Rearchitecting Enterprise IT for AI Readiness
Key Points
- AI’s current breakthrough stems from large language models that ingest and process the vast public internet, effectively “swallowing” it to gain broad text and image understanding.
- Within an organization, the relevant data and applications differ dramatically from the internet, making the straight‑forward “AI‑swallow‑the‑enterprise” approach a poor fit.
- Most enterprise AI projects that try to bolt AI onto existing IT stacks (data, SaaS, custom apps, networking) fail—over 90% according to the speaker—because the underlying architecture isn’t designed for AI integration.
- To succeed, companies must re‑architect their IT infrastructure with AI as a foundational layer, mirroring how the human brain’s structure informs intelligent function, rather than treating AI as an afterthought.
Sections
- AI Swallows the Internet Paradigm - The speaker highlights that today’s AI, exemplified by GPT‑driven large language models that ingest massive internet data like a brain, demands that IT professionals rethink and redesign their architectures to become AI‑ready, particularly for internal organizational applications.
- Brain Structure and Sensory Processing - The speaker outlines vertebrate anatomy—including the cerebellum, spinal cord, and three brain regions—and explains how these structures organize and process sensory data from sight, sound, smell, taste, and touch.
- Frontal Cortex Integrates Experience - The speaker explains how the brain’s frontal region coordinates sensory input and long‑term strategic thinking to decide actions, illustrating its remarkable ability to fuse memories of sights, sounds, smells, and emotions.
- Enterprise Applications Feeding a Data Lake - The speaker outlines how various business systems (CRM, HR, finance, legal) connect via a network to a central data layer, discussing the differing qualities and purposes of data lakes.
- Turning Apps into MCP Services - The speaker explains how converting existing applications into Model Context Protocol (MCP) services creates tool and data‑source interfaces that agents can orchestrate across a partitioned data lake, likening each application to a brain organ.
Full Transcript
# Rearchitecting Enterprise IT for AI Readiness **Source:** [https://www.youtube.com/watch?v=yKPAwTF1SGY](https://www.youtube.com/watch?v=yKPAwTF1SGY) **Duration:** 00:20:38 ## Summary - AI’s current breakthrough stems from large language models that ingest and process the vast public internet, effectively “swallowing” it to gain broad text and image understanding. - Within an organization, the relevant data and applications differ dramatically from the internet, making the straight‑forward “AI‑swallow‑the‑enterprise” approach a poor fit. - Most enterprise AI projects that try to bolt AI onto existing IT stacks (data, SaaS, custom apps, networking) fail—over 90% according to the speaker—because the underlying architecture isn’t designed for AI integration. - To succeed, companies must re‑architect their IT infrastructure with AI as a foundational layer, mirroring how the human brain’s structure informs intelligent function, rather than treating AI as an afterthought. ## Sections - [00:00:00](https://www.youtube.com/watch?v=yKPAwTF1SGY&t=0s) **AI Swallows the Internet Paradigm** - The speaker highlights that today’s AI, exemplified by GPT‑driven large language models that ingest massive internet data like a brain, demands that IT professionals rethink and redesign their architectures to become AI‑ready, particularly for internal organizational applications. - [00:04:30](https://www.youtube.com/watch?v=yKPAwTF1SGY&t=270s) **Brain Structure and Sensory Processing** - The speaker outlines vertebrate anatomy—including the cerebellum, spinal cord, and three brain regions—and explains how these structures organize and process sensory data from sight, sound, smell, taste, and touch. - [00:07:46](https://www.youtube.com/watch?v=yKPAwTF1SGY&t=466s) **Frontal Cortex Integrates Experience** - The speaker explains how the brain’s frontal region coordinates sensory input and long‑term strategic thinking to decide actions, illustrating its remarkable ability to fuse memories of sights, sounds, smells, and emotions. - [00:10:57](https://www.youtube.com/watch?v=yKPAwTF1SGY&t=657s) **Enterprise Applications Feeding a Data Lake** - The speaker outlines how various business systems (CRM, HR, finance, legal) connect via a network to a central data layer, discussing the differing qualities and purposes of data lakes. - [00:15:02](https://www.youtube.com/watch?v=yKPAwTF1SGY&t=902s) **Turning Apps into MCP Services** - The speaker explains how converting existing applications into Model Context Protocol (MCP) services creates tool and data‑source interfaces that agents can orchestrate across a partitioned data lake, likening each application to a brain organ. ## Full Transcript
Today, artificial intelligence is everywhere. And if you're working in IT or you're a developer, uh
you've probably either already come into contact with AI, or you may very soon find yourself to be
involved in some projects uh, with AI. And I, I think it's very important that we take a minute and
understand that AI is, is a launching off uh, of the human brain. And so, if we take a moment to
think about where we've come from with AI uh, and look at the architecture and the plan and the
plan of the human brain,uh, it's going to give us some hints about how we need to evolve our IT
architecture to be more AI-ready uh, across our technology. And so,uh, you know, here's a really little
cartoon summary. Um, so when we think about AI today, we have the
internet, and the internet's got tons and tons of data. And, and so really where, you know,
where the current uh, GPT-driven large language models and the sort of new paradigm of, of AI that
we're in today has, has real heavy overlap with the internet. Because basically,
you know, the, the GPT models kind of swallow the internet and process it and then create these,
these large models uh, that have a really uh, good understanding of text and images and these, and
these new applications, okay? So, this first paradigm is sort of AI swallows the internet, right?
Um, and so, when we think about this, this, uh, this framing, this is not really useful
for us inside an organization. Um, when we're inside an organization,um, we care a lot
about specific things that are applicable to our organization. So, you know, the data that we
care about inside uh, an organization is very different than all the data that's on, that's in
the internet. Um, and when we think about our, our organization, you know, we have um, important, you know,
applications and, and maybe some of that is, is SaaS and maybe some of that are things that we've
developed internally. But we've got kind of this, this executive functioning layer of our
applications. And then of course, uh, we have networking uh, that kind of connects everything
together. Um, but the, the challenge that we have with this sort of existing IT infrastructure in AI
is that, you know, now we're, we're in a paradigm today where we're in this kind of like plus AI
paradigm. So, it's kinda like, this is what we've got. You know, we've got data and apps and SaaS and
networking, and we're in this sort of like now we're trying to have sort of AI swallow the
enterprise, right? So, AI swallowed the internet, and now we're trying to kind of jam AI into the, into
the existing, um, enterprise. And the, and the problem, the problem with this, okay, the problem that
with this is that we're seeing 90% plus failure of AI initiatives that
are happening in this style. Okay? And so, what we, what we really probably need to get
to is uh, a world where, we need to get
to a world where our AI, okay, is a plus, but we need to get to a world where, where
we've got a little bit more organization and separation of kind of what our, what our data,
what our data sources are and what our tools and executive capabilities are, and then how that's
going to interact uh, with AI. And of course, the goal here is to kind of reverse uh, what's happening
down here, you know, and let's try to get to um, 80% plus success rates in our AI
initiatives. So we've talked a little bit about where we came from and kind of where we're at. And
let's take a step back now and let's talk about, you know, what are we trying to mimic. You
know, when we're, when we're talking about artificial intelligence, you know, what we're
really trying to do is we're trying to mimic human intelligence. And so, all animals, okay, all
biological animals have a body plan. And a body plan is kind of a biologist's way of saying an
architecture, right? An IT person would say an architecture, biologist says body plan. So here we
have a, a brain and we have, you know, this main region in the top where most of the action is,
right, um, uh, which is the cerebrum. And then we've got this, this sort of midbrain area.
And we've got a, a little thing back here called the cerebellum, which is, which is Latin for uh, small
brain. Um, and then you've got this nerve cord, right? So, this is your nerve cord and it runs
down your backbrone, backbone, right? So, so, so vertebrates have their nerve cord running down a
backbone and invertebrates have their nerve cord not running down a backbone. But what we're, what we're
basically, this body plan is basically designed to sort of process data, right?
So we're processing data from the environment and we're sort of, um,
generating responses,okay? So we're getting data here, you know,
from all of our, when our body touches things and feels things, um, you know, we're getting data
coming in here through our ear. Uh, we're getting data uh, coming in here through our, um, olfactory area
with the nose and the mouth. And the reality is that your eye is a little bit down from here, but
basically, you're getting in data here from smelling and tasting. You're getting data from
hearing, which is auditory. You've got a big uh, you've got a big you know, data center coming in from, from optical
data when you're looking at things. And so, you've got all this data coming in and,
and the brain is basically organized into, into three parts, sort of three regions.
Um, you've got the, you've got the uh, the lower brain, you've got the
midbrain, and then you've got this upper, upper brain area, right? And so, in the, in the lower
brain,uh, what you're doing is you're processing really primitive data and really primitive
responses, right? So in the lower and a little bit into this midbrain area, we're monitoring the
temperature, we know if we're hot and cold, we know if we're, you know, we know if we're, if we're being uh, eh,
if, if something is hitting us or punching us or biting us. As we move up into the, into the
midbrain, so the lower brain is like is like really, um,
primitive. Uh, the midbrain starts to be about connectivity.
So the midbrain is handling a lot of data exchange; what gets ignored, what gets stored in
memory, what gets moved around um, to different parts of the brain. And if you look at this picture from
the top, most people know that there's a left and a right hemisphere to the brain. So we're like
bi, bilaterally symmetrical. You've got this left and right brain, and right here in the midbrain as
well, you've got this very important structure that connects and allows for communication across
those two hemispheres. So really midbrain is, is about connectivity, lower brain is kind of primitive
functioning. And then if you look at this, you know look, at look at all of this real estate that we
have. We've got a huge amount of real estate left over up here. Um, and if you go and, and look at the
literature,this, this really big region up here in the upper brain is oftentimes described as being
in like four parts, you know, so there's kind of like four parts of this thing. And this frontal
part, the frontal area is where we're really getting our executive functioning, okay?
So, this, this is really the pilot. So, when you're, when you get up and,
and whatever you're deciding to do next right now—if you decide to pause this video—it's this
frontal brain area that's integrating all, everything that you're experiencing and kind of deciding
what to do next. Right? We've got these, we've got these sensorial areas,okay, that deal with uh,
processing auditory uh, data that's coming in through our ears. We've got a big optical area towards the
back of the, of the upper part of the brain. And we've got this, and we've got these upper regions uh,
that allow us to, to, to, to integrate and do really long-term strategic thinking. So one of the great
things a brain does is, is really, you know,the, themillion, the million-dollar, uh, you know, thing that
the brain does that we can't,uh, that we, that we have a hard time figuring out how to
replicate is the brain is like really amazing at integration, right.
So it's really good at integrating. So, when you're asked about a memory of like, hey, you were at the
beach last summer. Um, when you, when you think about that moment, you may be remembering salt
smells, you may be remembering the sound of the waves, you might remember the taste of some really
great uh, crab cakes that you had while you were there. Um, and you may even remember stepping on a
sharp shell, you know, and cutting your foot and experiencing pain and having to go get stitches. But,
but the thing that's amazing about the brain, right, is this ability to, to integrate and kind of
know when to take,uh, when to take bits of all these different senses and, and all of this different data.
Now,the, the last thing I'll say about the brain that I think is, is, is, is foretelling about when we
think about artificial intelligence and the enterprise is the brain ignores almost everything
that it experiences. So, what's amazing is that we, we completely ignore about
99.8% of all the data that we, that we come into contact with. And so, if you think about your drive
to work this morning, okay, you're, you're not you know, that there were other cars on the road, but
your brain knows to ignore the, the sequence of colors of different cars, or the make and models of
different cars. Now, if you, if a Ferrari went zooming past you, you might remember that cuz
it's really, it's kinda really out of the ordinary. Um, but, but the brain is incredibly good at
ignoring, um, like almost everything that you're coming into contact with. And it's very good at
storing things that really stand out, that are, that are likely to be important in the future.
Okay. So now let's, let's talk a minute about where we're at with, uh, with, with our enterprise sort of
IT architecture. We're going to keep this super simple. Um, so we're going to keep this in three
categories, okay. We're going to talk about applications, okay? So, akli, applications
you know, are you know,are, are things that do things, right? So it's kind of like
executive functioning applications.Two, um, is data. So, all different types
of data that we're storing. And three will be network, okay?
So network, network again is kind of how we connect things, it's our communication plan. Okay.
So, when we, when we sort of start looking at our applications, let's put just a couple of these
over here, um, you know, maybe we have like a, a CRM type of a system. Uh, maybe we have a
human, a human resources information system. Maybe we have a sort of financial,
you know, financial accounting type of system here. And, you know, maybe we've got, uh, you know, maybe
we've got something that holds. that's sort of a legal system that holds contracts or other kinds
of documents that might be of a legal nature. So here, here are applications. Um, and then typically,
you know, we're going to have uh, some, some type of data layer, okay? And so I'm going to write the
word uh, data lake right here. And, you know, don't be offended by data lakes,
you know.Um, I know that uh, some data lakes are constructed uh, to where we feel like everything is
just kind of dumping into them,um, and other data lakes have, have been uh, carefully constructed so
that you've got some really nice logical layers, and they're really useful, right? But whether, whether
your data lake is a, is a swamp or whether it's really useful, the idea is that we've got
these applications, the applications themselves have, have data and have users and have executive
functioning,uh, and we typically connect them uh, and sorta dump data into, into something that might
look like a data lake.Now, in the past, when we're trying to uh, when we're trying to
do things across these applications, we end up in, in sort of a bad, what I would call a kind of a
star structure, right? So I've got my, you know, my CRM, my HR, my financial,
you know, my CLM. And, you know, this may have been, you know, this is something else. Um, but what we, what we
typically do is we use an API type of paradigm. So we use kind of an API
paradigm here. And we say, well, you know,the, the HR system, you know, needs to go grab some information
from that system, and the CLM system uh, needs to push something to the, to the financial system. And the s,
the financial system also needs to talk to this other system over here. And so we kind of end up
in this, we kind of end up in this, in this connected world where we're relying heavily on
APIs and we're relying really heavily on very structured, um, types of integrations that do very, very
specific things. So, when we're building this type of architecture, you know, we really can't
leave much to chance. If we leave much to chance over here, things start to break. Okay?
So, the, the, the big question is how do we move uh, from this paradigm where we've got
kind of uh, AI kind of being jammed into the picture,uh, into a paradigm where this type of
picture becomes more AI-ready. Okay? And, and so let's take a shot at that. The first thing that
we're going to need to do is we're going to need to, we're going to need to scotch uh, this API
as the, as the dominant interface, because we're going to need to introduce a new character into
our story. Uh, we're going to need two new characters. We're going to need a new
orchestration layer, and the orchestration layer is going to
spawn armies of AI agents. So, we're going
to have lots of, lots of different AI agents that are going to be able to be spawned out of
this orchestration layer. Now, what good is the orchestration layer right now if it's not
interconnected to uh, the rest of this world? And you know, what good is it if we're kind of up in this
type of a world where everything is kind of already connected in very specific ways?
Well, without, without interrupting how we're operating today, we can start to do some new things,
which is to make sure that each of these applications
gets an MCP
service. So, MCP service stands for Model Context Protocol. And if we, I'll just put an M
here. So for each of these applications that we're using and for the different
areas in our data lake, maybe we partition our data lake a little bit and um,and,
and chop it up and organize it so that these things can start to become MCP. Um, we can have
MCP hosts that are providing MPC service to our agents. Okay?
And so, when we, when we, when we add orchestration and when we transform our
applications into now being able to be MCP services,um, what the, what
that's doing is it's exposing these applications um, as basically sets of, of tools.
Tools is like, what can I do? And data sources, which is like what do I know? So,
tools are what can I do, data sources is what do I know about. Okay? And when we start to look at
having MCP services kind of distributed here and, and where we can communicate and, and
ask and go, go, go ask agents to go do specific actions across this
infrastructure, we're kinda eking in the direction of looking more like this picture, okay?
Because the, the ability to orchestrate, you can almost look at, you know, these different applications
kind of as different organs in the brain. So maybe the CRM is kind of like my ears and my auditory
center. My HRIS system might be kind of like my olfactory, you know, nose and mouth.
My financial and accounting system, you know, might start to look more like my upper kind of
strategic part of my brain. My CLM system, you know, might be, might be back here in the rear as part
of motor coordination or something. But, but ultimately, we start to have sort of organs who, who
are, which remain specialized. But by putting in the MCP and expressing themselves as tools and
data, and by breaking into this data lake and creating you know what we call an AI-ready
data layer, okay?
When we get to an AI-ready data layer instead of a data lake, and when we. and when we and we get
these executive functioning uh, applications able to interact with MCP, then we can send lots of
agents out, and these agents almost start to be like synapses. Um, synapses are the, are the,
are the structure that connect neurons in the brain. And so, what the, what the brain is
able to do is when you want to do something super complicated,your, your forebrain, this executive uh,
functioning frontal area in here—I'm going to try to use a different color—so when you're, when you
go to do something complicated, you're activating this integration in this, in this executive
functioning here in this forebrain. Well, what, what's going to happen is that's going to be, that's
going to be this orchestration layer over here. So when we want to do something complicated, we're
going to want to activate this orchestration. And then what we're going to do is we're going to
know, are we calling on hearing or are we calling on our olfactory senses. Are we
calling on, you know, some really higher-level strategic functioning? Is, is the, is the task needing
to call on, you know, motor coordination? And we can start to develop a very, a very strategic approach
to doing very complicated things. Now over here, that's what this architecture starts to look like
cuz you've got sort of your orchestration is kind of your frontal lobe. It can, it can, you know,
it can basically look at, in the agent framework, it can basically say, look, let, let's talk about goals
we're trying to achieve and then outcomes that are going to be acceptable. Okay? And
so, goals and outcomes start to kind of be more the way we work in the, in the human brain. We can
use the orchestration layer and we can light up, you know, we can light up all these agents, kind of
like synapses, to go out across all these different, um, functional executive functioning
parts of the, of the application layer. And that they can also, the agents can also—let's go back
to our agent color. The agents can also come down and jump into areas of
the data layer as well. And so, the, the goal of this, the goal of this conversation,uh,
you know, was to kinda use a little bit of a cartoon and have some fun.Um, but it is to remind
everybody that artificial intelligence is literally trying to replicate the kind of, the
kind of uh, incredibly integrated functioning that we have uh, in the human brain with biologic
intelligence. And so, if we take a look at the human brain and we look at how integrated it is,
we look at how compartmentalized it is,um, we look at,uh, you know, how organized it is, we look at th, its
compact nature, we look at the fact that it runs uh, on very low power, um, and, and
then we can start to see sort of where we, where we maybe want to go with the enterprise IT um, so
that we can be more AI-ready across the enterprise.