1567 – Ai Co-pilots with Zero Systems’ Alex Babin

0:28Hey there, thoughtful listener. million in revenue. Just head to upmyinfluence.com and watch my free class on how to create endless high-ticket sales appointments. You can even chat with me live and I'll see and reply to your messages. Also, don't forget, the thoughtful entrepreneur is always looking for guests. Go to upmyinfluence.com and click on podcast. We'd love to have you. with us right now, Alex Babin. Alex, you are the co-founder and CEO of Zero Systems. You're found on the web at zerosystems.com. Alex, thank you so much for joining us. Thank you, Josh. Happy to meet you. Yeah. And what is Zero Systems?

1:13Well, in a nutshell, Zero Systems, I'll start with a nerdy explanation and then I'll go into the simple one. So Zero Systems and operating and orchestrating system to build, deploy, and run AI applications for enterprises. That was the nerdy one. Now I'll explain it in a more common language. Basically, AI is taking the world by storm, and enterprises, especially knowledge workers in enterprises, they need co-pilots, AI co-pilots. And the easy way to explain it is we are not good at some specific tasks as people, as humans, and AI can do much better automating those tasks. So we help people to focus on what's important to them, what they're good at, instead of doing the monotonous tasks where AI can be much better.

2:04So that's what we do, helping people with AI to be more productive. Yeah. And how does that happen exactly? And how are you engaging with enterprise level organizations specifically? So as you know, since November 22, the world has changed. And AI is now being viewed as the top priority for most of the enterprises. We know that this third industrial revolution started and it's AI revolution. No line of business is going to be untouched by AI in the next five years. And every CEO, every COO in a large organization now looking at AI, thinking about two things. Number one is how can I use AI to improve my business, to get more revenue, to reduce my costs, to make my people more productive?

3:01And the second thing they always think about is how it will disrupt my business. And it's not the question of whether it's going to be disrupted or not. It's the question when and how. And the biggest question for every enterprise is, am I going to be disrupting my own business with the help of AI making it productive? Or my competitor will get me out of business? So every forward-thinking CEO, COO, anyone in operations, even IT, think about it right now. And that's something that we'll need to live through. And there are a lot of things that are happening right now. but basically answering your question, our clients come to us because they know that we have all the components required for them to run AI securely, safely, efficiently inside the security perimeter.

3:53Because there are so many good things about AI, but right now it's such an infancy stage that a lot of risks also involved. We can talk about it, but those are great risks a lot of people don't even know about. Yeah. So I guess, you know, for someone who doesn't really understand the vocabulary of like why this could potentially be valuable for an enterprise level organization, how have you seen AI been adapted at that level for, again, for companies at that size, short of, you know, what I see maybe is a lot of, maybe little experiments being run here and there. But where have you seen AI been adapted at the enterprise level successfully and showing real promise?

4:45That's a great question. So what actually changed again since November 2022 when OpenAI released a chat GPT, which is an amazing product on its own. But what changed is that people saw the ability to do basically anything or everything with the help of AI. It opened up their minds and they realized that anything they can dream of now can be done. You can talk to AI, you can converse, you can ask it to produce the content. That's what is being called generative AI. So a lot of companies started building solutions around chat GPT and GPT-4 and other APIs for another large language models. We'll talk about what large model is in a second.

5:36But a lot of companies started building kind of wrappers around those APIs and those large language models. And enterprises were carefully looking at this saying, well, there are so many opportunities, but let's be careful. Let's start with one or two use cases and see how it being implemented, how risky it is, and so on and so forth. The biggest use case right now is, especially for knowledge workers, information workers, of course, is a content generation. And we see it a lot. A lot of companies that can help you write your emails, can help you write your blog post, can help you write the script for your podcast, summarize YouTube videos.

6:18So many applications there. I would say it's a very shallow and wide level of applications because, well, a lot of things you can do with content. But what if we start digging deeper? What if we start digging deeper where processes that enterprises are running every day are really time consuming, but also sound boring? Like when you talk about, hey, ChatGPT created a blog post for me and look how amazing it is, right? That's something people want to brag about. What about, hey, I have AI that processes 100 ,000 invoices for me? Doesn't sound that sexy. But the ROI for the enterprise, when AI can process enormous amount of information at no cost or little cost in seconds instead of weeks, is much more effective.

7:14So we need to separate those cool, fancy applications that everyone is talking about from real world ROI, unsexy processes that actually provide a lot of value. Yeah. And outside of content, what are you most excited about in the application of AI right now? Content's an easy one. And I think that that's super accessible. I can't think of a friend that I know it's not using that at this point. So, you know, heads up, you know, if you're not, you know, if you're not, if you haven't adjusted your operations to take advantage of technology, you're already getting, you're already lagging behind. And that doesn't matter what size organization you are.

8:07We're already evolving. that quickly. But, you know, these applications, obviously, you know, we're seeing applications with, you know, AI being able to write huge sections of code, for example, like that is pretty exciting. But, you know, what about stuff like, you know, and I think tools are getting better. And I think you could argue some of that is AI driven, you know, things like, you know, logistics and operations and some of the day-to-day how to run a company stuff that we spend big money on to support and make sure that it runs well so that we can serve our clients and customers well.

8:51What are you most excited about that maybe we're not thinking immediately about how AI is going to usurp a lot of what we're doing in very traditional means today? I would say the biggest pieces of ROI and the most visible, in the most visible way, will be happening, going to be happening where the intersections of those capabilities are going to be happening. I'll give you an example. One of the things that AI can do amazingly well and much better and faster than humans is data analysis. And in case you need to analyze thousands of documents, especially extract something from those documents, like dates, amounts, whatever you need to extract, that's what AI can do much faster.

9:41On the other side, we know that AI, especially large language models, they are good at generating content. So if those two basically pieces of the capabilities of AI can be combined together. And let's say you have a task of analyzing a lot of information and then writing a report based on the information you extracted. In the old world, that would take significant amount of time and a lot of work power and work time. AI can do it much better and much faster. So the best examples are going to be on intersections of multiple capabilities of AI. because those are going to be much much more visible um and that's where uh those and you mentioned logistics risks analysis financial analysis forecasting all of those kind of things and if you combine them together let's say logistics analysis and forecasting using the real-time weather forecast how our delivery schedule would work and what the risks are and how to mitigate them.

10:49That's where the gold nugget is, not just content generation. Make two of those capabilities overlap, and you have an amazing solution. Make three of them overlap, and you have something outstanding that your business can be taking absolutely next level. Yeah. You'd mentioned risks, and I was just looking at your social media, and you had kind of a funny post of the, I don't know if you remember, a runaway suitcase that could have been illustrated as being AI powered. It's chasing after its owner. The problem is the owner's on an airplane. So the suitcase is just kind of like running down the runway because it didn't get loaded on the plane because that's what it's programmed to do.

11:32You know, it's just kind of, again, kind of a funny illustration of, you know, considerations, right? And I think that because it's such, for many of us, or, you know, we're kind of running experiments, it can be easy, I'd say, for us to not consider, especially when we have, you know, mission critical systems or processes, or, again, the impact of, you know, a mistake or, you know, AI choosing something that we hadn't anticipated. You know, if you're enterprise level, you know, you got to have some stop gaps built in there. Otherwise, you know, all of a sudden now, you know, you're taking off 5 million customers and you're going to get a lot of negative press over that.

12:20And again, I think we've seen some of this, but what are kind of your rules for engagement or, you know, how can we sandbox and kind of adapt technology in a way that's going to be disruptive in a positive way, not in a negative way? Yes, it's actually a really, really important topic. A lot of people are trying to sometimes avoid or not to put a lot of attention to, but it's actually critical and important. I would say that example of the suitcase running on the runway is a great example. Yes, it was trying to chase the owner, but what the risks are here, it may have a collision with the airplane.

12:59It might be hours of damages being done. And who is responsible for that? Is the owner or the manufacturer of those suitcases or the airport themselves who are not using the right guardrails to protect airplanes from that? So there are so many questions. I would say right now we're at the stage where it can be compared with putting a nuclear reactor in the middle of the city and pushing a button to see if it works. that's how i would describe it uh and the reason is that we have all those great tools but they are unpredictable those large language models sometimes behave unpredictably of course companies trying to put the guardrails they're putting checks and balances but they are so powerful that no one can predict the results there are so many new technologies like auto gpt has been released recently open source project where people can connect multiple steps and actions and ask AI to basically perform some actions for them.

14:13Sometimes it works well, but sometimes it goes off the trail. And at some cases it can be very, very dangerous. So we also have examples when OpenAI have been releasing the paper, scientific paper on how they were training a GPT-4 model. And they highlight, openly highlight the example when the model actually was lying to humans to achieve the goal. So yes, and that's pretty scary. And we just scratching the surface. We don't know what's happening underneath. And in this case, reliability of an output and predictability of AI behavior is a priority number one for enterprises. I'll give you another example.

15:03We know models hallucinate. Sometimes they make things up. Well, in many cases, they make things up. There was a case when the journalist was talking to the British journalist, I remember his name. He was talking to ChatGPT and asked about himself. He was surprised to know that ChatGPT told him he was dead and made up an article in the newspaper where his death was described. And he was so convincing. And if that was not about himself, like, I'm not dead. I'm here, right? If it was not about himself, he would say, well, I would believe it because it was so convincing. So again, models hallucinate.

15:45But let's use this example. Let's say you're using a bank and there is a slight chance, let's say three percent chance that your bank instead of your account will send your money to someone else it will hallucinate just three percent not a big chance would you use that bank i would say absolutely not what about 10 how about 20 that's how big the level of hallucination is for those large language models right now so enterprises can rely on something that is not 100 % accurate. And in this case, one of the biggest things we do for enterprise is actually building, we have those components that are used as a source of truth or validation.

16:35They always check if the output is 100 % correct. Without that, AI, especially for enterprises, is just a fancy tool or even fancy risky tool. They can't afford it. Yeah. Your website, zerosystems.com, Alex. And by the way, congratulations on all the growth. It looks like you're doing a lot of hiring. You're going through a lot of growth right now. And it looks like you work in, you certainly are big in legal accounting consulting, but who among those that are listening should absolutely be reaching out and having a conversation? Or if there is maybe a resource or something that you would recommend for kind of the next step, especially if someone's listening to our conversation as part of their due diligence on XeroSystems, like what do they do next?

17:32So there is a section on our website with use cases and blog is also pretty powerful with all the information that how enterprises can take the leverage of the new AI technologies without basically risking the data because data is really really critical and what you mentioned legal accounting consulting financial services those all of the companies that can be or verticals that can be described as professional services so professional services are really knowledge worker heavy unlike manufacturing or oil gas they're knowledge worker heavy and knowledge workers a specific type of workers the only thing in common all knowledge workers have is they all work differently so trying to automate everything for everyone would fail from the very beginning because we're doing things differently You're doing your emails, managing your emails differently from what I'm doing.

18:36You're managing your time differently from how I do it and millions of other people. So in this case, AI co-pilots for knowledge workers that mimicking the way we work instead of trying to force the new way of doing things at us is the answer. And any enterprise, literally, if not, even not in a professional services, let's say pharmaceutical. They have a lot of parts of their business that are knowledge workers related, accounting, legal, internal legal, procurement, finance, and so on and so forth. So any enterprise has lines of business that needs to be and can be automated and augmented with the help of AI.

19:20So I would say there are great case studies how we're saving millions of dollars and helping our clients to actually get new revenue, millions of dollars more by using those new AI technologies safely and securely. That's what we do. And there are many materials there on the website. Excellent. xerosystems.com. Alex Babin, co-founder, CEO. Alex, thank you so much for joining us. Thank you, Josh. My pleasure.

19:52Thanks for listening to the Thoughtful Entrepreneur Show. If you are a thoughtful business owner or professional who would like to be on this daily program, please visit upmyinfluence.com slash guest. If you're a listener, I'd love to shout out your business to our whole audience for free. You can do that by leaving a review on Apple Podcasts or join our listener Facebook group. Just search for The Thoughtful Entrepreneur in Facebook. I'd love, even if you just stopped by to say hi, I'd love to meet you. We believe that every person has a message that can positively impact the world. We love our community who listens and shares our program every day.

20:36Together, we are empowering one another as thoughtful entrepreneurs. Hit subscribe so that tomorrow morning, that's right, seven days a week, you are going to be inspired and motivated to succeed. I promise to bring positivity and inspiration to you for around 15 minutes each day. Thanks for listening and thank you for being a part of the Thoughtful Entrepreneur Movement.