Strategic Enterprise AI Deployments: Mapping out AI in your Enterprise Business with Reuven Cohen at AIE 2024

Hi everyone. I'm Ruben Cohen, and today we'll be presenting strategic enterprise AI deployments. So if you're interested in how to deploy large scale AI with a specific focus on generative ai, this is the presentation for you.

So, to get started, I need to tell you who I am. I'm, I'm Ruth, and I've been doing a lot of enterprise work, probably for close to 30 years. This shows my age a little bit, but this has been a kind of transformative period, not just for technology, but for me as a enterprise architect.

And I've been lucky to sort of surf multiple waves in my career. And this generative AI wave is just a second of, of many waves and that I've seen over the last 30 years. And the, my work has been most notable within the cloud computing world.

And as an early sort of innovator within cloud computing, I was lucky to be sort of at this forefront of a lot of the technologies that were become prevalent within enterprise architectures around cloud computing. I was the one of the guys that was involved in these, the infrastructure as a service movement as an early advisor to some of the larger companies, including Amazon, in their AWS service in 2006. Uh, and that led to a whole variety of, of work in, in some of the larger enterprise deployments, uh, including large companies and governments, and just really interesting projects o over the years.

In 2015, I was lucky enough to sell my company to EMC, and that kind of was this nexus of the next generation of what I was looking to do. And that led me to Microsoft where I was, uh, briefly CTO in the Microsoft's cloud and AI division, not, and that, that really opened a door for a lot of the interesting things that have sort of followed more recently within cloud, the sort of integration of cloud computing and generative ai. 5, which kind of led to the chat GPT environment and launch in November 2nd, 2022.

Now, for me, the launch of chat GPT was really that sort of point that led to a lot of interesting opportunities. 'cause it was the first time that really a broad and accessible generative AI platform was made available to really the com the broader consumer sort of world. And all of a sudden, within a matter of months, we had more than I think a hundred million users of this particular platform, making it one of the fastest growing commercial technologies ever sort of seen.

But this also sort of spread the awareness of the technology and created a kind of bottom up approach where suddenly you had all these people using generative AI and kind of chat like experience, and they're saying, well, why can't I use this within the context of my business? And it was, it was really sort of an interesting inflection point, especially for those of us working in enterprise technology. Suddenly, not only was this technology broadly accessible, but companies were saying, well, people are gonna use it regardless of whether we sort of say you can or can't.

So we have to get ahead of the curve. And a lot of companies started saying, well, how are we gonna do that? What is the process that we can go about integrating these types of generative technologies into our companies?

So being, being, I was a few years ahead of the curve, you know, working with OpenAI and, and others, I sort of found myself at, at this sort of point where companies were coming to me and saying, how can we do this? 4 billion budget. So this presentation is gonna explore some of the challenges and opportunities for integrating the use of generative AI within the context of a enterprise deployment.

So the first question that you need to really look at and ask is why ai? You know, there's a lot of discussions, well, this isn't just a chatbot, is this isn't really that different than a lot of things that came before it. But from my vantage point, it, it's, it is truly different.

It's differentiated in a number of different ways. So when you look at the earlier stages of generative AI and, and things that sort of came before it, things like machine learning and deep learning, you'll see that those early iterations of this technology, although powerful and valuable in many ways, they were hidden from, from view. For a lot of users.

You weren't able to, to go and just simply use a machine learning system, although those systems underpin a lot of the technologies that we take for granted every day, whether it's finding fraud in a, a credit card transaction or using a search engine or social media and finding those patterns. But the ability for just the average individual to go and be sort of empowered by this technology wasn't there. It required huge amounts of sort of technical capability.

You have to be a programmer, an engineer to really understand how to use it and how to implement it. Now, suddenly with generative technologies, generative ai, you, you had this sort of human-like natural language interface to this technology that allowed you to just ask a question and get this kind of response. It was almost like a kind of alchemy.

If you ask the se the, the right questions in the right way, way anything's possible, it's almost like a spell. And what really happens is it empowers you, empowers you to find the opportunities that you never knew were possible. It provides this information in a way that's tailored specific to the way you think, the way you need the information provided to you.

And that's a critical change in how this technology works. Suddenly, you have this kind of limitless pill, you can access this information and not only does it give you the information, it gives you this information in a way that's tailored specifically to the way that you need that information. So there's no worry of saying, well, explain to me like a 10-year-old.

You can, you can have it provide to you in a way that's unique to your way of thinking, the way that you need to consume that information. And it kind of acts like, again, like a superpower. It gives you the ability to have this information, whether it's just discovering something new, the kind of ai rabbit hole that a lot of people will, will experience when using things like chat GPT for the first time, I've discovered something that I didn't know was possible.

I find these areas of, of interest and, and you, it's unique in the sense that it's catered specifically to me. But there's also drawbacks when you play with this, you realize that a lot of the, the information that's being provided to you is provided to you in a way that isn't actually always true. And now that's a double-edged sword, sometimes referred to as a hallucination.

Now the hallucinations interesting because it provides this kind of creativity. It provides this area, this gray area between what is real and what is possible. And when you play within this area of the kind of creative confines of ai, you start to discover that AI sometimes will invent things, it'll hallucinate things that it thinks should be possible, but isn't true.

So there's this sort of balance that you have to have between, you know, this is an opportunity versus this is completely invented and isn't true at all. So on one side of the spectrum, you have companies and organizations saying, well, we need to make sure that this technology is, is provided in a way that's guided and and shown in a way that can be really specific to how our customers need it or how our employees need it. And we, there is no room for interpretation.

It has to be deterministic. And deterministic is one of those words you'll hear a lot of. And the basis of a lot of these AI platforms is this idea of a kind of prediction of the next word or the next sentence related to what came before it and what most likely will come after it.

So you're, you're playing with this kind of dichotomy between sort of the invented and what needs to be factual and true. This is a challenge for a lot of enterprises, and we'll, we'll dig into a little bit of how that works and how you can sort of negate some of the challenges, but also embrace some of the creative opportunities for using this technology. So as I mentioned earlier, I have 30 years of experience.

I started young. I'm what a 8-year-old who gets a, uh, Tandy computer in the 1980s looks like grownup. I really enjoy the sort of emerging technologies, whether it was, you know, the internet in 1994 or the first online systems when I was a kid.

I've always had this sort of interest in the emergence of, of new and interesting technology. So when, when I looked at the future, uh, as a as as a kid, I, and I thought, you know, what does the future look like? I was inspired by things like star Wars and robots and artificial intelligence, how, and things like that.

And it wasn't until really recently that a lot of those things were possible and they weren't just possible from the fact that, you know, um, it, the technology was there. It was possible in the fact that now I can build it and anyone else can build it as well. So I came up with a, a structure, as you can tell, I named it after myself.

And, and, uh, I have a tendency to do that, but it's based on the concepts of what I call the roof method. I'm r. And so when looking at consulting and implementing these systems, I'm driven by this idea of a kind of human-like approach to the technology.

The, this technology should serve us. We should not serve it. So to create this sort of approach, when I look at implementing this within the structure of some of the Fortune 500 companies that I work with, I, I always have a kind of guiding light, a guiding principle that drives me through a lot of these opportunities in terms of responsiveness.

I don't wanna be painted into a corner and I don't want my customers to be painted in a corner. Choosing technologies that aren't, aren't adaptive or capable of being changed after the fact. It's unifying.

There's always multiple technologies. There's always multiple ways to solve a problem. And there's should be people as part of that process.

And this technology piece should be a unifying agent that combines all these opportunities into a singular sort of, uh, point of view, but also one that's not so specific that it can't be adapted to the ever-changing sort of nature of the technology itself. Just go on to social media every day and you'll see that there's some new and amazing AI that's been created. Now that speaks to the visionary part.

I can't predict the future more than anybody else, but what I can do is help build what that future may be. So the visionary aspect assumes that as part of the sort of adoption of this technology you're gonna be creating with the technology, it creates a kind of reinforcement feedback loop where each new advancement sort of builds on the previous to create some new and interesting new opportunity based on the previous. So think of a kind of feedback style loop in any of the approaches that we take here each.

And, and this AI itself should be part of that loop always sort of investigating, providing guidance and insights into the things you're building. So a lot of the approaches is that kind of amalgamation and integration of smart capable AI based systems into these, the flow of the people who are working within the, the construct of your enterprise. So think integrated, unifying, and responsive to change.

Now, what does that mean for the modern enterprise? Well, without doubt, AI is one of those transformative elements within any enterprise today. And something like 77% of enterprises right now are to some degree, directly integrating generative ai.

And probably within the next few months, that number will probably be a hundred percent. You basically can't run a modern business without some level of ai and more specifically generative AI within the context of that business. Now, what does that actually mean?

It means likely that a lot of the processes of business, which are human-centric, HR people are at the core, a lot of businesses, you're not gonna replace the entire business with ai. That's just not feasible. Nobody wants to work for a company that's only dominated by AI and or, or interact with a company that doesn't have real people.

So what you're looking at doing is the sort of integration of these systems and these processes in a way that's seamless and augments the people and the processes that exist within your business in a way that's, again, seamless to the end user. You don't want a lot of the AI processes to be overtly obvious. You want them to be empowering, you want your people to focus on the things that are important, but you don't necessarily want that to be the, the, the core sort of tenet of your business.

It's like back in the, the older days when people started adopting enterprise DA databases, you never saw someone say, Hey, we use an Oracle database. That wasn't really the point. You used the Oracle database to do things that weren't possible, give you access to large amounts of information or, or be more streamlined in, in the integration of that information.

But it was never really the central selling point. So right now we're in this kind of strange mode where a lot of people are saying it's, you know, AI is the whole reason for our existence. And I, I don't really see it that way.

I think it's one part of a larger puzzle, but it, it can't be the main reason you're, you're, I think the French say the re long deck for why your business exists. So a lot of what we're, what I look at is that sort of integration of these autonomous systems and these systems that continually operate on our behalf that can integrate in much the same way an employee does. But their intention isn't to replace those employees.

It's to find the right tasks that augment those, those, uh, employees and people and processes in ways that are going to free them from the more mundane parts of their work. Should you be spending all your time doing time sheets? Should you be, you know, doing, you know, detailed reports when you could literally just ask or have a system that does those for you in a more optimized way?

Now that's a, that's a balance, right? And if you look at some of these technologies, you'll see that, you know, one of the most obvious is in the adoption of things like code and any sort of development environment at this point within enterprise are using some form of copilot, whether that's GitHub or some other system. And the early sort of analysis of those within these businesses have been tremendous.

We're seeing something like 40% increase in productivity. Now, a 40% increase in productivity across a larger enterprise could mean tens of thousands of people and, and hundreds of thousands or millions of hours of, of time that's now saved. What does that mean?

Well, it can mean two things. Most likely it can mean the fact that those people can now focus on more valuable tasks. It could also mean that you may have more people than you need for certain areas.

So there's a, there's likely gonna be a sort of adaptive sort of change to the way that we organize and how we sort of structure these businesses from a person and personnel point of view. Having a way to understand and see that is critical as, as we sort of start integrating this within the business confines. So again, a kind of analysis approach is gonna be very important.

Now, when I look at this, I see two, two paths. As I mentioned. There's this sort of strategic path.

This is sort of a cultural transformation. This is how you integrate the technology from the point of view of the sort of culture of your business itself. And you're not, you're likely not gonna integrate AI everywhere all at once.

You're gonna take steps, you're gonna start with things that are probably not mission critical. These are administrative, these are back office, they may not even be customer, uh, facing at this point. And you're gonna get your sort of sea legs for what, what and how you can integrate that.

Once you're comfortable, then you are able to sort of more strategically integrate the, those various facets of AI into your business. Now the technical path, as I mentioned previously, is actually a little easier when it comes to things like development and scale and deployment of applications. AI is particularly well suited for code centric activities.

So if you're looking at increasing productivity quickly, one of the, the quickest path to doing that is to look at integrating that with the technical parts of your business who are probably more comfortable with these types of technologies. So I often take this kind of choose your own adventure approach to the consulting work that I do with customers where we, we on one path, say the executives and the leadership of the company need to be familiar and comfortable with the technology. And that's generally this idea of a chat bot or sort of responsive approach to ai.

Ask a question, get an answer, ask multiple questions, get better answers, ask structured questions, this idea few shot or multi-shot approach. So for those who are not familiar with this, at this point, a zero shot within the context of a lot of AI is asking a simple question like a Google search will be a zero shot. Tell me about something.

But I mean mean no context to me and what's important to me. Now for a lot of enterprises, a zero shot approach, just tell me something about, you know, create me an RFP response isn't gonna be very valuable. Create me an RFP response for a particular customer that we've been working for, we're working with for 10 years, that drives a billion dollars worth of revenue and we know all this information about is gonna be more valuable.

So having a structured approach to how the intelligence is both re requested, integrated, and used is critical to the success and the value of these types of technologies. So the, the next part is this kind of pick and play. The pick and play approach is this idea that I can pick certain things that are, that are highly valuable to sort of get the buy-in from leadership, the get the most, the most value to sort of have those quick and early wins and to have any kind of adoption.

What you don't wanna do is say, I'm gonna boil the ocean. I'm gonna do everything all at once and I'm gonna transform my company overnight. That makes no sense and it doesn't really work.

The, the idea of a kind of micro transformation on the other hand is really the approach that we, we need to take many little small steps that are transformative small on a, on a group level or even an employee level can add up to major and massive change. A agility is also critical. The ability to actually have this technology used by your team is a of critical, critical importance.

A lot of times in enterprise we're forced to use new technologies. There's a new portal, there's a new thing that we all have to do for our time sheets or whatever it happens to be. And it seems like a daunting task.

I gotta learn this new thing, I gotta, I have to use it. The key here isn't to make this a a kind of friction. You don't want your you user base to have to be forced to u use this.

You want them to, to see that there's value and empowerment by using it. So like, uh, a big part of this is the user experience itself or the ux. So small changes like integrating it into a teams or a Slack channel or a mobile application that's more seamless in how they can access it, really will increase the adoption level of this technology.

Lean teams, one of the key parts here isn't, isn't necessarily for large adoption, but very targeted specific adoption that can act as a kind of catalyst for others within the structure of your business. So even if you have 400,000 people, don't deploy it to 400,000 people, but look at the individuals or, or the groups that could really use that to iteratively and adopt that technology on a, a sort of quick basis to prove to others that's it's viable. So I like this kind of pick and play approach.

Now, the other approach that we need to look at is, and this is particularly important within the enterprise context, is the build versus buy approach. Now, a lot of applications you can just turnkey pop in and be often running quickly, and that's often, you know, provided through a kind of SaaS like interface. And if you go onto pretty much anyone, anyone's website at this point, you're gonna hear about gen AI or AI in general being sort of infused into pretty much everything.

But there's always gonna be a sort of balance between this idea of a kind of one size fits all AI, where it's sort of built for everyone and no one in particular versus this idea that I need something specific to my particular business, whether that's regulatory, legi, you know, the idea of logistics or jurisdictional limitations or even data limitations. So I've created a kind of buy and build matrix here that sort of helps guide you through the process a little bit. So on the left hand side, you'll see, well, there's application only.

If I need something quick and easy, let's say it's email, and then the email, uh, you know, I it's not really a mission critical application. The data is pretty straightforward. You know, I might need to insert some information from let's say a repository of, of documents in, in, let's say Microsoft or Google, things like that.

And it's fairly broad in, in sort of its use. I don't need to train a found educational model, the foundational model being the kind of underpinnings of a lot of these systems. Then I'm probably in a position where I can just buy the thing and, and have instant or value from it.

Now, as you move to the right, you'll see that there's this level of customization that starts to occur. Maybe I need to have the insertion of specific data from a, a unique database, um, sometimes referred to as retrieval, augmented generation or rag. Now that's where you start thinking, well, I might need to have a kind of mix and match approach where the application itself might be something like an email program or your outlook or what have you, but you might actually want to integrate that with some of the, the sort of unique data that exists within your organization.

Now, I don't need to fine tune the fine tuning is the idea of taking a foundational model and mo modifying it to the particular data structures. Maybe I, I can get away with just inserting the data in parallel to the, the requests being made and a lot of applications fall into this realm. So this is a kind of, well, I'm gonna build, I'm gonna buy most of it and I'm gonna do a little bit of prompt engineering.

And prompt engineering is the kind of term that you'll hear a lot of, right? It's this idea that I can guide the AI through a, a really structured approach, which I mentioned earlier, this kind of few shot a structure of, of the information being provided to it and the middle, it was probably the most u often used. So it's a combination of a data retrieval, a unique sort of application, and then potentially fine tuning based on the, the particulars of, of that company and the, the jurisdictional or or regulatory limitations that are placed on that company.

And the larger the company, the more regulatory hurdles you're gonna face. So as you, as you get larger in terms of the enterprise space, the more you sort of sit in the middle to to, right, the smaller organizations probably sit a little more to the left. The things on the left fast, the things on the right slow takes longer to fine tune, it's more costly.

Now, as we go forward, you'll see that this, on the right hand side, I have this build models and logic from scratch. Now there's three basic principles that are driving a lot of these applications. There's this idea of logic, the structure, how decisions are made, the reasoning why the decisions made in the comprehension.

Does it actually understand what decisions were made and what the ramifications of those decisions are. Now if, if all three of those points are required, you're going to skew more towards the right of this matrix, and you're likely gonna start building your own foundational models, taking things like LAMA three and minstrel and other models and making those sort of optimized particularly to the needs of your business or, or structures in terms of your, your customer requirements. Now the drawback to that is it's gonna be a lot more time consuming, it's gonna be a lot more costly to do that.

And, but you're ultimately gonna have a much more differentiated approach than, than having a kind of one size fits all on the left. So it's a balance, cost, time, value. And this is something that, that I spent a lot of time with my customers sort of explor exploring.

Now, once we've sort of come up with this sort of approach, are we gonna build it? Are we gonna buy it? Or most likely we're gonna take a kind of mix and match approach in the middle, then the approach is human centric.

What we don't want this system to do is drive us. We want, we want to be in a position where we understand the underlying, um, structure, as I said, the logic and the reasoning of these systems, how they're making their decisions, why they're making their decisions, and ultimately how those decisions really affect us as a business, the people within the business, and ultimately the opportunity for the business. So for me, this, this is really driven by this idea of a kind of human-centric ai.

And this is personalized. It's for you. It's, it's optimized for the needs of your, of your people, whether that's internal, uh, an employee or your customer.

The people need to be the sort of central tenant of how this is rolled out. If it's an afterthought, you're going to be in, in a position where you have a kind of black box approach where it might have value, but you don't understand how that value is derived and, and the more worse it's taking you like kind of autopilot mode where it's driving you down a path that you're not even aware that that's even happening. And one of the things that I do often to sort of get around that kind of black box approach to AI is I, I try to pick the ideas of design thinking, a kind of human-centric, empathetic approach to the use of technology, understanding what my, what my users want, how they work, creating kind of personas for how they, they do their sort of jobs in the day, in the, in the life of.

And then I, and then putting in tools that allow for a kind of human to always be part of that kind of loop. So if there's certain things that happen, it might be escalated to a person, or at least there's, there's feedback from people as part of that process. And that's critical importance.

And, and going forward, as you see some of these legislative and regulatory sort of frameworks come together, one of the key parts of that is understanding, understanding how these systems work and having people as part of that understanding. So long story short, having people as part of the process is gonna be very, very important. Sometimes referred to as human in the loop.

Now I mentioned earlier this idea of micro transformation doing one giant, uh, sort of transformative project never works. And whether it was the adoption of cloud computing or early internet or other sort of technologies where that were infused into the enterprise, this idea of I'm suddenly gonna roll this out to everybody all at once, generally doesn't work particularly well. So this idea of micro transformation is critical, small, manageable bite-sized pieces that are capable of, of guiding the use of this technology and providing a kind of measured KPI as part of that process is always gonna be important.

If you can't measure it, then you probably shouldn't do it. The other part of the process is, is the idea of quick wins. Not only should you be building things that are, are valuable, but you should have the ability to demonstrate that value quickly and easily through a series of, of different types of analysis and ways to sort of show the value quickly.

So, and the great thing about AI is you can roll these out in smaller pieces, show that value, even if it's in a smaller targeted group, and then use that as a basis for a broader implementation as you grow and use that more broadly within your enterprise. And as I said, culture, culture, culture and more culture. It isn't the da the danger of AI is it fundamentally changed the culture of your business.

What you don't wanna have is an AI driven business with really, uh, no sort of human sort of oversight. And, and this is gonna be a sort of balance that you're gonna have to do. So really having an environment that empowers your users is very important.

Aligning AI and business goals, having that kind of matrix in place, understanding where and how those objectives are actually met, and integrating AI in a way that seamlessly meets those objectives without disrupting the sort of flow of your organization as it currently stands, a lot of organizations are older and and established, so you wanna do so in a way that keeps that kind of vibe that already exists within the context of your business in place. Now communications is key. So a lot of the work that I do as a kind of independent contractor in AI is I help articulate that.

What does it mean? Where is it all going? How is this gonna benefit us?

So that ba that breaks down into a kind of CIO based, um, sort of commentary where often I'm, I'm the guy on speed dial for a CIO or other senior leaders to sort of help them in the daily work that they do, seeing where and how they can integrate that. And the other part is just articulating that to the, the broader customer base and broader user and employment base of these organizations. So a lot of these types of presentations, the presentation I'm literally giving you right now is the type of thing that you're looking to do within the construct of your, of your business going forward.

Talk show don't, and, and you can tell, but showing is always gonna be more valuable in that scenario. And again, it comes from leadership. This, right now we're seeing this sort of balance between the sort of bottom up approach.

Folks are gonna go use this AI regardless of whether you make it, uh, available or not. So the question is, do you wanna drive that through a kind of bottom up approach that says they're gonna go and use whatever random AI that, that they decide to use to do their job? Or do you wanna have a sort of involvement in that ai a kind of guidance of where and how they should use that?

So creating the tools that allow them to have that kind of chat, like, you know, experience that those intelligent agents that run on their behalf, but having a sort of framework for how they can actually do that and have a kind of continuous integration, a continuous learning process, whether that's the system getting smarter or just your people getting smarter in terms of how to use it effectively. So that for me really starts with a leadership that understands the opportunity and the fact that we're in this kind of transformative phase in, in technology right now where this technology, like it or not, is going to be used within the, the sort of structure of your business, whether you mandate it or not. So the real question isn't really whether or not they're going to use it, it's how they're gonna use it and how you can have a part in that sort of adoption level at this point.

So again, I've been lucky to work on a number of projects. Right now I'm, I'm doing a major project with the folks at Baxter International. They, they're in the process of, in integrating various types of AI in their organization.

A lot of what I just described to you is the sort of work that I'm doing there. And they've taken a similar approach, a kind of a micro transformation approach that looks at different parts of the sort of ecosystem. One of the things that I think they're doing a great job of is creating a kind of data fabric that creates a kind of unified approach to all the information that pins the Baxter business.

Now, if you're not familiar with Baxter, they're one of the largest, if not the largest medical device, uh, companies on the planet. More than 300 million people use their technologies on a daily basis. The, when you're in a hospital room or an o or or emergency, you're seeing Baxter.

And a lot of that, that is create looking at creating technologies and implementations that are geared towards, again, a more human-like experience. What we don't want, especially in medical and life sciences, is technology that is, is lifeless. This should be for you.

And the opportunity for these technologies is to optimize it in a way that's particular to a individual patient. We have the opportunity for the first time not to create broad implementations of technology, but to have uniquely individualized treatments and, and diagnostics and implementations of these technologies that hyper specialize, specialize for an individual. We are at a point where we could actually optimize this for 300 million or more individual people who need individual care.

This is meant to be more empathetic, more health, uh, more optimized for you at the end of the day. Now another, another example would be the work I've done with the, you know, folks at ey, they're early in terms of adopting this technology. They, we did this rollout last year.

It was one of the largest enterprise, uh, deployments of generative technologies. They rolled this out to, I think 400,000 of their employees. And the first step was essentially a sort of optimized chat bot.

And that's what was really interesting about this project. They, they looked at sort of, here are all the various parts of our business. I think they had something like 1700 different use cases, and they looked at sort of how they could integrate this on a user or, or group basis.

And so they, they optimized initially for the more technical parts of the business developers and, and consultants and things like that. And ultimately, this was an empowerment for those consultants to really do be, do more with less. And it, it was a, a, a really great sort of success, um, in terms of looking at how they adopted this technology.

Now, the last part and is really an interesting sort of idea that, that i I do a lot of work around is this idea of a readiness assessment. Now a lot of what I just described is sort of understanding. So what, before you start, you really have to take a deep look internally of the things that are meaningful for your organization.

And this really comes down to this idea of an AI assessment. You know, what are the sort of, uh, technological infrastructures that currently exist? Do I need to augment those?

Should I just buy? Should I, should I build, you know, what are the particulars of my workforce? What does the, what does the persona of an individual within my workforce actually look like?

Back to the ethical considerations, you know, understanding the, the sort of our ethical current guidelines for an organization. Regulatory compliance, certain businesses have more regulatory sort of hurdles to pass than others. And understanding how AI will affect that, those regulatory compliance, uh, and structures are really important.

Organizational readiness can, is there a way to actually get this broadly or even specifically implement it within my organization? So before you begin, I always like to say create a readiness assessment. Have a structure, understand as best you can and adapt that as you, as you move forward.

Now, the the next part of this assessment is evaluating the technological infrastructure. Like, okay, we understand that there's a need. How do we actually deploy this?

Are we gonna use a cloud infrastructure? Are we gonna use our own in infrastructure? Is it a combination of both?

So the technology and the sort of cultural components that I described earlier are always gonna go hand in hand the assessment of the workforce capabilities. Who's using it currently? How are they using it?

What are the benefits they're seeing from it? Ethical considerations, ensuring that these are are in line with the ethical guidelines that are likely already in place and making sure that this just augments those in a way that's seamless to what's already in place. Now the other part of the equation and a lot of large enterprises are doing is they're putting out RFPs and just about every large enterprise at this point is balancing between build it ourselves and, and have some consultants come in to help us build it in partnership.

So when doing that, it's important to have that kind of readiness assessment in place to understand those and, and as part of the RFP process, that could be literally part of the requirements of it. The other is a technical detailed requirements. A lot of the things I've been describing, um, the vendor expertise prove it.

A lot of, uh, folks in the space are saying, you know, we're experts. We've got 10 years of generative AI experience. The real the reality is this technology is nascent.

It's only been around for about a year, year and a half. So looking at sort of technologies that are similar but related, but don't overtly say, you know, I've got 10 years experience with generative ai. We're gonna know you don't have 10 years experience with generative ai.

It's only literally been around for a few years. And now scalability, don't paint yourself in a corner. Build things that are gonna scale scale over time.

You don't have to do it all at once. And I think that, that that's the, the critical takeaway from today's, uh, presentation. Now to help guide, you know, various folks through this, I've open sourced my entire guide.

It's about 200 and, and something pages of information. This will go in a lot more depth, but everything I've been describing in this presentation, it's geared towards a chief information officer. But the reality is it's a kind of pick and play, choose your own adventure kind of structure.

I've open sourced it, you can grab it off my GitHub at, uh, roof net if you're interested. You can contribute to it if you wanna update a section. Everything that I've been describing is in much more in depth in this, in this guide.

Feel free to use it as as you please. And we've had, I think there's been thousands of downloads of, of this over the last few weeks. Um, so feel free.

There's lot, lots of code examples, there's practical implementations. I've got a whole guide to things like the data fabric, everything you kind of need to get going. The, the guide is broken into two parts.

One is more management, consulting and sort of the understanding. And the other is like the practical sort of requirements of implementing this within the structure of your organization itself. Um, I'm easy to get a hold of if you're, if you wanna chat, I love to talk about this stuff.

Again, I'm, I do a lot of my sharing on LinkedIn, so you can look me up as on LinkedIn at Ruben Cohen. com or just shoot me an email. I'm happy and excited to, to be part of this adventure and I love this.

So hopefully you found this presentation interesting. I am just about outta time. So thank you everyone for uh, taking the time to, to view my presentation today and, uh, thank you.