Join Andreas Senie and special guest Charli AI Intel’s CEO Kevin Collins as they dive deep into the state of Artificial Intelligence, AI’s place in Commercial Real Estate, and talk about how AI can be used and misused in the workplace.
Kevin has more than 30 years of experience and extensive knowledge in artificial intelligence and machine learning, both as an entrepreneur and a corporate executive. As Founder and CEO at Charli AI, he oversees a team of expert scientists that are pushing the boundaries of innovation in the AI-driven intelligent content management space.
Before founding Charli AI, Kevin was CEO and Co-Founder at Bit Stew Systems, a data intelligence platform, which was acquired by GE Digital for its AI and ML capabilities in 2016. Prior to his time in Silicon Valley, Kevin worked in the high-tech networking and security field, and led technology firms specializing in cryptography, public key infrastructures and high-performance and scalable networks. As a serial entrepreneur, Kevin is passionate about sharing his expertise in building and growing successful startups. Before founding Charli AI, Kevin was CEO and Co-Founder at Bit Stew Systems, a data intelligence platform, which was acquired by GE Digital for its AI and ML capabilities in 2016. Prior to his time in Silicon Valley, Kevin worked in the high-tech networking and security field, and led technology firms specializing in cryptography, public key infrastructures and high-performance and scalable networks.
Sector Interviews are bonus episodes of CRECo.ai Real Estate Roundtable - Your comprehensive all-in-one view of what's happening across the real estate industry -- straight from some of the industry's earliest technology adopters and foremost experts.
This bonus episode of the RoundTable included:
Host, Founder CRECollaborative
Non Profit & For Profit Business Technology Transformation Champion
CRETech Thought Leader, Founder & Brokerage Owner
ABOUT THE ROUNDTABLE:
Your comprehensive all-in-one view of what's happening across the real estate industry -- straight from some of the industry's earliest technology adopters and foremost experts. Join us live at 6 PM EST on the 1st Thursday of each month, across all major social media channels and wherever you get your podcasts. This three-part show consists of:
Part I: Introductions and what's new for each panelist and the business sector
Part II: Sector Focus on the past month's most prominent news and paradigm shifts
Part III: What does all this mean for real estate businesses, and what you can do for the next 30 days
Join us as we dive deep into Technology, Marketing, Capital, Construction & Cyber Security trends affecting your business.
#datadrivenbusiness #businessmanagement #commercialrealestate
#podcast #AI #business
Sect Int_Kevin Collins_Transcription
[00:00:00] Andreas Senie: Welcome back to the AI Real Estate Round Table for this expert sector. Interview with another then Kevin Collins, founder and c e o of Charlie ai. Leading intelligent content services platform provider that is driving the transformation of human decision making. I'm Andrea Seti, technology growth strategist, founder of C R E, collaborative, brokerage owner at EAC Properties, and the host of the Kreo AI Roundtable.
You're all in one comprehensive view of what's happening across the real estate industry, straight from some of the industry's earliest technology adopters and foremost experts. And I could not be more excited to be joined today by Charlie's ai. Kevin, why isn't it named Kevin by Kevin, the c e o and co-founder at Pits two Systems, now c e o at Charlie ai.
Welcome. Hi, Andrea. Thanks for having me here. So, why isn't it kevin.ai first and foremost? Uh, that wouldn't go over
[00:01:12] Kevin Collins: well. Uh, there is, uh, a story behind the, the name of Charlie. Uh, And it really comes down to that we're all time poor. Uh, we need to clone ourselves essentially, uh, get all the stuff, all the work done that we don't want to do.
And Charlie was really a play on chief of staff. So we all need that chief of staff to do all that stuff. We just don't
[00:01:35] Andreas Senie: want to do anymore. I love that. Well, and before we get into to the AI of today, let, let's talk a little more, and I still need a chief of staff, so that, that's a good one. Let's talk a little more about, uh, uh, bit STU Systems and your background in AI and ML capabilities two, 2016 and before that cryptography, security, public infrastructure.
I mean, you've been a, you've been a brown. Why ai? How'd you end up where you are today?
[00:02:04] Kevin Collins: Yeah, I, it's been, uh, quite a journey over the past 30 years and I was intrigued by AI back in my high school days. Now that back then was more vision, uh, than reality. The technology has certainly exploded, uh, recently, but that whole journey was going through many different iterations.
Of what AI is gonna look like. Uh, and at TU you mentioned Tu that was a fantastic experience because we were on the industrial Internet of things, trying to take massive amounts of data from internet devices, uh, on the industrial side, make sense of it, and get insights. And we had to apply AI back then.
So it was very heavily involved in ai.
[00:02:53] Andreas Senie: Well, so there's, and there are different types of ai, right? So Mo Yes. This audience, global audience, top 20% globally for ranked shows. Um, there are different types of AI for different reasons, and as you just said it, AI is kind of exploded on the scene. Chat, G p T and then chat G P T four.
You know, they, they were this paradigm shift, and I always, I always relate, COVID was the paradigm shift that forced a good old fashioned industry, like commercial real estate to actually get online. The die hards. The people who have been doing it for 30, 40 years said, we'll never change. We're just gonna do what we're doing.
And they didn't have to. And then Covid hit and everybody was stuck at home and every one of them got online. And this was a huge moment in our industry. Cut to today chat. G B T is exploding. Every founder is revisiting their business plans. Meanwhile, chat, G B T is, I understand it. Hallucinates their words, not mine.
I mean, what has happened in AI in your, as you said, since high school to today, industrial to today. Why is it, why is today the time? Why does it explode?
[00:04:02] Kevin Collins: Uh, the big thing that it's really exploded, and I think this is where CHATT really. Had innovation is that they made the AI technology accessible to everybody.
Uh, before that it was a lot of scientists, it was a lot of, uh, deep learning people that played with these scientific tools and capabilities. They were fascinated and it took a lot of engineering effort to do something with it. Whereas now everybody gets access to it. Everybody can engage. With the AI and get answers back.
And I think that's really the innovation is the ability to engage, get accessible or get accessible ai uh, and you don't have to be a rocket scientist to use it right now. No. The open AI folks, and a lot of other industry experts say democratization of ai, but it's really just the accessibility of it.
It's easy to go in and have a chat now and get an answer. And that's just fascinated the world
[00:05:10] Andreas Senie: and, and whether the answer's right or wrong or, or now as you said it, you can, you can start to investigate it a bit more. I've seen some really cool ideas from the thought of it, like, give it to your four year old on a little robot on a table and your, and it will get smarter as your child does.
That scares the crap out of me, but I think it's. It's something as accessibility is king. Um, and, and actually that was a great point. So here you are 10 years ago, maybe 20 years ago, you've, you're using AI in, in the industrial worlds, connecting it to the internet. So this is not new technology, it's just readily available to the consumer.
[00:05:47] Kevin Collins: it is. I mean, AI has evolved quite a bit over the last few years with. Things like transformer technology, which is, is the science behind the models that you're seeing today. Uh, and that's obviously a leap from the technology we were using back in the bits, two days and definitely a leap from decades ago.
But these new models and approaches of machines that learn is gonna advance rapidly
[00:06:19] Andreas Senie: over the next few years. Ab, so, so let's, let's go back to the go. Right. Go. AI is applied to go, and it's gonna play itself for the chess game. There's a few famous examples of this. Yes. Where do we break from machine learning?
This something just running against itself to get you an answer. Compute computational power, I think is a layman's way to say that. Yeah. To ai.
[00:06:43] Kevin Collins: Well, the, there's, uh, a bit of a. A hype and still a Hollywood romance around what AI is.
[00:06:51] Andreas Senie: And then there's the reality. Okay? The reality
[00:06:55] Kevin Collins: is we're still scratching the surface of this and.
You know, there's no
[00:06:59] Andreas Senie: critical thinking that, you know,
[00:07:03] Kevin Collins: these machines aren't thinking for themselves right now. That's not where the technology
[00:07:07] Andreas Senie: is. Um, you can't teach a common sense or you can't program common sense. Yeah. Yeah. Or
[00:07:13] Kevin Collins: they, where, where we really, uh, within our own team and our own science and research efforts around this is that we really want the machines get to get to the point where humans are brilliant at critical thinking.
We can't go though through those critical thinking elements, those what if scenarios in our heads. Um, this is where we want the AI to eventually get to, but right now it's large language models and one, and I unfortunately know a bit too much about the limitations, including the hallucinations that you
[00:07:47] Andreas Senie: alluded to, alluded,
[00:07:49] Kevin Collins: uh, and they're significant.
Uh, the reason they're significant is that the large language models that we're seeing today, Are fantastic at predicting the words that have to go together
[00:08:01] Andreas Senie: in order. Does it sound good? Y yeah. They're, they're, they're really
[00:08:04] Kevin Collins: good at knowing the language and knowing how to give you something that's very coherent and, and fluent and you go, wow, that's fantastic.
And I believe it. I don't care what you're saying, but you said it so well. I believe it.
[00:08:18] Andreas Senie: Uh, but that's kind of, that's, that's, that's, let me reframe that. So that's like running into someone who isn't very well educated, but happens to have a, a slight grasp on some larger words that sound good together.
And then they try to string them together. And, and I, and with a person, I would go, wait a minute, that's not what that means. But with the AI, people are going, okay. Yeah.
[00:08:44] Kevin Collins: Now the reality is that we are getting to the point now of challenging the ai so you get beyond the fascination and the romance. Uh, yeah.
And then you're now getting to a bit of a reality check of, did you really just tell me that? Is it true? Uh, the same thing that you do with a human? Okay. Uh, that. Now, don't get me wrong, I think what OpenAI has done with chat, G B T is really made it accessible and I think that's the real crux of the innovation.
At least in my head. The ability of AI is gonna dramatically change, but the fact that you can now engage with it, you're at the point now, you can question it, which is still a good step forward.
[00:09:25] Andreas Senie: Yeah. All the while teaching it, gathering more data, I always remind people, you know, you get to that cap show where you have to flip the donkey over till it's standing up.
You're teaching an AI something, I mean, that's where that data's going at the end of the day. Um,
[00:09:39] Kevin Collins: and that, that's part of even the, the challenge that a lot of people don't understand because they'll see the output. But behind the scenes there's a lot of teaching that has to go on. There's still a massive effort to train these models, uh, to make sure that the veracity or correctness of the information is there, requires
[00:09:58] Andreas Senie: a lot of background training similar
[00:10:01] Kevin Collins: to how
[00:10:02] Andreas Senie: you train a team of people.
Well, so, and then so, so a lot of training, background training. There are, there are famous scenarios where these, these bots, chat bots have come on and been taught the wrong thing and said the wrong thing. And I'm, now, I'm gonna talk about Charlie AI for a moment. You are, you are executive, uh, chief of staff, which I love.
You're chief of staff type tool. It's much easier to train good behaviors and good traits into something specific like that, as opposed to, Chat, G P T, which is doing everything all at once out in the wild. And how do you control the, I mean, one is Wikipedia, one is brittanica,
[00:10:43] Kevin Collins: uh, and that's actually, those are good analogies.
I haven't thought about that in detail, but I do want to go through that because you have to for a couple of reasons. If you think about what chat G B T has done is really. Has opened up this capability for a broad audience, but they hallucinate, and that's simply because they've taken this vast amount of information and they've put it through their models to teach it.
And understanding what's true, what's not true, uh, is up for interpretation in our world. Uh, Charlie, uh, we have to, we address the commercial real estate market, the financial market, the legal market, and everything that we produce has to be based on fact. Right? That becomes significantly more difficult because now you're basing the information on training data that is highly vetted, um, that is.
Knowing how to auto correct itself. So there's, it's a check and balance that we have to do such that we've got a set of AI models that will generate some great output. Yeah, and we love it. Similar to chat b t, but then we also have a whole set of other AI models that gone. What did you just tell me and I'm going to go and verify it and I'm gonna go and redline it
[00:12:08] Andreas Senie: and auto correct it.
Well, and so that's an, you raised an interesting point. So deep fakes and, and the ability for AI to create things. Yes. That, that's where people get excited. It's like, well, AI can create a story. It can tell a story, and there's a whole team that does this have the trick, the AI into telling you things it shouldn't.
Right. And, and people can read that. I'll leave a link in the show notes, but so your, your approach to AI with Charlie, this, which is a sector specific ai. You're chief of staff, you're to replace someone, not to replace someone that could help someone ease, uh, in their work. You are, you're, you're building in a crosscheck for that specifically.
At the same time. It's like, here's the, here's the AI that's gonna do it better, but here's the AI that's gonna help teach it to do it better. Just like they did with the, the first, the first computer to win it. Chess. Yes. Run it against each other over and over again. Yeah.
[00:13:03] Kevin Collins: It's the challenger. Um, you always have to challenge yourself.
Even humans have to do that. Um,
[00:13:10] Andreas Senie: Well, I don't play chess. Uneasy. I I got it.
[00:13:15] Kevin Collins: But the, there's a, a concept that we have had, if you think about all this hype around chat, G B T, uh, it's really a single model and a single brain that is producing some output. The principle we have more on the industrial and the enterprise side is more of the thousand brains.
[00:13:36] Andreas Senie: Yeah.
[00:13:37] Kevin Collins: Not the one. We need the one, we need a collaboration, uh, of a team of models that produce an output. And then we've got another set that does verify. Uh, and that is how people work in business today. Uh, it is a team working on it, and this is why we have more than that thousand brains philosophy and how we build it and use it, rather than just relying on a
[00:14:05] Andreas Senie: single.
Right. That's an interesting point. The, so, I mean, Microsoft made, its made its mark on the world, let's say it, or, or continued to make its mark on the wor on the world when they shifted into a company that focused entirely on developers being able to access it and make things on it. Yeah, and they've kind of done that with the AI model.
Here it is, open ai. Let's keep going. Yeah. Um, but as you just pointed out, or rather as we know, Microsoft, because that's so open and wide, a lot of good can happen and a lot of bad can happen. Programmers will do what they can do. Yeah. Whereas in your model, the. And in others like you, are there others like Charlie ai?
[00:14:51] Kevin Collins: that's, that's gonna be
[00:14:52] Andreas Senie: my, my Chainless. So, so Charlie AI's model you have, you have a thousand Charlie AI's brains all working independently with their own access to. Certain knowledge based on, on yes. Credentials, which incidentally, in my own shameless plug, is very much like kreo AI as far as connecting data is why we're excited to work together.
Um, and so you're teaching them independently, but they all benefit in a collaborative way as a. How, how?
[00:15:27] Kevin Collins: Yes, exactly. And, and part of the reason we do that as well is that we have to worry about protecting the data, the privacy, and the intellectual property associated with all of our customers. Uh, and that means that their data can't go in to train the bigger brain.
Uh, so it has to train their own brain and their own team of brains. So there's, that's part of the reason that we've designed it this way is protection of information and those aircrafts that you have to have. Yeah. Uh, a good example of it is your data. You don't want to go and give it to your competitor or allow your competitor to use it or to unnecessarily or inadvertently expose some of your client information, cuz all of it, sudden it's stuffed inside.
[00:16:16] Andreas Senie: AI model. Right. The AI doesn't know any better not to share it. Yeah, because you didn't tell it, you didn't train it. Nope. Uh, there, there's a news story about this. It was, I forgot which hardware, but I guess some engineers at one of, and even if I remembered, I'm not gonna say their name, company X uh, engineers went in and were, and were asking the AI to write code better for some sort of proprietary software.
Company B'S engineers went into the same AI and said, Hey, which one of the software is that company? X should is the best, should I use. And I, as I understand the story, to, to be the AI system spit back out, oh, use such and such. Because they have this brand new and it was proprietary information that that first team of engineers were entering because it didn't know better.
So now all of a sudden you had, you had trade secrets just out there in a brain. Yeah.
[00:17:10] Kevin Collins: And that, that's, that is the. It's a very real danger and, and I saw a lot of the recent government, uh, panels on this and discussions even with open AI there talking about the dangers. And it's a reason we do not allow our own engineers to leverage any output from open ai.
We're still liking what they're doing as far as opening this up from a. An accessibility perspective, but we know the real dangers and that could include, I don't wanna expose our company to liability because we ended up using someone else's code that was inadvertently exposed. We don't want the liability around if they're leveraging GPL or L G P L licensing.
[00:17:58] Andreas Senie: a big issue for us. But, and GPL stands for.
[00:18:02] Kevin Collins: The, the, the licensing mode, so the software licensing, and there, there's certain open source licenses that. Protect your code versus still allowing you to use this open source. There's others that can put you in a position where you have to expose
[00:18:19] Andreas Senie: your code as well.
And so that's, it's funny enough I went, so we have a platform cracker as a platform, as our audience knows, and I joked when, when chat G P T exploded on the scene, I was like, To Steven, our cto. I said, Steven, come on. Go get it done. Just dump it into the system. It'll, it'll, it'll speed you up. And he goes, I, I'm not doing that.
Are you insane? This is our code. And once we give it to someone else, at that time, at that point in time, and I don't know how, this isn't a regulatory rebuff, I don't know where they are today, but he knew, he said, oh, you can't do that. Yeah, you can't just dump your coat in and say, Hey, clean it up for me.
Because then they own the cleaned up version of that code. Uh, that,
[00:19:04] Kevin Collins: that's the other one, is that, that's a feedback loop to them on your code.
[00:19:09] Andreas Senie: Mm-hmm. Which was incredible. I went, I didn't, and, and then I started to actually dig in and I'd been lucky enough we'd been working together on, on how to integrate Charlie AI with commercial real estate, with reporting specific to commercial data and, and we had.
Because chat g b T hadn't exploded on the scenes. I wasn't even thinking these things you have already basically covered them. So I went, makes perfect sense. And then all of a sudden they went, uh, chat G Petit shows up in such a loud way. And I, and I did. I, a friend of mine goes, just open it in a browser and try to run your day on it.
And as you're running your day, if you have a question or a thought, ask it. Uh, I was not impressed to say the least. Uh, and I, I actually went back to Charlie AI team. I'm like, do, do you believe what this said? And that's when I, I really started to digging and that's why I was excited to get you on the phone because I see my colleagues, partners across the country, they're all like, well, chat g p t.
It's the answer for writing your descriptions. It's the answer for so many things to make your life easier. And really, I understand on an educational level, if you need a foundational. Blueprint or if you needed to summarize something, maybe it will do it better. Me? Yeah. Well,
[00:20:24] Kevin Collins: there, there is a, it's a philosophical difference as well.
Uh, and I, I know I'm probably not gonna be in the majority, I might be in the minority here, but I like to research. I'd like to get a wealth of information.
[00:20:40] Andreas Senie: And draw your own conclusions. Draw
[00:20:42] Kevin Collins: my own conclusions. So even the philosophy we have is, and you mentioned at the beginning, it's all about supercharging human decision making.
It's not about replacing the human
[00:20:52] Andreas Senie: right. Uh,
[00:20:52] Kevin Collins: we wanna augment our capabilities that even as the chief of staff, it's not about replacing somebody, it's just. I, I need somebody, I need to clone myself cuz I'm tired of doing all this stuff I don't want to do, but
[00:21:06] Andreas Senie: I still want ar arbitrage of information. It it, there are certain things that computing power, you know, computing power becomes your exo cell skeleton, right?
It's, it's, I can lift a car if I have enough power, but that's physically Now if, if AI can accelerate my decision making by giving me arbitration that information, giving me the right information, that would be great. But it's not, as I understand, it's giving me, it may be giving me the right information, it may be giving you
[00:21:35] Kevin Collins: the right information, but it's also giving you a biased opinion of information as well.
And that's the other real danger. Uh, Even with social media, we still end up in these loops because these formulas are designed to keep you clicking. It's a click bait, uh, thing that you'll see on social media all the time.
[00:21:55] Andreas Senie: Echo chambers, right? You're gonna end up where it wants, it's where it wants you to be and, and so let's, it wants you to be, and then you have coding bias, which our audience might have never heard before.
So, yeah. Uh, can you speak to how AI is solving for that or avoids that? Or is it. It's just there. Oh, I think it actually makes it worse. Uh, yeah. So, and for those that don't know, coding bias is exactly what it sounds like. Yes. If, if you are of X, you're an X type of person, you're gonna naturally type in characteristics for X type of people.
It's exactly what it sounds like. Yes. And that,
[00:22:31] Kevin Collins: that is where I think the danger of AI is. And this is why I'm a big fan of regulatory. Uh, there has to be oversight, there has to be regulations. Uh, it doesn't have to be the government controlling it, but there has to be some type of responsible approach and accountability in place that we're not falling into traps because the other thing around coding is you can find is there's gonna be bad actors and they can inject Trojans through these AI systems
[00:23:04] Andreas Senie: that are auto generating.
Yeah, and now you're in a really
[00:23:08] Kevin Collins: tough spot because it's a, it's basically AI doing the generation and it can do it a lot faster and change and adapt a lot faster than what a human bad actor
[00:23:21] Andreas Senie: could do. Well, and that goes into the cybersecurity side. If you put a, a bunch of ai, you know, bad acting AI against a traditional security system, and this is a different sector interview, please come join that one.
That'll be a fun one coming up on the 18th. Um, but yeah, I mean it's, it's a, in theory it is a, uh, for lack of a better word, it is, it is your chief of staff. It is a workforce of, of. Things going out there and doing the heavy lifting if done correctly for good or for bad? In theory. Yeah. I mean, part, part of our
[00:23:56] Kevin Collins: approach is to, we, we knew this problem and a lot of other companies have been there, so the AI has been around for quite some time.
Even the problem around hallucinations, you would've seen that from Google and Meta,
[00:24:09] Andreas Senie: uh, a couple of years ago. Yeah. Uh, this is
[00:24:12] Kevin Collins: not new. We've seen it with our own large language models. It's part of the reason we have invested heavily into ensuring the veracity, correctness, traceability, uh, the auditability of the AI to guarantee your customer's, uh, privacy.
[00:24:34] Andreas Senie: Uh, edu educated information or verified verified
[00:24:39] Kevin Collins: information. Verified verified information. Mm-hmm. And, uh, so you can't get to the point where there is responsible ai. Um, the dangers are very real, but there is more to this than just a large language model. And I think we, we talked about this right at the beginning of the.
The chat here is that there's more to AI than just the chat p t
[00:25:03] Andreas Senie: or large line response. Yeah. Well, and chat PT, as you pointed out, just makes it easy to access. All of a sudden you have a buddy to chat with online, and that's what's kind of exciting is, is it's easy and you're chatting and it's responding, and now you can click and investigate.
But what, and many large corporations have already jumped in and said, Nope. Don't use chat c p t for the reasons we just discussed. Don't, don't you, you shouldn't use it for this cuz of proprietary concerns. All of these things are jumping up in major corporations. Uh, there's still a huge opportunity, I think, for some of the smaller corporations to use it the right way, uh, to, to come in and have best practices if they in fact want to use an AI model or, and or just come in and use a purpose-built solution like yours.
And while yours is the only one of its kind, I imagine more companies will try to, well, actually, let me ask you. Will companies try to create their own AI internally? Uh, they
[00:26:02] Kevin Collins: will, uh, and we call it the diy,
[00:26:05] Andreas Senie: uh, the do it yourself ai. Yes. Yeah. Is that the janitor as opposed to the chief of staff at Charlie?
[00:26:11] Kevin Collins: Well, it, it's the, uh, I think there's a lot of fascination. There's a lot of availability tools, and again, I'm a big fan of having those tools available and the open source community around it. So there are organizations that still need to use these tools and build their own to solve their own unique, um, challenges with certain use cases.
But we would caution them. You need to build it around a framework that allows you to adapt. Um, and we have talked a lot about adaptable AI because now you have to go from one use case to the next use case to the next use case, and you can't scale, uh, a bespoke training exercise
[00:26:55] Andreas Senie: on all of these models.
So it's not as easy as going out and changing the, the tires on my fleet of trucks cuz it's winter. Yes. This isn't something you can just say, all right, we're gonna, we're gonna make a simple investment and we're there. Yeah, exactly. It's the opposite. It's, it's a major initiative. It's a. It's almost a, uh, a shift in the way your company works if you want to go down this road.
Yes. Yeah. And this
[00:27:19] Kevin Collins: is where we, we do caution a lot of our customers, you still have to go and think about building your own models, but they have to be put into a framework that scales. And this is part of the reason we want with our own platform, uh, we call it the ANCs platform that allows customers to plug in their own models, but support the, the training and adaptability of those models across their use cases.
[00:27:44] Andreas Senie: Um, now one
[00:27:44] Kevin Collins: customer brought up the term of, you know, house of Cards and, and a lot of organizations, there's a lot of bespoke, isolated, siloed work going on, and that can really become a house of cards.
[00:27:59] Andreas Senie: Whereas if you put it into a framework
[00:28:01] Kevin Collins: and you've got the management around it, um, you know that you can't easily adapt from case
[00:28:08] Andreas Senie: to case to case.
That's interesting. So the, so your model or your system is, Dare I say plug and play, you're gonna give it the inputs that you have available as a company. It's gonna take its brain. That brain will grow as, as needed and trained by your team and the, the cl the client, the customer. Yeah. This is what we're looking for.
This is the answers we expect. Uh, and very much so giving the something you can walk out with confidence on and be sure that it's not hallucinating. Yes. Okay. I mean, that's a night and day difference that's showing up. That's showing up to your, you know, your, um, uh, showing up to the bar and trying to guess your an that, the answer is C on probability because it's a multiple choice test as opposed to actually knowing the answers and going through and, and getting the right answers.
So that's exciting. Well, just on that
[00:29:05] Kevin Collins: one as well, is that we always design it so that the human is never out of the loop. We want to keep the human in the loop.
[00:29:13] Andreas Senie: That's part of supercharging humans. Yeah. That
[00:29:16] Kevin Collins: gives the ability of the human to look at the decisions. And help feedback and correct.
[00:29:24] Andreas Senie: Um, well, I mean, speaking into said electronic devices, I have both types in front of me.
I'm not gonna say the name or else it'll interrupt. A, a call wasn't a thing until it turned the lights on at all. I mean, that's what. That's what made it a thing. So if we can, and I can't literally can't say it cause I have both on the other side. So whatever
[00:29:44] Kevin Collins: words you're saying, and if you were in your car, you would, uh, not be able to say anything.
[00:29:48] Andreas Senie: Yeah. So something's gonna respond. Um, and, and they're always listening, but that's a different conversation for a different call. The. So here you are, you're making it applicable by industry. You said you're in financial services, legal and commercial real estate. We knew, and that's why we have you on the call, how long have you guys been in financial services, in legal?
I mean, how long has, have you been building this ai? Because you gotta teach it, which means the longer you've been around, the better you are at doing it. Yes, in theory and,
[00:30:22] Kevin Collins: and part of it is a, again, a, a. A design approach that we have versus what is happening on the world of chat Yuki, cuz that could take a long time to train.
It's massive amount of data and massively
[00:30:37] Andreas Senie: expensive. If
[00:30:40] Kevin Collins: you go into,
[00:30:42] Andreas Senie: uh, specialty
[00:30:44] Kevin Collins: verticals, including financial services and that's brought by itself or commercial real estate or into legal, you have to be able to go from one use case to the next use case to excuse very quickly with the same level of accuracy, uh, and veracity and checks coming up.
So we've invested heavily into, uh, techniques around transfer learning. Uh, that allow us to quickly adapt from one environment to the next. Now, we started off building this tech, seeing the problems in the industrial world. So years ago, 2016, even before there was a whole concept around digital twins in the industrial world, they're still around.
Yeah, and we saw the issues of how to build digital twins, and it comes back to a lot of Achilles heel on data training,
[00:31:38] Andreas Senie: testing.
[00:31:40] Kevin Collins: So we ended up investing heavily into a platform that allowed us to scale a little over four years ago to solve some of these Achilles heels problems, and we've proven it out over the last four years.
Uh, so, so that we have been in financial services now for a little over two years. We have quickly transferred training from that sprayed into commercial real estate, uh, for things such as market research, property research, uh, going into doing valuation exercises, uh, and giving insights into different markets.
And we've even adapted that into the legal side because there's a l a lot of similarity. You know, in financial services you have to deal with legal clauses. Oh, by the way, when you're in commercial real estate, you're also looking at legal agreements,
[00:32:33] Andreas Senie: Lisa landlords, tenants. It's fair to say that you, you are training in legal and financial services prepared you to work in commercial real estate, cuz that's what bridges all three.
Yes. Right. Um, again, it's, it's, here we are in the background chat, G p t paradigm shift of the industry going, oh my god, AI is here. And it, and, and it's the only way I can say it's Saul Klein in the, the regular show. We talk about technology a lot. Everything we do in commercial is one, it's 10 years behind residential on purpose.
It's still a closed network except everybody's been excited about this idea of ai. And now that I'm talking to you and the way you've framed it as a chief of staff, I think actually what we're really more excited about is the chief of staff part to make, there's a lot, there's a big part of being a broker, an investor, developer.
That we just don't want to do. It's just part of the job. And this will do those parts, that'll it'll look into that market, analyze that market from all the different data sources available, and come back to us with a recommendation that we can then prove, disprove, or improve upon. But at least it's from a place of, of confidence.
Definitely. Yeah. A hundred percent. Ugh. And, um, it's an exciting world. So what, so let me ask you this and, and we're, we're running through the time here cause I could talk about this all day. What do you see as the biggest hurdle specific to commercial real estate? Either on the language model or the, or your model or both.
And what's the biggest opportunity for people in the industry? The, the biggest challenge
[00:34:13] Kevin Collins: is, uh, I don't think a lot of people understand just how difficult the AI elements are. So Chei fairly straightforward given their focus on a relatively known market that they're going after. But when you get into commercial real estate, there are so many different use cases and so many different.
Proprietary data sets.
[00:34:39] Andreas Senie: This isn't
[00:34:40] Kevin Collins: the public internet that you can train on. You're going into some very closed, you, you mentioned the term closed. A lot of this is closed
[00:34:47] Andreas Senie: network. Yeah. So the data sources are buried Yeah. And are still analog. Yes. Not even, let's not even go into the analog data. Yeah.
Buried, buried, hidden OPA analog in that.
[00:35:02] Kevin Collins: And part of it, just from a Quicken analogy perspective, it's the challenge is not the data itself. The challenge is unlocking that door. You get access to it, and once you do have that access to it, You can open up a wealth of opportunities.
[00:35:20] Andreas Senie: Uh, well, so unlocking the door and access to it as commercial brokers, I'd say it's a conversation I have with Saul a lot.
I mean, data is the digital oil of the future and commercial real estate and even residential real realtors, I mean, they sit there, they collect data and they input data into systems and it's, it's a treasure trove. Some don't know how valuable it is. Yeah. Um, so we're at the point of capturing the right data.
We're at the point of having the right data, and it sounds to me like with the right, a AI application, Charlie, in this case, you're actually at the right point to maximize and leverage that data for. Intelligence action, what I call actionable insights. Yeah. As opposed to overload of information. It's great to have data in 12 places and know every statistic on the map, but what good does that do you if you can't have an actionable insight from.
And that's, that's the, the biggest opportunity I see with AI is it can help get you to the actionable insight faster or identify one that you may not know is, is there definitely, and that,
[00:36:25] Kevin Collins: that's where the opportunity is. So that, that's the other part of the equation when you ask the question is those insights we're the cusp now of gaining very valuable insights.
Um, because the biggest challenge is unlocking the door and some of those doors are being unlocked now. Yeah, and it can be leveraged. So now we're on the cusp of far greater insights, and I'm seeing that as of yesterday, uh, I was actually looking at some, what the AI was showing me. I. That's fascinating.
Uh, so it, it's almost there. I'd say we're on the cusp of far greater insights that can drive the commercial real estate
[00:37:04] Andreas Senie: Well, and, and as you said it, you're not replacing anyone. You still need the, um, I, I, I've heard this in the finance real, you soon need that human overlay, right? It's still a combination of both.
No matter how smart the computer gets, you'd need the nuanced intelligence that only resides in a human brain. Yeah. Not, that's not, that's not being developed yet. It's not even been thought about. There are, there are brains being grown in labs, but they're not, we're so far away from that happening and replacing people just like, uh, machines and factories.
Everyone was worried that, you know, a conveyor belt is gonna replace. Workers in the factory. That's not in fact the case. Somebody still needs to be there to turn it on, turn it off, to train the arms, to move the arms. We're so far removed from AI replacing anyone, and I've gotta ask it cuz we're talking about AI and we are so far removed from AI taking over the world by rise of machines.
Yeah. That's just not, not, that's not a real threat to this industry or any industry.
[00:38:07] Kevin Collins: Yeah. And then that's, uh, people are worried about this. There is going to be a shift in the workforce. Similar to the conveyor belt, people needed a shift from doing the, the basics to more assembly line. A, adjusting their skills, adjusting their work to the assembly line.
But the same thing's gonna happen with ai. There'll be a shift, uh, and I think it's a positive shift because I think it has the ability to create new jobs. Yeah. Uh, And it is just a shift in the market. Uh, but people have to
[00:38:40] Andreas Senie: be aware. Is that means new training, which creates jobs? Yes. And on and on. Yeah. I do it with my granddaughter.
[00:38:49] Kevin Collins: I, I do have a granddaughter and I've got granddaughter and grandsons. And my thing with them is that the world that you're going into tomorrow, you have to be prepared for, so it'll be quite a bit different. Then today, which means you need new skills and you can always upskill,
[00:39:06] Andreas Senie: relearn. What you can't do is stick your head in the sand.
Uh, one Saul, Saul and I keep prime on Saul. Saul was the internet evangelist for the residential world. He helped bring the first listings online. Um, and he said, and he said it. He goes, look, you don't need to be an expert. You just need a little bit of it. Have a little bit of it in your business, and then grow that from where it needs to grow.
Yeah. Um, but do not stick your head in the sand. And I have never seen an indu, this industry commercial real estate move so quickly in any direction and it's this direction. So I'm, I'm hoping that thanks to this call others educating and regulation, as you pointed out, that we get to a point of operating from confidence and not just to a point of, uh, fear of missing out.
Which this industry's never had, and now all of a sudden, brokers and investors and developers are going, oh, I have to have that because it's because it's, it's what's what my competitors are using. No, you have to be very smart about what you're gonna apply and why? Because I. Um, while this is not a fad, it is a very, it's a flash fire.
Things have happened very quickly in this space.
[00:40:13] Kevin Collins: Yes. Yeah. The, uh, and as you say, that flash fire, that height you see today will settle down into reality.
[00:40:21] Andreas Senie: Right. Uh, and that's what we're seeing now. This regulation, this, the, all these concerns are happening. Yeah. Um, alright, well, so where's the one place?
Last question. Where's the one place people should be paying attention or the most important place for people to pay attention to the world of ai? Is it, is it following, uh, Charlie AI's blog? Is it following regulation? I mean, is it reading a certain book? I mean, what do you follow? The big movements? It's a
[00:40:48] Kevin Collins: part of it is.
I, I would say, I mean, even from the three that you mentioned, I, I think the regulatory space has got, A lot happening in it. Um, and that's going to have a big impact on commercial real estate. And that's where, if you're in commercial real estate,
[00:41:06] Andreas Senie: whether you're a broker or whether you're,
[00:41:10] Kevin Collins: uh, on the landlord side or whether you're on the tenant side to pay attention to the regulations because the regulations are gonna be coming to surface key issues that are very real.
Uh, now the hype is there, but if you're going to just bank on the hype, You will probably miss the signal and the noise. And this is where I actually view AI as gonna have the be. It's figuring out the signal, all the noise. And there is so much noise on the hype cycle. So be careful
[00:41:39] Andreas Senie: what you read. And do not ask Chat, G p T or other AI to keep you informed.
That's not Model Verify,
[00:41:48] Kevin Collins: ask. You're gonna be doing a lot of verification.
[00:41:52] Andreas Senie: Uh, I love it. Uh, sorry for the background. If, I don't know if you can hear that background noise, but we're a bit noisy today on my end. The. As I said, that was the last question, Kevin. I mean, this was fantastic. It's, it's a mushrooming space and maybe not, it's, maybe it's not mushrooming, it's now hitting its hockey stick moment in growth.
It's been around forever, but people are really paying attention. Um, I can't say enough how much I appreciate you coming on the call, how much I appreciate in partnership, Charlie and correct AI are gonna be servicing the industry on a commercial real estate standpoint. But just in general, cutting you hear as a signal, cutting through the noise.
Because everywhere I look, I'm seeing chat, G P T and AI and all of these, this confusion, understand, verify, investigate even. Uh, more importantly, train it. Pick what you need when you need it, and train it to be what you need it to be, just as you would an employee is what I'm hearing. And Charlie AI today is the only one that can do that.
Is that accurate?
[00:42:55] Kevin Collins: Uh, yeah. Cause it's all to do with our focus on the veracity qualification, double checks. Focus on privacy and security around the
[00:43:04] Andreas Senie: data and the models. So, so I, and so in that case, ignore the emails that I'm now getting inundated with, with companies promising me a solution to write better descriptions with their proprietary AI or their, their, it was like the SEO thing, like all of a sudden everybody was an expert and I'd get these firms writing me blast emails.
And now I've started to get ones about AI and they're like, we're, we are a true, we're an AI company and we can help your business. Well, these mom and pop AI companies really don't have the potentially, I can't speak for all of them. I haven't met all of them. Most likely don't have the skillset to be training it the right way.
Is that a fair statement?
[00:43:45] Kevin Collins: Uh, it is a fair statement. This is still a highly challenging sophisticated industry around these models. You do have to know about how to train them. You have to know about the risks, uh, around it. And that does not, the expertise for that is few and far between. Uh, open the eyes obviously got that expertise that Google's the Microsoft, they have it.
But when you're getting into, there's just not enough talent there.
[00:44:09] Andreas Senie: Yeah, well, I'm getting into third parties, you know, rebranding and white labeling and, and tinkering, I guess is what they're talking about. But our chief
[00:44:20] Kevin Collins: scientist comes out of the world of quantum computing. She competed in the mathematical Olympics.
She's, wow. She's a physicist that has very significant experience in the industrial world dealing with complex models in ai. Uh, At such extreme levels and that type of talent,
[00:44:44] Andreas Senie: few far between. Yeah. It's not that, for lack of a better way to say this, it's not your 13 year old who's a whiz on his computer.
Yeah. That's not gonna be your next AI genius. Your AI geniuses have are, as you just said, they're in, they're, they're, they're competing in math in. The Olympics of math and otherwise, um, at levels I don't even know about, which is why I asked. Well, that being said, the one, uh, magician and math, uh, not math magician, but magician.
I do know Mr. Mendoza, if you wouldn't lead us out, I wanna thank you again, Kevin, for joining us, for giving us this, this in-depth conversation. People can reach email@example.com, or at least they can reach your chief of staff there to ask questions. They can. Yes, and I wanna thank our audience for tuning in with Without You, this show doesn't exist.
We're available everywhere. You get your audio and videos through Cross social Media. Please do rate and review us. It really does help. And don't forget to join us the first Thursday of every month for the round table. Kevin, thank you. Um, I'm, I'm, I'm sure we're gonna have quite a few questions ongoing, and I, I'd love to have you back and I'm afraid to bring you on to the, the, the cybersecurity call, but we'll see.
Be good, uh, panel discussion. Yeah, yeah, yeah. Um, alright, Mr. Mendoza. Thank you, sir.
Perfect. I think that was great. I, the time flew, I looked up at the clock and I'm like, oh my God, we're 45 minutes into this. Yeah. Um, so pots give you a little prompt. What? No, I know you are. And, and then of course I'm working from home. My landscapers all showed up and I see 'em on the cameras coming closer with the happy machinery.
I'm like, here comes all the noise and I'm trying to mute why I
[00:46:55] Kevin Collins: stick myself in the basement, uh, just to stay away from all the noise that's happening up
[00:47:00] Andreas Senie: there. Oh, yeah. Well, now lower level and the, the only time you'd hear it is when they come by the back windows. Because I have, you know, the exits in case there's a fire.
My wife wants to kill me. I have a, I have an escape route. Um, yeah, mines are all
[00:47:14] Kevin Collins: bared up. She doesn't bother me
[00:47:15] Andreas Senie: out.
Well, um, this was truly great. Your team's busy at work. We're sending 'em some stuff on the a on the, um, that they need to get their. The manual stuff done. We are giving them the keys to our data. So you're gonna have a lot of 27 million property records pretty soon. Okay. Nice. Uh, yeah.
[00:47:36] Kevin Collins: I'm actually, just so you know, I'm, uh, heavily embedded on the commercial real estate stuff today.
I love it. Since Friday, it's been deep dive into unlocking data. You asked me about that challenge and it's like, oh, I gotta something else I gotta unlock. And it's not so much we know how to deal with the data. It's like, how do I get through to
[00:47:55] Andreas Senie: get that door open? How do I get it? Well, I know that, um, after you started working with us, you got introduced to Cherry.
They've got a lot of data streams available to 'em. Yes. Um, Aldi's a great guy. Brett's a good guy, but it's still opaque on purpose. I mean, a lot of operators just don't want their stuff out there. And yeah,
[00:48:15] Kevin Collins: we find that, and that's the, uh, it's a bit of a finding a needle in a haystack sometimes,
[00:48:20] Andreas Senie: and. Yeah, I, I do have a hacking background and
[00:48:24] Kevin Collins: I'm going through and it's like, I know how to
[00:48:26] Andreas Senie: get to your data.
Yeah, yeah, yeah. Just share it. That's funny. Um, I'm just cur I, now I'm gonna ask the question cause I do have that, I do have Darren and, uh, this guy Steve Oron. What is the likelihood that people are using AI today to hack. A hundred percent. Yes. So I read last year that 70% of homes are hacked every year, most of which just pass by.
Cause they don't find anything of any, any value. Yeah. Now take AI into that equation. The AI is gonna have enough time to sniff around and say, wait a minute, they've got 10 grand in that checking account. Yes. To keep checking. So what does a homeowner, what does the average person do? Besides have that million dollar insurance policy from Norton.
[00:49:15] Kevin Collins: There, there's quite a bit that, I mean, I rely on keeping my
[00:49:20] Andreas Senie: firewalls up to date. I've got a physical one, not that I'm inviting you to try. Yeah. But I've, because I was hacked and I was like, and I didn't listen to my, my guy and I've got a good guy. I've got Darren, but I've got the, I've got a mask phone number.
I started to do all the things. But yeah, go on. But I also,
[00:49:37] Kevin Collins: I mean, there's a lot of things I don't do. I don't do. Uh, digital
[00:49:42] Andreas Senie: banking, uh, I'll use my credit card all
[00:49:44] Kevin Collins: the time, but there's no
[00:49:45] Andreas Senie: reason I use my credit card because it's insured. And you can, you can, yeah, you call 'em up
[00:49:51] Kevin Collins: and some cause they get hacked.
I don't want my bank account hacked, but my credit card get hacked. I don't care. I'm just gonna phone off the credit card company. So there's a lot of things that I do to mitigate, and I'm not
[00:50:02] Andreas Senie: saying eliminate, mitigate. Can't eliminate. You're reducing hard surface. You're reducing points of of risk and that
[00:50:08] Kevin Collins: that's part of it as well is I always pay with my credit card if somebody wants my bank account.
It's like, no, I'm not giving you that. Yeah. Somebody the other day says, can you eat transfer? I'm not e transferring
[00:50:19] Andreas Senie: shit. Yeah. Well, imagine my world. Can you give us proof of funds? No, don't send me the thing with your account number routing and all the other stuff. Just gimme a letter from your bank. Yep.
[00:50:30] Kevin Collins: Yeah, that's the, the other one. It's always a separation of channels for communication as
[00:50:35] Andreas Senie: well. Hey, fax is still unhackable Last I checked. Yep. And a fax machine.
I'll use fax. Yeah. Yeah. Hip. It's HIPAA compliant. And it's secure. Yeah. Not that I'm gonna tell people Onco to go get back. Go back to fax machines. We want to keep the momentum, we want the data online. Um, well, good. Anything else I can do for you? Uh, Mr. Mendoza's gonna clean it up. Make it pretty. It'll air on the, we have a set air date.
We do not, don't have a set air date yet. We'll have to go over that, but, uh, um, we're gonna, yeah, we'll have a, we'll, we'll schedule the air date. You don't have to do anything on your end except, except the speaker invitation on LinkedIn, because then we share it. Okay. And if you don't accept the invitation, you're not shown.
Um, but then that's it. Make sure I do that and we'll, we'll do the rest. Yeah. Um, so Chief Charlie, chief of Staff, ch how did you get to Charlie from Chief of Staff?
[00:51:36] Kevin Collins: So the, uh, it was basically looking at a bit of a plan words with, uh, uh, chief of staff, the lettering. Now we, there's no e at the end. The reason for that is to be gender neutral.
Cause I view Charlie as a she. Yeah. My wife's going, no it's not. She, I views Charlie as a hot Australian guy. That's the, uh,
[00:52:00] Andreas Senie: the inside joker around. That's hilarious. My wife's got, uh, I think English male on her Siri and I. Don't. Right. So that, that actually makes a lot of sense. There's a bit of a battle on the car cuz one of my
[00:52:14] Kevin Collins: cards has got a female British accent.
Yeah, this sounds
[00:52:18] Andreas Senie: so good.
[00:52:20] Kevin Collins: Um, so that, that one's a bit of a dropping the, let's be gender neutral on it then.
[00:52:26] Andreas Senie: Okay. Yeah. I don't envy the battles you'll rule face on this, in this front with, with the world. Yeah. Um, perfect. Thank you for your time. This was fantastic. Um, As far, we have his informational slide to throw up.
That's right. Yeah. I'll, I'll create that. Some of the stuff, some of the stuff we're using right now is just temporary, just placeholders and we'll create the, the final product, make it all very pretty. Yeah. And then, um, and we'll go from there. Super excited. If you need anything at all, please don't hesitate to reach out.
I'm happy to assist. No, thanks.
[00:52:55] Kevin Collins: That was a lot of fun. Thanks. Enjoy it. Thanks. Cheers. Thanks. Thank you. Thank
[00:53:04] Andreas Senie: All right. Here we go. Okay.
[00:53:07] Kevin Collins: That works.