Interview by Jonathan Kersting
Is AI more than just chatbots — and are you leaving serious business value on the table by thinking it isn't?
Jonathan Kersting hosts TechVibe from the Pittsburgh Technology Council, sitting down with EY's Natalie Freedline and Aaron Rakes for a high-energy conversation about the real-world AI revolution happening right now in business.
Why you should hit play:
• The human side of AI — Why your people aren't being replaced, they're being redeployed, and the new skills every professional needs to stay relevant (hint: it's not coding, it's interrogation).
• AI beyond the chat box — Discover the "seven layers of the cake" and why chat is just one of them. Autonomous agents are already traversing ERPs, CRMs, and email systems — and they don't take coffee breaks.
• The security wake-up call — Non-human identities, silent segregation of duties, and why governance isn't optional when your agents have access to everything.
• Pittsburgh's AI moment — From $100 million in startup investments to local powerhouses like GreenCabbage and Gather AI, the Steel City is quietly becoming an AI stronghold.
• The "just start" philosophy — Why waiting for perfect data, perfect documentation, or the perfect tech stack is the most expensive mistake you can make right now.
Whether you're an AI skeptic, a business leader trying to see real ROI, or just curious about where this technology is heading, this conversation will change how you think about what's possible.
TechVibe is your front-row seat to all things innovation and tech across the Pittsburgh region. Press play and get ahead of the curve.
Transcript:
EY
===
[00:00:00] But I love it. It's a great time to be doing AI, talking about AI and I'll tell you today is the time to start if you haven't. I'm all dressed up, I got my makeup done, my gown on, and he looks right at me and he says, "You look like AI." It's Jonathan Kersting here with the Pittsburgh Technology Council with TechVibe. This is your front row seat to all things innovation and tech across the Pittsburgh region. And I don't think you can have a conversation this day and age without two letters, A and I, written in them.
It's just everywhere. And what I love about my job is I get to hang out with some of the coolest people, the smartest people, who are on the front lines of all kinds of tech, but especially with AI in this day and age. And today I have some great guests from EY hanging out with us. Do they still call it Ernst & Young or is it just EY these days?
We've rebranded. Rebranded. We're EY. Yeah. It's just EY. Okay. It's all- But we'll always be known as Ernst & Young, so yeah. I think so, yeah. It's like we're still known as the High-Tech Council, and we lost the High- Yeah ... literally 30-something years ago. Yeah. Yeah. It's kinda crazy, but that's the way it is.
But we have Natalie Friedline and Aaron Rakes hanging out with us today. [00:01:00] Very excited to talk to youse guys. Our conversation's gonna hit some really fun points that I think, our viewers are just gonna really appreciate, and I'm going to get to learn a ton as well, too. First off, I wanna learn about how you guys are using AI in everyday lives,
I wanna learn more about the skills that really matter. And the big one I cannot wait to dive into is AI more than just chatbots? 'Cause I know there's a whole other thing going on there as well, too. And you guys do a lot of work with ERPs, with manufacturers, and big corporations that are using ERP.
And how is AI kinda playing in that field? There's a lot to cover. Let's jump into it. I'll stop talking. So now let's start. Like first off, tell us about EY, since you were now EY, and Piat EY for quite a long time. I'm still stuck in 1995- Yeah. Yeah
with Ernst & Young, so EY, we're a global company, huge consulting arm, especially in the technology space. Me specifically, I've been with EY for about 18 years now. I started- Fantastic ... in technology audit. It looked a lot different back then. But now I focus more on ERP transformation, specifically in securing, [00:02:00] controlling, and governance.
So we do that across the globe, and we also do remediation optimization type work, too. Good lord. You stay busy- and what I love is that the global presence of EY sits very strongly here in Pittsburgh- Yeah ... which is a tremendous resource- Yes, and I'm sure-
to our ecosystem here ... you're familiar with our tech report that we do every year for the innovation work. Oh, yes. So it's a great view of the powerhouse that we have locally in technology- Ex- exactly ... especially in robotics and AI. Little plug, if you go out and read it, it just validates all of the great things that- 100%
Pittsburgh has to offer with investment. I look forward to it every year. I usually get to interview- ... the folks involved, which to me is always fun, and it's great to see what is driving technology here, introduce yourself, mister Yep Please. Aaron Rakus. Been at EY for 15 years. Always been in tech consulting.
I didn't always do AI, but I tell my kids I was doing AI way before it was cool. Nice. And I did it the hard way, through brute force, and AI's a very big topic. It is. It's broad, it's ubiquitous. But I'll talk a little bit today about what it is, what it's [00:03:00] not. Yeah. And I help organizations scale it across the enterprise and it can really go wall to wall if done correctly.
Exactly. I'll talk a little bit today about, what does it take to start. Hopefully not as scary as you think it is. Yeah. Tell some real stories about the work we're doing. But I love it. It's a great time to be doing AI, talking about AI and I'll tell you today is the time to start if you haven't.
I think we're now at this really cool inflection point where it's not just doing it to do it, but actually seeing the business value from it, which is where UI comes into place. Because people who are investing, companies that are investing lots of resources and money into this wanna make sure they're getting that return on investment.
And now we're at that point where we're really seeing that really starting to happen. And what I find exciting is you've been, you said you've been kinda like the OG of AI in so many ways. Yeah. When I actually do my research, I look back through old Pittsburgh Technology Council magazines from the early '90s, late '80s, there was talk of AI type stuff, you know- Yeah
like 30, 40 years ago. But now it's in the past few years where things have really started to come to where we all- we're all talking about it like we know what it is. Yeah. It is certainly [00:04:00] mainstream and it, to one extent, it's almost like a commodity. Access to tools they're all over the place.
It's crazy, isn't it? But i- it's very easy to do it incorrectly and, we're gonna talk a little bit today about how to do it right. Let's jump into it. What I wanted to think about is what tools do you guys use?
We developed a tool, ey.ai. We use it for risk, for ERP, so you know, and we've been doing this for 10 years, right?
It's just looked different. And we have benchmarking, we have capital, we have IP And we have enablers, but now the way we can use it at scale and how much data you can churn through and how you can- Yeah ... just tell it- ... to intelligently do what you would do manually it's pretty, it's crazy.
Do you ever feel guilty sometimes- Yes. ... where it's I'm like, "Oh, I don't mean to ask you to do this." Yeah. But this would normally take me a whole- Like picture editing ... weekend of going through things. Yeah. I know it'll take you 12 seconds, but would you please? Yeah. Yeah. I'm also pretty apologetic when I'm the, when I'm putting my asks in there.
Yeah, they say that t- puts too much power in it when you say thank you. E- exactly, yeah. It was [00:05:00] running through too much- Yeah ... churning through too much. I gotta keep it in its place. Yeah. Do this for me now. I'm not saying thank you, but thank you. Yes. In case you take over the world, I want you to know- Yeah
I was the nice guy. Oh my goodness, that, that's crazy. So Aaron, how should we start off with? Trying to think what's a good way for us to jump in- Yeah ... around like how, like we're seeing all these tools, other businesses are seeing all these tools. Everyone wants to say they're doing something with AI.
Lots of people are now, have been experimenting. They are finding things that are working. They probably have learned many lessons about what you can and can't do. Yeah. But as the technology keeps strengthening, it's just what you couldn't do six months ago is now easy. That's why it's like it's hard for me, like where do we start?
Yeah. Why don't we start actually on, on the human side? Like- Let's do it ... what's the role of the human now? Okay. Some folks are thinking are, is human capital going away, right? Yeah. Are we gonna have a shorter work week? Yeah. But we're seeing quite the opposite. We're seeing a lot of opportunity to redeploy human capital and to upskill.
So I do a lot of insight sessions, talks like this and also activate a roadmap and implementation of AI. [00:06:00] Excellent. And I typically have three questions. One is what's the role of my humans in this AI world? Okay. Two is how do I measure my people, right? What's the measure- Very cool, yeah
right? What's the definition of success for people? And the third one is a very important one, is around the skills that people need, AI upskill. And maybe I'll start there first because it- That's a- I love the way you framed this out- Yeah ... 'cause I think that just gets us right on a track.
Okay, cool. Yeah. So it breaks my heart when someone uploads a document to ChatGPT or whatever and they say, "It didn't do a good job." I'm like that's because there's a gap in AI literacy." And I don't mean that in a demeaning way- Of course ... just that we need to learn how to talk to AI, how to use it, and how to engage, and we shouldn't fight it.
We need to work alongside it. Exactly. And we need to embrace it. It has become my official team of interns- Yeah ... at the tech council- As it should, yeah ... as far as I'm concerned, and they all treat me very nicely. Yes. You've trained them quite well. They don't even talk behind my back.
It's wonderful. Yes. At least that I know of, so yes. Yeah. Yeah. But literacy is a big part of being successful with AI, [00:07:00] I think at all levels. We'll use it differently, but we need to know how to talk and engage and when's the right time to, to prompt it. The other is just the role of humans, and this goes back to successful versus unsuccessful implementations.
We should assume that our humans are escalating and collaborating and managing exceptions, but they shouldn't be in charge of the process anymore, okay? AI is now built into the process end to end. And I want my humans to just manage the exceptions, interrogate it. But some times you think, "Oh, my humans need to be leading the charge."
And AI has advanced quite a bit in the past few years, so that's really the role of humans. And then the third is just human measurement. The old way was my best employee can crank through all these invoices or- ... reconcile this data, so it's almost just volume. Yeah. Okay. Yeah. But now it's really about creativity and innovation or, finding areas for improvement or waste or just great initiatives.
That's a great point that you brought up- Yeah ... 'cause I think that's where I think the magic happens is when you have the creativity and ingenuity of [00:08:00] the human mind- Yeah ... with the monotony that an AI bot can bring to you. Yeah. So you can be creative, but not how to do the steps- Yeah ... to do the thing, but know you want that done, and the AI knows how to do that, but the AI wouldn't think of that by itself.
And that's where I feel like that's having the human in the loop. Exactly. Yeah. Yeah. The humans are def- still definitely accountable for the outcomes. Exactly. There's definitely some emails that I've had drafted and I'm like, "Oh, I would not say that."
And Green made such, raised such a
great point. So we have associates that work for us and help us do work and projects. But I know the associates are using AI to, to create content, and as they should. We need to be more productive. Yes. So the new role of that associate needs to be around interrogation of the AI because I've told them before, "If you're going to show me something, be prepared to answer any question I ask you.
Any slide, I'm gonna ask." And show sources. Yeah. So as an associate- Where did you find this from? You need to be able to- Yes ... defend anything you put in front of me that came from AI. So I think it's just an evolution of skills that we're te- [00:09:00] teaching our associates to learn. Yeah. But just how to work, how to interrogate, and almost how to coach almost bilaterally AI whether we're coaching AI or AI's coaching us back.
Interesting. Yeah. And if you think about the control over that too- Yeah ... it's what do we have in place? To mitigate or to catch anything, right? Are we just relying on that data? And when you think about from an audit perspective, you still need traceability. 100% you still need to know where the data came from and to be able to show that.
But we will think about auditing AI differently, but it's gonna be more... Whereas today it's, what went wrong? Let's detect it. Let's show it. In the future, it will be more preventative, and it will be, tell me why you think this model will produce future- ... correct information. So it's definitely a new thought process and how we're gonna handle.
And h- if will an unreported exception go, and we- ... nobody knows it? How are we gonna de- How are you gonna- ... how are we gonna mitigate that? How are you gonna trace that down? Yeah. I think what's really interesting as I think this through, and I think about EY [00:10:00] and just your whole, the background of what EI does with auditing and how important that is to AI.
Yeah. And the fact that as you're using it, you're really making sure there's a responsible use of it, to the point where it's wow, okay. So that's where I think there's this special sauce that comes together when you have a firm like an EY that, that's really putting all their guns into AI and being able to leverage it, but doing it in this way that, there, there's guardrails and there's traceability and proof and all that kind of thing.
So it really brings the best- Yeah ... out of the technology. And we're working on both sides. I know- Okay ... I help audit teams. We have conversations and, they bring to us what the new regulatory landscapes looks like. How are, what are they gonna have to report on? How are they getting comfortable over it?
So we're teaming across the lines where we can and- ... making sure our clients get, the best of both worlds. Yeah. Yeah. At some point, though, do you ever feel like, I feel like this sometimes I've be- I might become too reliant on the technology, but I say to myself, "This is what we have as a tool."
It's not going away, so why would I [00:11:00] not use it? Should I be training my brain in other things? I don't know. What are your thoughts on that? Yeah. I think to Aaron's point I, as I got more comfortable- ... I find myself in Copilot all day. Yeah. I have it up all day.
Yeah. Whereas before it was like, oh, if I thought of it- Exactly ... I would utilize it. Yeah. But now it's just, it's starting to become a piece of my every day. It's part of your every day. Yes. Yeah. Yeah. It's- and the way it's being integrated into everything, it's becoming more of our every day, and sometimes we may not even fully realize it- Yeah
Which is interesting. Which makes me think about there's just a whole new set of skillsets. The first thing that I really picked up on where this question started forming in my head was, Aaron, when you were mentioning that the fact that we need your people to interrogate- Yep So you, the idea is you're learning how to prompt.
You're learning- ... how to manage something and interrogate something. That's a skill set right there. Yeah. That's a, it's a way of thinking- It is. It is ... to do that, and that's just one of many skills that we're all gonna have to develop as business just changes- Yeah ... right now. I think, as you mentioned and I mentioned earlier, just critical thinking and the interrogation is going to be so important that, the model, if done right, is grounded [00:12:00] and is sourced truth.
And one of the things I would say is AI, if built correctly, there's no guessing, okay? Everything is cited, fact-driven. When you build some of these things like RAG architecture and DAG architecture- Okay ... that means that I ask a question, it understands the intent. It goes to fetch the information.
It's cited. It loops back up to humans, kinda in this loop. But you also have the ability to interrogate and, as you said, prompt and follow up with questions. So it- it's a whole new skill set because we are benefiting from the productivity of AI. One of the things I say a lot is it's compressing time, and that is such a- Is that why the years keep going by faster- yeah ... and faster? Come on, now I know. I know. Okay. Thank you, AI. And it's such a important asset, time. No one has enough time and AI's helped us in all areas compress time, whether it's using AI to actually implement AI, and so sometimes we say using AI for AI. I know. It's an infinite loop, but we should be using AI to help us- write code, to, to design. It's actually great, it's starting to open our eyes up to new ways of working, new ways of thinking. So it's really exciting- [00:13:00] Yeah ... to see how we're using it. Yeah, 'cause it's exciting times right now. That's why I just love these conversations. So many of us think of AI as working through a chatbot. Like we, we- ... we log on to chat.
We type our thing in. But there's more ways that AI works than just the traditional chatbots. Yeah. What am I missing here? Where else is AI being deployed where I'm not, like- Sure ... typing in "Hey, I need this report by Friday"? And it says, "I'll give it to you in 12 seconds." Exactly. Yeah. Which is a very real example, but, AI went through a boom when ChatGPT became- Yeah
Democratized and was using it. That's when it came out to the public- Exactly ... really, 'cause businesses and industries were using it behind the scenes to automate some things- Sure ... but you never saw it as a consumer or a person, so you, I think you raise a great point. Yeah.
So keep going. Yeah. Yeah, because AI was statistical models- ... machine learning algorithms, pattern recognition, and then all of a sudden, boom, we had large language models and GPT, which was creation of content, and we were serving it up through chat. And so a lot of folks said, "When I hear AI, I think of chat."
But you're leaving a lot on the table if you're thinking AI is just chat because what that means is that as a human [00:14:00] I'm the bottleneck, right? I'm the one that's typing the question- ... aI is responding, but we don't want that. We want AI working all the time. So sometimes I'll refer to AI as- Not even a little break?
We don't want- A coffee break or something like that? You'll have time for that. Okay. But I refer to it as seven layers of the cake. Okay. And chat is just one of seven. There, there's source data, there's systems of record like your ERP and CRMs. Okay. AI is a great way to activate those, so it's actually a system in motion, not just a system of record.
And so AI should be traversing across systems. And the way that agents work today, you've probably heard agents a lot is that they have identities and they have access to tools. And so what that means is that business events happen. As a human, I don't need to say, "Go do this," through a chat. Instead, the agents are triggered, and they have access to tools to go do the work for us.
And when they do the work for us, they're accessing our ERP, our folders, our emails, and they're traversing across all these different systems. I should only be [00:15:00] chatting by exception, right? If I want to interrogate, as we talked about- ... or if I want to- I like it. Yeah ... send a- Okay
one-off command or escalate. So when I think of AI and chat, I actually think of it more as the exception, not the rule. I like the way you bring that up 'cause it, it makes me think about how the next level above chatting is building your own agents. I've built a couple- Yeah ... myself just for fun.
Okay, cool. And it's kinda like... And it's pretty easy to do. It's basically drag and drop. And I'm like, "I think I might have to buy a vibe coding T-shirt or something like that- There you go. ... and be like all the cool kids out there. But any- I, I digress, but I think the idea that as we move to these next levels, we're gonna be building these agents for ourselves to get through our tasks.
So it is working behind the- Yeah ... I'm, it- it's kinda like the ultimate prompt to do a lot of this for, until I tell you to stop or something like that. Yeah. So that gets really exciting to me 'cause I c- I start thinking about the things that I could do so my Monday mornings are a lot faster- when it comes to posting a radio show, to putting something online- ... and all that type of stuff. And to know that these agents have access to our CRMs and can go in there- ... and update a file. I think about when we do score [00:16:00] carding at the tech council, we like to keep track- Yeah ... of all of our interactions with our members.
What a great way to, if we could get a bot that would go in and .. just be doing that all the time. Yeah. It's exciting. Yeah. Yeah, and we start to talk about identities and non-human identities. Yeah. This is where I start to put my risk hat on. Okay. Yeah, ex- Yeah, you're like, "Wait a second."
My mind is just... Yeah wait a minute. So- Is that... I think the Ernst & Young in you is coming out, right? Yes. It's been embedded in me. But it's- It's not EY now, man ... it's that... Yeah. It's Ernst & Young when I'm thinking about this. Yeah. But it's... When we start thinking about these non-human identities- and the ownership- ... and the governance over them, we have to be careful. And the liability too. The liability. Yeah. I mean- I don't know if you've ever heard of silent segregation of duties. Oh. So- No. Okay ... we're giving agents all of this access- Yeah. ... Looking through access to ERP, CRM We have to be careful.
We have to control and manage and monitor what they're getting access to because, agents talking to agents, they could be creating, approving, and posting Right that's a segregation of duty [00:17:00] issue. So- Gotcha ... it's gonna be very important- ... as we think about non-human identities,
We have to give them broad permissions in some scenarios- ... so they're, they work. But- ... how are we managing that so that we're not- ... inadvertently creating additional risk? And this is why people come and talk to professionals like at EY- ... so it gets done right, 'cause it's something you don't wanna you know, be in Lindy building Yeah.
Be like, "Hey, I think this is gonna work." All of a sudden they're like, "Wait a second. That talked to a bot that didn't have any security on it." "Now all my data is out there." Yeah. And so security has to become literally the most top of mind thing at that point. Absolutely. But it goes to show you just still how Wild West some of this really is.
But we are seeing this gel in certain ways as we're talking now, so to me, that's why it's just exciting. I just, the future keeps getting a little bit brighter as to what can be done. Yeah. Yeah. Ear- earlier you brought up two things that you're doing about you could build your own agents and maybe you integrate with enterprise systems.
We're also seeing some challenges with that because we have what we call diffusion of effort. [00:18:00] Yeah. And what that means is that some organizations say, "I have dozens of AI models. I have this, that, and the other. But I-" I'm not feeling the benefit, right? I'm not seeing the productivity on my P&L, or we're not doing different things.
And so our recommendation is pick a few frontier models- Okay ... pick a few enterprise applications. You want people engaged, and you wanna encourage that culture of, innovation- ... let people play. But you have to also put boundaries around that because y- you may have seen in the news there's things called tokens.
Tokens is another way to say how much money did you spend on those AI models? And if not done correctly, folks are just exploring and playing and then racking up a lot of tokens. Exactly, and that tokens is cold hard cash at the end- It is. It is ... of the day for somebody out there. Yeah. Which is why so many of these AI companies are still hemorrhaging a lot of money right now- Yeah
'cause it's like someone's paying for the electricity- Exactly ... to make that happen. So you need to find a balance between allowing folks to, play around and innovate, but also have maybe a core team- Exactly ... to focus on enterprise solutions. 'Cause I probably don't want an [00:19:00] individual connecting to my CRM because they probably don't have access to what they're trying to do anyway.
So you need to just be careful and find the right balance around that. Yeah. I, everything I think obviously comes down to some sort of a balance- Yeah ... but I think it comes very easy for us to go very gung-ho when we see some results. "Hey, we're all in." It's like, "Hey, yeah. Keep it down." Yeah. "We can figure this all out."
Interesting. So we've been talking about like ERPs and CRMs and things like that. And obviously that's like the soul of a company. That's its truth. Yeah. And and to have, AI and bots and every- connecting into that and being able to access your data, manipulate your data, it's creating lots of really cool opportunities for companies to actually get better data and to get better served customers 'cause they can do things.
Like what is going on between AI and ERPs? Is one putting one out of business? What what's going on when it comes to this? ERP is not dead. Okay. Contrary to... Maybe not yet. Okay. Maybe, some point in my lifetime or my seven-year-old's lifetime, but, right now, ERP it's changing.
It's evolving where it will be the [00:20:00] future, context engine for AI, we'll say. Like this is gonna be where it's- Oh, interesting. Okay ... as before, it was the core of everything, right? Everyone's spending a lot of money investing. It, that's gonna evolve, and it's gonna be more of the brain more of the engine where we take and we rely upon and we trust the data coming from it.
It's our s- kind of our source of truth. But who knows what that will look like, if it will be outside, inside. Exactly. And, w- the future hopefully will be able to take what's in that ERP, and if we need to move it, we don't need to just throw it away and start over. We can evolve and we can- Yeah
build the context con- to kinda continuously improve. I get fascinated with this 'cause I just- ... I just think about these massive companies. Think like a U.S. Steel or something like that out there, and you're looking at how their business is run, and When you can fundamentally really start changing that, and provide more insights, provide more data, like it just becomes a whole new world out there.
And you get to have so much fun 'cause you're actually [00:21:00] working on this and working your clients through a lot of this as well too. So you come to work every day like, "I got some fun challenges today." The future, so hard to predict. Yeah. It, I feel every couple months there's something dramatically different- or a new level of something where- Every week almost ... it's getting better. Every week. Yeah. What has you most excited on a business side of what's happening with AI right now, where businesses can really start seeing value? Yeah. I think for me, one of the things that is intriguing me is just the pace of innovation right now.
Yeah. You think about these old monolithic systems, could be an ERP, could be something else, but the pace of innovation was every few years there was an upgrade. And my prior life, I was doing big system implementations. Okay. The client, we were on old system for 10 years, end of life, moved to the new one, and it felt like a lot of the things they did were similar.
Not a true step change. But what excites me is the pace of innovation, week over week, new models, new capabilities, just helping businesses better serve their customers and suppliers, [00:22:00] better serve their own employees for just a better work-life experience. I'm really excited about what AI's doing to transform companies just to better s- help them serve and offer the products and services that they deliver.
The other interesting thing, and these are real things we're doing with clients, is just around better predictability of customer, whether it's customer churn or better negotiations with suppliers. I think it's just about better hygiene of the way we do business, because AI is really helping us kinda see around the corner better, right?
Yeah. I have fewer blind spots. Yeah. We'll have no excuse I think by one or two years from now to say, "I didn't see that coming." Interesting. 'Cause we absolutely should. It's having the right dashboard with all the little mirrors so you can see what's going on. Reminds me of like the transition to the cloud we saw- Oh, that's a great parallel ... 10- I love that ... plus years ago. So I had a funny memory as I was thinking about this. We did a CIO roundtable probably in like- ... 2009. I was there to pour everyone's coffee. Okay. But I was listening and engaging- Really?
Okay ... in some of the conversation, and I recall two CIOs, I, locally I believe [00:23:00] they were from a bank, I don't really recall, but they were basically saying, "We're never going to the cloud." Really? "This will never happen. Our data is too sensitive." And I'd love to sit down and see- Yeah
Where they're at. But it's funny how it's kinda the same paradigm shift right now. It's we're gonna change the way we work- ... we're gonna change the way we operate, but it's gonna be built on this trust, and the trust begins with this experimenting- ... testing- ... and curiosity.
And I think that's what I'm most excited about, is seeing, people will come to me and say, "Look what this agent I built. This agent's talking to another agent, and this is how I'm making my job better," and it's allowing us to be entrepreneurs in our own- in managing ourselves Yeah, exactly And it's really exciting to me to see how, especially the younger generation is taking this and really running with it.
I just feel like the future's wide open, and that there's so many- ... cool possibilities- Yeah ... because of this. And even locally, I know we have two extremely impressive [00:24:00] companies that are in our Entrepreneur of the Year program. Okay. We do an awards ceremony every year.
They are finalists. GreenCabbage is one. Okay. A local Pittsburgh company. Yeah. They do procurement intelligence. . And also Gather AI, inventory intelligence. We know those guys at Gather. Yeah. Yeah. Awesome. Again, powerhouses out of Pittsburgh. It's a real testament to who we are and, we have the universities and just this tech powerhouse locally. Something that I think that really got , my brain spinning, and we're actually seeing it happen here in Pittsburgh to what you're talking about with some of the startups.
At Huntington, they had their economic outlook breakfast a few weeks ago, and one of the big insights was something like $850 or so billion being spent this year on AI. Like- ... that's a lot of money, and you look at where that's going, servers, grid all the stuff that goes into, to making all that happen.
And I look at what happened here in Pittsburgh during that same time period. We had two startups- ... rack in $100 million worth of investment with Sooth and Greyswan, right? And like Sooth, this isn't... they're very nascent. $50 million, nascent in [00:25:00] Pittsburgh. That doesn't happen very often, right?
Same thing with Greyswan. Grey- the cool thing about Greyswan is you have Ziko Kolter there on the board of- ... OpenAI, and then also runs AI at CMU, and it's all about security around AI. So you're seeing that the money's going into these things- Yes ... that are gonna make AI better, that's gonna make it safer.
I get really excited about that, 'cause when I see that happening in Pittsburgh, that's super cool. And I'm not trying to undersell Pittsburgh, but I just say that Pittsburgh's not that kind of a town very much. And that goes to show that with our chops in AI, I think we could be seeing a lot more of these GreenCabbages and Sooths and things like that- Yes.
Absolutely ... that are just taking the rest of our region forward. And not only that, solving really tough problems. These are companies that are really trying to build something. I got nothing against people that make dating apps and stuff like that. Whatever, that's fine. But when I see someone that's can make AI more secure- ... that's got... It's a little cooler to me, but not, this is me. I digress. Yeah. So that's where my brain goes on. Yeah. I'm excited. Yeah. Very excited. So so much fun here today. What have we not covered? I'm always wondering, like I get [00:26:00] so excited- Yeah ... and I say to myself, "Did we miss something that I could've totally talked to you guys about?"
What I would say is Now is the time to start. Yeah. I think organizations are wondering I probably don't have the perfect data. I don't have the model-" You never will. No. Yeah. Yeah. "And I don't have the perfect tech stack, and my documentation, I need some time for that." It turns out, start.
Yeah. It turns out, start. I'll tell you, from a data perspective, you don't need perfect data. Okay. You need data that's focused on the areas that you're using AI. Okay. I have an acronym called LUMA, which is light use case driven, it's modular, and it's AI-ready data. Man, you got it going on. I know. Jeez, Aaron.
And like what that means is just a few thin connectors makes a nice light integrated AI system. I like that. Yeah. Your data is better than you think. It's more ready than you think. Just think about, we can take this entire call, or I take calls at work and load it into an AI system. It can tell me my action items.
It's really good with unstructured data, so if folks think they need really organized data- Yeah ... you need [00:27:00] better governance policies. I like that. Yeah. But the data's probably ready for a lot of AI. But at the same time, I've had issues where, we just leave a feature, in the backlog because the data, just needed some more time to harden.
That's fine. The other thing around documentation, I'm less concerned about the current state. I'm more concerned where you wanna go. Okay. So organizations- ... say, "I don't have, click by click details on what I do today," and that's totally fine because if we focus too much on AI in the current state, we're gonna have an, a old process with AI.
And what we wanna do is raise the ceiling, one of the questions I say is, "If you had AI on the day you started this process, what would it look like?" It'd be drastically different. That makes total sense. You would have different approvals. Your data would be sourced differently. So let's not worry about not having documentation.
Let's worry about the outcome and- I like your approach. Yeah. Just makes me, makes it feel more approachable. Yeah. And I say that it's gonna be a learning process. Things get messed up along the way. But that's just part of it. You need to be experimental. Yeah. You need to learn, right? I- if you don't learn, you're not doing it right, I like it. We've said that, too. We've built things that we're didn't end up using, but [00:28:00] we said, "You know what? It's not throwaway," because we learned. 100%, yeah. We learned a ton. We learned about our data. We learned about our process. We learned that we were probably trying to put AI on top of our manual process.
Yeah. So we said no. This is not gonna work." Yeah. "We need to start over, throw it out the window." And change is hard, I get it. It's hard for me to change, too, but- now that we've opened and accepted, I feel like we're all in a better place. Yeah. Great insights. Oh my goodness, I have learned so much today.
This is something where I'm gonna encourage everybody out there, if you haven't- Start dabbling. Yeah. Start thinking about where you think it can make the most impact- ... and don't worry about how nice the data is. ... It's 'cause everyone's, no one's data is perfect. It's as simple as that.
So that's great insight. So much to walk away with today. So glad I got to pick your brains here- Thanks ... on all this type of stuff. Makes for great conversations. It's why we love hanging out with folks like you here in the Huntington Bank Studios. Yeah. Oh my goodness, Aaron, Natalie, you guys are the best.
Thank you for being here with me today. Thank you. Yeah, thanks for having us. It was a fun [00:29:00] time. Great stuff. In case you forgot, this is Jonathan Kersting with the Pittsburgh Tech Council, just loving doing Tech Vibe. Who cannot love having conversations like this? I cannot wait to see you on the next one.