Session Information:

This session will include a review of a recent worldwide Digital Transformation survey of 4,300 decision-makers from mid to large enterprises.

Session Transcript:

Toronto in Canada, he is an AI and data analytics leader for Dell, and I'm talking about the Great belong who is here with us today. Bill, is the AI and data analytics practice leader for Dell Technologies, responsible for supporting Canadian commercial businesses and public institutions.

He spends his time developing AI strategy for firms, providing guidance on how to accelerate artificial intelligence development, working with universities, and advancing AI research, and publishing a quarterly AI newsletter, focused on AI adoption in Canada.

Previously, Bill has held roles as an Enterprise Architect in consulting and AI, analytics in development, and product management, during his tenure at Dell, Microsoft, Oracle, and IBM.

He also published numerous books and white papers on advanced analytics and data management, and speaks often at conferences. And I'm always fascinated by the insights that he provides to us here at beat those.

Bill, thank you so much for sharing expertise with our global audience today, package Joseph and hi everybody, it's great to be here.

And what we'll talk about today is architecting an AI ecosystem, and I'll present to you as enterprise architects as I used to be, and really enjoyed my time as an enterprise architect, and I continue to use those best practices as I consult in the area of AI.

So, we'll briefly talk about, where we are, digital transformation. And then we'll talk about the AI ecosystem.

We'll talk about the hardware, the software, And, I'm going to, where possible, talk about a use case that we can relate these two, and this is a live use case or a case study, that, I had the opportunity to work with, one of our hospitals here in Ontario.

Then, as part of the ecosystem, some of the questions that you'll be getting is, how will this integrate with cloud technology, will end there.

First of all, let's take a look at why Canada and AI.

So if you've done any research in AI, in in universities, likely the papers will reference Canada. Very fortunate to kind of be at the epicenter of the big boon for AI.

Imagenet.

That was done at University of Sharjah, was the first time team put together the compute platform and GPUs, new algorithms as well.

This, at that time, neural networks, and then a large amount of data.

And that just was kind of the big bang, as the invidious CEO calls it, or AI, and it's been growing ever since.

This recent survey about the Global AI Index that's taken last year shows a few small countries here, the US, China and the United Kingdom, ahead of Canada, but, you know, in all fairness, you know, and about being realistic about Canada, while we're viewed as great innovators, we don't have the financial clout as the other countries too.

And so, the US, China and UK, you're seeing lots and lots of investments and focus on AI, but care.

We'd like to think that we're great contributors to this movement.

Now, regarding digital transformation programs, this is taken out of a survey that dose, every two years, and on the right hand side of your screen here, the five top programs, cybersecurity continues to be the number one.

No question on that, very important in the age of the internet, remote, working, again, from the last year and a half, everybody's had to work through. Again, no surprise.

And, where I come to is number four here, And so there are lots of great digital transformation programs, but you'll always find that leveraging data is in typically the top five investments that people want to have true.

This is tech knowledge that people think they can really differentiate their offerings.

As part of the survey that we do every couple of years, we also ask what are your key challenges here?

And not surprisingly, data privacy and security.

And as we go forward, I think they'll continue to be very important factors.

Now, last year perhaps accentuated the need for funding and the pandemic actually increase funding in certain sectors. and we will talk about one of those sectors to health care.

And then finally, number three, this is an ongoing challenge for many firms, Large and small, is how do I extract insight from my data and we'll talk about some of the recommendations we have.

Now, I'll start here rather than concluding on why enterprise architecture is so important, especially in the field of AI, and the leading firms will have enterprise architects, leading innovation, and that's 24%. And if they're not leading, they're certainly part of it.

And there are firms that we call, like, it's where ... Teams do not participate, Which I think is an error on their part because the EA is a unique pool where you have a holistic approach of how the business works and what's the value to the business.

I had a meeting with the CTO.

And the first question I had, after he explained, there's challenges where I ask them to do have an ear and they said no that they were looking unfortunately they did hire one and that EA reports directly to the CTO.

And, again, that's the best practice I have is person with these kinds of skills really needs to be involved in these kinds of programs.

So, a bit of a paradox that we're seeing.

And if you're interested in learning more about this, I'll be presenting about some of the paradoxes we see out there, there's no doubt that companies think that AI and data analytics is important, as you see by this chart, by Gartner, on the left-hand side.

But the paradox is that, while they say is important, the investments continue to live.

And there are many reasons for that.

But in the industry that we've seen today, there's still, kind of, they say it's important yet, they're not fully both feet in investing, But there are significant investments to say, in AI. And right now, it is the largest growing footprint in the data set.

Now, AI, these hype cycles dated 2020.

And I took a look at what Gardner's ready to 20 21.

They should be out shortly, but there's not too much change here. She took, Take a look at the second half.

Right, half of the screen here.

So, there's going to be a focus on generative AI And this is the technology that can create deep fakes, how to combat that.

You'll hear terms like composite AI more and that the term for more container based and using AI from individual projects, combining them as well as adaptive ML.

That's a term that we're starting to hear for ML programs that are changing and adapting based on real-time data, not just based on historic data.

Uh, now, as an EA, I go back I used to use the Zachman Framework for those of you who go back that far and but I I do like to talk after co-chair.

So if you are based in, and that kind of background where you have an application data technology architecture, business architecture, and a lot of companies respond to that well.

So what I'm going to do is take you through kind of a case study here, and some of the artifacts that we created here, beginning with the vision and use cases, understanding, and documenting for the stakeholders, what the current environment is, talking about, you know, what is the art of the possible target architecture.

Based on time, I'm not going to get into details on exactly the steps on the strategic roadmap here, for you to ask those questions of your life.

But again, it's part of any kind of transformation of governance, very important to have that as part of the program.

So.

So, this is a real live artifacts not gets created for this presentation, but we did present to the C level executives for this hospital.

And their vision was to create an AI Center of Excellence center of excellence, a sandbox where researchers would come in with their algorithms to test them out.

Their environment was that every researcher and they have about 30 projects going on.

Everybody was doing their own kind of linear IT project.

Everybody's using different algorithms. Everybody's using different hardware.

Everybody's getting private data and they thought there's got to be a better way to do this, to save money.

So they look for opportunities to standardize consolidate, and they really wanted to get deeper insight.

So there's a lot of data locked in proprietary type of databases. If you're in healthcare, you will have heard of databases called Epic.

They have a lot of healthcare data, are very difficult to access.

And so, when we talked about possible use cases, AR has really kind of changed the landscape of what's possible out there, what some of the things the hospital's looking at.

And what consumers were, where for them adding is, hey, can I use my, my Fitbit? Is that useful information?

So lots of smart devices there.

People have personal devices, but also at the Bed's remote monitoring, the hospital, that was certain width also had a huge research arm and teaching on.

So, we're looking into Augmented Reality today, too, help doctors perform, let's say, surgery, and it's all done virtually, so that our technology skills for today, virtual assistants or frequently asked questions from patients as well on the right hand side.

Digital typography, precision medicine, these are all things are enabled by image recognition, AI, algorithms, and neuroscience, as well.

The use case as part of the.

This transformation exercise, the one that they wanted to focus on, was reducing emergency wait time at hospitals.

So, it's a bit of precision medicine, and right now, like wait times, is in the hours, And that's not bad, but they thought, hey, are an opportunity, really, to reduce that?

On the bottom, there, are kind of, what business drivers for the institution had, so they want to be known. And one of their charge is to deliver the best patient care possible.

They also wanted to foster research, and a great learning environment for new physicians.

And, as well, do new types of research. They had.

They have an HPC cluster available to the researchers.

Now, this is a kind of a set of architectural principles, and typically, when I get involved in developing a strategy, usually I see about 12 architectural principles listed out.

And these are, are things that help design and help the team select what technology you want to incorporate.

So it's a useful thing to set up upfront, and these are high level, want to adopt open technology course. You want something that can perform and scale, et cetera.

But you want to get sometimes to the next level of detail here, And I'll talk about the ones that were relevant for this project.

So, always useful, too.

Put out what your current stages, and, outside of listing, the asset's here, it's good to paint a picture of what the key challenges are with the current state, and the Mall's development.

So, again, I mentioned before, each researcher kinda does their own thing, but swell, they're challenged.

Outside of just developing in Python, There were no tools to automatically promote code into production, and then two, what happens to actually deploy in production, and I monitor, etcetera, Do I have to change the model?

Another thing they found here with their researchers is that there is a big opportunity to somehow consolidate and standardize data. Everybody went after their own dataset.

Everybody had to vet it, and data really wasn't a shared resource.

And even if it becomes a shared resource, as we've seen a lot of firms go, hmm, where? Where is the data? Or what is this data?

And so, we'll talk about what best practices are here now.

And then, finally, we saw some very inexpensive solutions on people on where they've got their servers and storage infrastructure.

When you have something under the desk, people would beg, borrow, and steal, and with some people who are going to the cloud, et cetera. There was no standard infrastructure out there.

So, on the right-hand side here, again, it was assembly requirements that came out to talk to the researchers, as well as with the IT, as well, is a platform that they could leverage, that's scalable.

They will perform and support their favorite tool.

OK, so, now, it's part of the ecosystem.

People have a choice on the hardware, and why that's important is that for some areas, medical research, some of these require just them intensive amount of compute power.

So I'm not saying that, you know, for, for all AI type of work, you know, you have to pick GPUs.

However, having said that, if the work is really, you know, deep learning intensive, and really needs kind of were the best and latest algorithms for natural language processing and image recognition, likely it may skew toward GPUs.

Today, there is a breadth of the choices on your far left here, CPUs, which everybody's familiar with, the chip that powers your laptop and the big players, Ariel Intel, AMD.

And the advantage of this is everybody's familiar with the applications that live here.

Um, and then as we go to the right, you get more specialized processors here.

Some video really, the £800 gorilla when it comes to providing GPUs.

And then FPGAs, field programmable gate arrays, this is more flexible and but you have to code at the firmware level.

So we don't see as much adoption here then. We do see some adoption on certain industries.

Then on the right here, on the far right, a six year application, specific integrated chips, These are chips that are especially designed to run deep learning algorithms really quickly, efficiently, takes less power, should be less expensive.

Most of these will be in IOT devices, where they build the intelligentsia, but there are places.

Companies like Graph Core, a unicorn from the UK that has done deep learning and claimed that they are the fastest server out there.

Don't, we were one of the only ones that offer that choice there, but there is a choice, and, again, this should be driven by the workload.

Other things, when you develop an ecosystem for your environment, is to understand what kind of trends are out there?

So, when we understand our, this new technology, those AI hype cycles and types of algorithms, also, what we're seeing, a trend, is kind of a merge of workloads.

And if you have experience with simulation workloads running on HPC systems, there is this new, kind of, data pipelines that requires both simulation and AI deep learning algorithms.

So one example of this in health care is actually research that was done to better understand and to create a virus against cold air.

So they used actually AI to predict the molecular structure and if you're have a background in structural biology and chemistry, the approach they take is they want to develop an inhibitor to the buyers. In order to do that. Well, what they really like to do is understand the physical structure. And that's why it's important to understand that three-d. structure there.

Once you have that, rather than going to the wet lab and actually testing this under a microscope, we wanted to do this in silicone in computers, wherever possible.

And that's where you need lots of power, compute power, and HBC does these proteins simulations, so very compute intensive, And it's great faith that they were able to produce this vaccine, in a short amount of time. They did.

I will tell you that, in order for, for them, to create that, massive amounts of compute are used across world, and it's great to see if you, if you're in this field, there's a lot of collaboration, a lot of sharing, a lot of companies, giving up cycles, three compute power, just so researchers can better understand structure of the molecule and what would inhibit the virus.

Um, now, for a lot of companies, they're looking to machine learning as a service, or what kind of software tools are out there.

And for the first time, the quadrant on your left, they actually put the cloud vendor tools with the on prem tools.

And most of the cloud vendor tools are in the bottom right-hand corner.

Today, with today's technology here, the best tools are appear to be. by adoption.

Are the on prem tools, say the cloud vendor serpent stands still there, but today I think it's fair to say the on prem tools have an edge of integration With the development tools.

Now, that's a foundation, very important to have a data platform, and this should be optimized for the industry and for the workloads that you're going to support.

So for health care, or medical research, on the far left here, these are the kind of data sources that you see.

And in this use case, the EMR, or, yeah, are tronic, now, called, Records.

This, this is the kind of data that's held in Epic databases that we are trying to extract, and we wanted to extract it in real time.

Which means that we needed a streaming solution as well.

So, yes, that itself was an interesting exercise, but as you can see there, there are other types of data here, from research papers to real-time monitoring of patients.

Medical imaging, these are very popular data sources that are being consumed today.

And as you move to the right, what you want to do is create a platform that we're just this enriching, curate the data, and then make it available.

And a platform that will perform well for the personas that we see on the right and our use case here, we really want to make this available for both researchers and the clinicians, the people on the frontline.

And part of this ecosystem is another way of putting it is, you know, selecting vendors that kinda meet your architectural principles.

And they're single, so you would not kick.

Platform with all these tools, these are just examples of tools that you pay.

Let's say for, let's pick in the middle string solution.

So elastic search, Splunk, very, very popular tools, casco, if you're going open source, et cetera.

And then, as a data lake, Cloudera very popular for Hadoop, in our use case, it was interesting. This hospital did not want to use Hadoop.

So if you're familiar with Hadoop, very popular platform for unstructured data, this hospital thought that there's going to be newer technology signal I'd be tied to, to, to do.

So they just picked object storage as a storage device going forward for unstructured data.

Just a couple of points about the cloud.

What to talk about is, we're seeing a lot of companies kind of doing a pause here, and I see a lot of firms that they have mandated, everything's gotta go to the cloud.

And I think what people are finding is, well, there are some benefits, putting everything in a central spot.

But the original ones about saving money that seems to have fallen by the wayside, a lot of people are going toward cloud.

Really four.

To try to be more agile, time to market, which are great nuisance.

Bye.

Cost isn't necessarily the justification of the cloud.

Um, and what they're finding it is that in some cases, especially with AI, the costs are quite dramatic on where you place the data.

So if you're training an AI model to do, let's say, image recognition here, I have one firm here Ontario, who will tell you that it's, it's at least a tenfold difference.

And so, what we're seeing as most firms are picking a hybrid approach, many would like to develop.

Some would like to develop on the cloud as well.

But, if you have a lot of data, they put the development on prem, and when they finish with their model, they, they put that to execute production in the cloud.

And so, they step back and ask themselves, what, what makes sense? Where should the data go?

And especially with healthcare type data.

May and hospitals are very concerned, and wanted to make sure we got the best security.

And then today, they, they could, that opera does say the cloud security isn't as good, but they just wanna, like a security bug right there.

Public cloud adoption should be no secret that AWS has.

Market share, leadership has had for a number of years.

Asher's number two, actually, Azure is actually doing better by comparison in Canada. For some reason, why?

Relatively speaking, they seem to be doing better here.

Number three, Google Cloud.

And number four, which, probably most people don't know, is v.m-ware.

So, v.m-ware folks, that virtualization technology can build something or work together with containers and migrate that seamlessly to the other cloud platforms.

So, for your reference here, I've listed, or AI, data analytics type of applications, what each cloud vendor would offer there.

And just because they have an offering, it does not mean it's the same capabilities, functionalities, as its neighbor column there, you really have to kinda understand what are the functions that you're asked to deliver?

And then take a look at who's got the best mousetrap.

In addition to AWS, Azure and Google, I have listed things that Dell offers and Deleuze approach here is really to partner with you out there.

So most of them in that column are independent software vendors.

Costello really doesn't own software.

You still want a company called Gloomy, and they'll become independent next, fiscal quarter. And then lastly in the last column, the open source, if everybody honest.

The hyperscale offerings, You can pick a source.

So with that, and understanding all that, then we tried to craft our future state.

And what we're moving toward here is a model driven environment.

And wow, I know a lot of folks like to code in Python, a lot of people do now. There's all equally large and people who don't want to code.

And likely, what we're going to see is more and more tools that don't require coding, and really, to get more citizen type of data scientists to be able to create these applications.

So, machine learning as a service, to reduce the time development from months to weeks.

Then, as I mentioned before, in the middle layer here, this data's so you have all this data, and if you do mesh consolidate all, how do you figure out to where the data is? And the technology that's come out of the way is this data catalog.

And there are a number of vendors out there, and Boomy has a data catalog capability.

But what you want to do is, to try to remove IT from the equation of a self-service portal that researchers can come in, your data scientists, and they take a look at a menu.

And say, OK, here's the data, Pick this data, and description, dictionary a glossary of what the data means.

And so, companies are using that today.

And then, finally, Platform as a service, how to deploy the infrastructure, which includes hardware software using Moodle, and leveraging Kubernetes, having everything kind of, containment.

And so, as I mentioned before, know, for this project, and these were kind of the main architectural principles that guided their selection of technology, on Agile thing, they wanted us to free.

They didn't want one vendor, either Cloud or on prem vital tools.

The things open-source tooling, I mentioned. Or hybrid.

So, they recognize that there is use and benefit of the cloud, but they didn't wander. Warwick.

This is just to highlight some of the resources that Dell can bring to bear when it comes to AI, and we have a bunch of data scientists that I Leverage.

And this place, just around Austin.

So we have a super cluster there, made of GPUs that I've leveraged for one of our universities, super cluster for AMD, processors, and superclusters of Intel, So whatever the workload wherever the customer is.

Going or favorite platform is, we help test those algorithms.

And it's a service that are chargeable.

So this is my last slide, um, and what's the best practices here?

Some of the things that we learned here is, 2, one is to adopt this data driven culture, and it's very important that the benefits of doing this before, in the past, sometimes with similar organizations I've seen, is the, guard, their data really closely. They, they don't want to share it, they think that it's more valuable, very valuable.

Yeah, and they have to go through, you buy, in this day and age, the data driven, world oriented, You're going to get more value to more people if you share it.

I know it sounds kind of simple, but one of the hospitals I was working with, government, know they they. They written a paper, a great paper talking about what their digital transformation strategy is, then another great strategy paper, or what their data architecture should be.

But when it came to execution and practice, what we found is the IT people really guarded the data. Say, Hey, it's either. We're not going to share it with the people who could really use the data.

And these are researchers working in healthcare. Say, Hey, we need that data, So, we can kinda, there will be small Z, They couldn't come to bear that.

We couldn't, fortunately, agree on how to do it.

So, as many of us have probably seen, what happened is, these researchers, their line of business, they have their own budget, They bought a cloud database, so they created another silo of data.

Yeah.

They brought me in to try to say, hey, while it's great that you're drawing, this line is kinda making problem more complex here.

Really got to find a way to kind of work together.

So, again, sharing data, Let's say you do, and then, as an enterprise architect, take an architectural approach to use, uh, and involve the stakeholders, make sure that there are business stakeholders researchers involved, that this is not just an IT driven project, and we can get into details on the governance, but, again, if you have a tool, like a data catalog, those tools help automate them, make people adhere to governance policies.

So whole bunch of vendors there, again, pick the ones that are best for you, Strategy that we have here at Dallas, we want to be, you know, OK.

That's it for that, and now we can for Q&A.

Excellent, Bil, excellent, though, I'm gonna ask you to do me a favor, actually, keep the presentation sharing, because someone asked about slide number 22, the one where you put the different vendors and the providers there, yeah. Would you share that one momentarily?

List that I know you have made, you have asked the question and others have pointed out should this slide as well, You know, you have access to the recording of the presentations, but you can go ahead right now and take a screenshot here.

They thought this was an incredible insightful, uh, data that you're sharing with with us.

And thank you for doing that great work here and kind of distilling down the options and the there are available there for different levels of processing and applications.

So, just the audience asked if you could give you, give them a little bit of time. So I'm assuming that by now, people have taken the screenshots if they need to.

But this is very great, great, yeah, very generous of you to compile and share this. As happy as, some people have found it, a benefit. I did take off the tech to create that.

In my kinda, after hours job, I guess, I was kind of fun too, I haven't worked at, yeah, uh, at Microsoft kinda knew their tools.

Here, here's the thing, It's funny, if you try to do this yourself, what you'll find is some of the companies here have changed the names, even those, the same tool.

And here's kind of a kind of a unlit Ms. Tennis and I use for companies.

If the company changes the name of the product repeatedly, it tells you that they're not happy with their market adoption.

That's kind of one of my rules of thumbs here. So similar tools here of change, 2 or 3 times in the name, same tool.

So I'll talk about that.

Um, but when the unfortunate, for some reason, a lot of marketers and technology think if they change it kinda repackage it, maybe again, it will get better option or sell.

So as far as I know, these are the latest names. I always have to keep checking it. If not, apologies but I do try to keep it current.

If you're interested in getting the full presentation, feel free to look at my profile LinkedIn.

I have we'll be posting this presentation up there as well.

Fantastic.

Fantastic. And of course, you also receive an e-mail with a link and password to Bill's full presentation. Not only the slides, but he is color commentary on everything that's not build. Fantastic. So you, you may now stopped showing the presentation screen. I'm going to come back on here and relay some of the audience quite other audience questions to you.

Um, and you if you go into the goto Webinar interface and just hit that button that says Stop Showing screen.

Yes, I'm trying to bring it up right now and I can't find it anymore.

Yeah, it can happen. It can get hidden behind applications.

Sometimes it's It's under, You see that little red arrow, just where you toggle your camera off to the right is where the tabs should be.

I see the sharing and Narrative to thank you. There you go. There you go. Thank you, Bill. Bill, so we have a few minutes here for for the questions. I'm going to start with one of the themes that has a merge and kind of an overall catchall.

Real type of question, if you have a blank slate, share company, you know, has not really become serious about developing an AI ecosystem. You know, there may be an application, a little application here and there, but you have not really built the AI ecosystem.

What would you do to implement in a way that, you know, that's not a large implementation scale, but something that you start small and scale?

What are some of the things that you think that organizations needs to take their time to get right upfront, and then, in building this AI ecosystem?

Yeah. Great question.

My recommendation, and I do get requests from customers, questions, exactly that. Where do we start? Like, no!

It was like, cloud was first, introduced C level execs are going, Well, what's our strategy? Well, what is it?

Is this is, I still think, from an enterprise architecture approach, you have to build a strategy, and now we're going to start small.

And first part of that strategy is to get in line up with an executive stakeholder and pick a use case that's going to get high visibility, That's low risk. I know that's asking you a lot.

Usually, the higher risk ones means it's much more difficult to put out.

But look for those use cases where you can get a six quick success story.

And that'll help get people on board, just because, I'll tell you, This is not an easy task. This is one of the most challenging.

technical things you can do out there, but the rewards are, I think, are worth it, or why everybody's interested, to digitally transform a lot of things that we're doing out there.

So, again, like you said, Josey, start small, but something that is high profile.

So, for banking, something with customer experience, customer insight, with health care, something, that frontline person clinician would benefit or a patient, and it doesn't have to be difficult, really.

Calm, complex.

So here's an example, Hospitalized worker just a couple of weeks ago.

What they found is we're a certain weather patterns, they have found that they've they get more people coming into the emergency ward. And those weather patterns tend to be under really kind of harsh loans.

And even though they have all these weather programs at the adhere to, it's never good enough in terms of predicting.

And so they got their own IOT devices.

Or maybe they leverage it from the government.

And they get that data stream, and they see that in and then, to their model and combine it with something else to help them do resource planning.

So that when certain weather hits, sounds like an inkling that it's coming.

They got more doctors on staff and nurses. It's a simple thing. It didn't take them more than a week to put that together.

But something like that, people are gonna see.

the status is good stuff to be able to combine that data with the resources. This IOT fee, it's just such a simple thing, but really high impact. That that's the kind of application.

That's, that's very good. The other question and theme that has emerged is related to the open source versus proprietary algorithms, AI engines. What does the landscape look like today, in your opinion, with open-source AI, and the, and then, you know, proprietary? I know that this can be like very application specific, and very industry specific, even. But just a few playing, a broad picture of how open source AI versus proprietary AI has continues to evolve. What is your assessment? What does it look like today?

Yeah, well, you know this, this field is changing so quickly.

I think open source has the advantage of getting things out there quicker, usually, because the commercial ones want to test those out.

And this is the kind of feel that, I've seen that kinda merges a lot with academic. A lot of areas in computer science you don't see things from academia and until much, much later, Jack. Daniel's really kind of research things, but with AI, it's quite different.

Some of times we see academic papers published, and then there's something there that they can commercialize almost right away.

And, and the, the advantage of the commercial vendors is, they can take that application specific expertise and many people forecast, and, I'm probably part of that group, too.

That's the big transformation, shouldn't see an AI, are from people who can take application specific knowledge, and there are the AI.

They will produce more transformative applications, then Google or Amazon, or are actually because they are kind of general providers of the platform.

They don't have industry specific knowledge, although that is changing.

You will see Microsoft and then, you know, investing in a healthcare cloud, same with Google. So there are some exemptions there because they go, hey, that that industry is really quite profitable as a march toward that. But let's say manufacturing. None of them have a manufacturing specific.

No line of business, so the real breakthroughs there will come from a company who goes, hmm, hmm, there's something I think we can improve here in the factory.

That subject knowledge, combined with AI, either commercial source, those are the ones who aren't yet.

That's very good, original submit and then Karen Kirby. both say fantastic content and content and industry depth. So, thank you.

And I was just a bit specifically, is asking if you can elaborate a little bit more on the approach it took to overcome the Data Silo mentality that you'll find on IT people, you know, how do you overcome that? Because it seems that maybe she may be experiencing something similar.

Yeah, Unfortunately, it's not easy.

What we had to do was, we had to go to the highest sea level, and IT, so, in this case, it was a CIO, and we had the line of business and their skill level exec.

Yeah, we had to get them to kinda agree.

And then that percolate down, if you worked at the ...

level, they're just gonna follow their mandates that they're not going to share or they're going to want to share.

And in a way, that IOT will allow it.

So, unfortunately, um, yeah, it's not an easy thing.

And sometimes, if there's a independent person, a trusted advisor, somehow, which is what I was brought in as, they go, OK, I didn't really have a bias or whomever. I wanted something else best for the firm.

And so, going back to again, what is, what helped them was understand what is that business, division of the hospital.

So, in this case, my case study, was teach in hospital, research hospital, but patient care.

And it's those things, at that level, you have to remind the IT folks and remind, or in the business, we're all in this together, alright, here's the things here, and let's try to align them so that, you know, we're all play nice together.

Yeah. They're always via chat.

Thank you, Bill.

It's always a masterclass with you, whether it's artificial intelligence, data analytics, you're one of those fuel leaders that combined the technical expertise with the collaborative approach in your leadership. And it's highly effective, it's a blessing for all of us to have an opportunity to learn from you and share it with you.

So, on behalf of our global community, on a say, our deepest, thanks and appreciation for you to share your expertise with us, thanks, Andrea.

Say, I always look forward to meeting you and your clients, I'm grateful for this time. I'm glad I can assist.

Thank you, Bill.

Chairs.

Ladies and gentlemen, that's belonged.

Leader of Artificial Intelligence and Data analytics. Adele, directly from Toronto, Canada to our global audience. Always a masterclass on on Artificial intelligence, Data analytics, and the Enterprise Architecture, and technology related fields. We are going to take a break, and then when we come back at the top of the hour, we'll have our last session of day one of Enterprise Architecture Life, and we're gonna bring you another tremendous practitioner. I'm talking about Andrew's C four, who's Director of Actuarial Data Governance for Nationwide. And Andrea will be talking about designing a customer focus governance model that aligns the business architecture for competitive advantage and continuous improvement.

So, again, a real industry practitioner, not only talking about this concepts, the living, the implementation of this disciplines, and the in the organization. So, taking a break now, I'll see you all back at the top of the hour. Thank you.