered that so that's also pretty important for some of our customers. So Booby think about it like it's a platform wh's ubiqulous that anybody can see, whether you're a business executive, an I executive, a developer, an operation person, you can see and integrate and
play the same platform across the board. But with type security and controls and access, you can make sure the only the right people will have access to that data. And that's critical and that's why many of our customers choose Boomy. No, that makes a lot of sense. What do you see with AI and AI being infused in the system. Where do you see the low hanging fruit of how that can be applied and how it can change day to
day operations. That's a great question, right. So Tom Davenport was a keynote speaker today and he said, the only people who lose with AI are people who don't work with AI, and that's true for platforms and companies. What Boomy is doing with AI is required in this day. It is required, and it's also a great stepping stone because they've been doing this for the
last ten years. They have two and twenty terra by building gazillion data that's what Ed mentioned, and they've been doing this secretively, honestly they've been doing this any so, they've been doing this for the last ten fifteen years. So when this chat GPT was released, it was a youtubeling. Uh they are they leverage what they already had and just made it available for us. I think some of the features are gad today, which is pretty good.
So we are seeing the evolution of it. There will be iterations where it's it is going to get better today with a single With a single prompt, I can generate an integration from CRM to it's crazy crazy Yeah, but that's still step zero, step one right, Yeah, that still requires some work to make sure it's productionalized and deployable. What I would love to see is you know that being production ready code and that code being generated based on what
I've previously deployed. AI should help And I see in the roadmap that at some point of time AI will help with operational issues. When can I schedule these deployments so I do not impact an existing system? Right? With a loader in today in our customers, we do about twenty two million documents a week. Wow, how do I know I should be doing twenty one or twenty four? Am I missing a million? If am I getting one million
too much. I don't know, to be honest, And we are reactive in this place, and the whole industry is more reactive in this This is an area of an AI can shine and help HI, which is human intelligence, to do a much better job. I've always believed that if a machine can do a job faster than a human and more efficiently, let the machine do the job. You don't see farmers using yaks to till their land anymore. They have combines and big machinery to get that done. And that's the
kind of the world we're in the right now. Yeah, yeah, yeah, I mean I drive a Tesla. I don't knowssemble wheels, but you're right. I mean, AI is a friend that we can employ with care and love. It's a mutual respect and love. I mean, we kind of use it, and we've got to make sure that there's governance and control on the AI itself, but we can. We can do a lot more. Today. I'm able to write in languages and never learned. Right, that's right, that's right. The generative AI. You know, I was
afraid of using new languages or even using some complex things in Python. But I can ask AI to make it better with with privacy and everything considered, and I think I'm getting slightly better in my job and and that's that's that's always helpful. Well, that's wonderful. We've been talking to a Jing not orogen from United a techno. Thank you so much for your time. We'll be right back. You're listening to Inside Analysis. Welcome back to Inside Analysis.
Here's your host, Eric Tavanaugh. Okay, folks back here on Inside Analysis times for the podcast bonus segment. And I'm pleased they have Damien Black with me. He was the founder and CTO and I guess for a while CEO of Sequel Stream, and he's going to tell us a bit about what they're really trying to do with these large language models and kind of how they work. So Damien, take it away. Yeah, just I can just sort of demystify this. And so first of all, the first sort of
big breakthrough is the transformer architecture. And what really the important part about that is before they would what they're doing is they're looking at in neural nets. You're basically trying to predict the next say word they say token part of a word. But really you can think of it as predicting the next word, and you do it from what's been before, and so you can then train
neural nets. You can change train neural nets based on for example, Wikipedia go through and I you'll be trying to do each time is train it to what is the next word. The problem with that is you know, you have a whole context of what's gone before and then the next word that's coming up, and it's a sort of very inefficient way of having to retrain. When you're retraining, what effectively doing is you're kind of solving thousands and thousands
of equations, and that's what's really going on here. I can tell you a little about that what the neural nets are, but that effectively is what is really going on or and it's not actually solving, but trying to get better solutions or better approximations to the solution. So what the transformer architecture instead does It allows you to have a number of words predicting the next token,
so you have a buffer of them all at the same time. And so there's a benefit in that because you can actually parallel You're not just sort of moving on one with one word at a time. You're actually doing one word in the context of a sentence of words. Now, part of the challenge in trying to actually get the sort of the full semantic meaning from that is
there are a couple of things that you have a problem. One immediate problem is then is that you actually have a kind of look ahead capability, because when you've got up you know, fifty words say that are being used when you're training against the next word that's coming, it's got all of the past history these fifty words, and then predicting the next word. The system when it's solving the equations can look ahead and you can actually be looking ahead at
words that are yet to come. Right, that's a problem. So they have a mechanism to try and stop that. Add in some masks so that the words we even when you're solving the problem any in trying to predict the next word, you're not allowed to look forward. Then the model can't actually see can't take any waitings from the words that are yet to come. So that's one thing that it does to try and stop. Otherwise it just you know, it's a bit like you know, playing cards and you know what
the next card is. It's going to be delta on the top of the deck. It kind of ruins all the predictive models. So that's cool. That's a call of masking. The attention behavior is basically provides a number of different extra waitings where words can refer to other words or take waitings from other words that have happened in the same area in that same buffer, So you
allow and that's really important because of the context of a word. When you for example, particularly using things like pronouns and so on, you don't necessarily sure what the thing is that you're actually referring to. So the attention mechanism is to allow in the training algorithm waitings to come in for words based on the other words that are present in that buffer. And so the you could
it can when it's solving the equations and changing the waitings. It has a number of different ways of changing parameters and for each of the sort of the attention references or like the attention head, it can actually tweak waitings to try and come up with the best possible solution. And just adding in that extra capability where there are waitings based on references from basically one from the position and one word to another word, it allows the it allows the extra expressiveness that's
required to be able to capture spoken language or natural language. I don't know if I put that very well. But the way to think about this, if you understand what what a you're on net is, it's basically, you know, it's it's modeling information as equations, so you can write, you know, it can have ax one plus b x two plus b x three plus you know, c x three plus d x four and so on, and then it's all about how you actually what are the waitings you're going to
put onto these different waiting factors on how many variables you have. Then they then those equations then roll into they feed into another variable at the next level of equation. These equations all basically slot in and you what they're trying to do is is have a series of equations that is trying to model what's going
on in the real world. If you had enough of these equations, you know, rather than if you're just you should be able to measure, you know, model any kind of shape, and as you have enough dimensions, you can model anything. That's the idea and the way it's so what they're trying to do is change change through these complex equations that roll into one another, which is what the neural nets are, and the waitings are finding the
waitings on those variables. What they do then they then say, they write a function which is describing how close you know the function which describes those equations compared to the ultimate answer that you want. The difference between them is the error function. So by taking the separation of those you have in another etceter equations that describes which describes basically how accurate your model is for the natural language.
So if you have your equations describing how you represent the understanding of English, and you've got your predictions, you can see then okay, what is the difference Because you've modeled it all with mathematic equations. Now you think, well, how, how on earth what can I do about that? Well, what they do is they use simple calculus. It's basically how the actual algorithms are working to train the network. They're trying to basically reduce their weightings
to reduce the error function. So it's all it's all, you know, It's all about solving very large numbers, literally thousands and thousands of equations, thousands and thousands of parameters, billions of possibilities, and that's what's going on with these systems. Well listen, thank you so much for your time today and great stuff. Will talk to you next time. Folks, you've been listening to Inside Analysis, Southern California's mind spread progressive talk in Southern California,
the legends you love and the best talk open for takeout and delivery. Hell 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, machaka charizo, wavos, Ranchero'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 logo 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 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 take out for breakfast, lunch, or dinner from ten am to six pm called nine to nine threeoh seven zero zero one seven. That's nine O 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
corpusols in the blood, which carry oxygen to organs and cells. Our organs and cells need oxygen to regenerate themseelves. 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 t Hebotea Club dot com. T EBO is spelled T like tom, A, H E E B like Boyoh. They continue with the word tea 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 with sixty
years of fascinating facts. This is the man from yesterday and from this time. In October of nineteen fifty five, the new Mickey Mouse Club over ABC TV is going to add a filmed serial series titled Spin and Marty and their Adventures on the Triple R Ranch. Spinning Marty was an unbelievably popular thing in our time, but we were the only thing on that was for kids. You know. There were pampa shows, there were a cartoons, but there
weren't any live action shows. And from this time. In nineteen eighty eight, Leondard nied Way of Star Trek Fame directs a new Disney movie called Three Men and a Baby. Two of its stars are known to TV, Tom Selleck and Ted Danson. It also stars Steve Gutenberg. It's a baby. It's a baby. Of course, it's a baby. It's you a baby. No, it's not my baby. It's Jack's baby. If your turn to changer. I'll give you a thousand dollars if you'll do it, And
from this time. In October of nineteen sixty seven, at Sullivan signs the Calcils to an exclusive TV contract for a minimum of ten appearances on the Ed Sullivan Sunday Night CBS TV show. Their first date is October twenty ninth, and they're going to perform their big hit, The Rain, the Park and other things. Many know it as I Love the Flower Girl. She sent me done with more at Man from yesterday 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 Berratge 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 eighty five. Their services free and after forty two years of the business, their agents are trained to help you pick the plan that's right
for you. Was your car involved in an accident or just need help with dense All Magic Paint and Body Collision Centers in business for over thirty years, they're highly trained staff and certified technicians and friendly staff are the best in the business and treat each car as if it was their own. All Magic Paint and Body Collision Centers are family owned and offer state of the art equipment and tools to ensure optimum results. They use the latest technology in computerized color matching
and specialize in frame repairs with their modern laser measuring systems. They're OEM certified and they have four locations to serve you. All Magic Paint and Body Collision Centers offer rental car assistance with free drop off and pickup services too, and their work has a lifetime guaranteed. All Magic Paint and Body Collision Centers are in Narco, East Vale, Marino Valley and in Fontana. Call them at one eight hundred sixty one Magic. That's one eight hundred sixty one Magic.
All Magic Paint and Body Collision Centers one eight hundred sixty one Magic All Magic Paint and Auto Bodies says, drive carefully. Listen the CACAA Loma Linda at one O six point five FM, K two ninety three c F Burno Valley, NBC News Radio. I'm Chris Gragio, Secretary of State Anthony Blincoln will be heading back to Tel Aviv tomorrow to continue a world win diplomatic mission. He made stops in Egypt and Jordan today to urge Arab nations to work together
and keep the conflict from expanding. Blincoln says the US stands with Israel, and that's one of the four key objectives of his trip to the Middle East.