ScreamingBox Technology & Business Rundown

Dave Erickson 0:03
Running a business is hard. So how can you run and grow your business while keeping costs low? We're going to examine the latest AI tools for business growth and operations in this ScreamingBox podcast interview.

Dave Erickson 0:36
In this fourth episode of our podcast interview series, we will be interviewing Garik Tate, co-founder and CEO of Valhalla.Team, who has a deep background in applying the latest AI tools for running and growing his businesses. Garik is an authority in leveraging AI tools for business expansion and operational efficiency. With his unique expertise, Garik has mastered the art of selling businesses, while simultaneously nurturing their growth. Through a proven process and company AI development and integration, hiring elite talent and operational automation. Garik enhances the value of the companies for his institutional buyers, while bolstering their profits. With over a decade of entrepreneurial experience under his belt, Garik has left his mark in the various sectors, including software development, outsourcing and publishing. Today, we'll delve into some compelling use cases and insight from Derek, shedding light on how AI tools can propel businesses forward in today's competitive landscape. Well, Garik, welcome to the ScreamingBox podcast interview, just so we can start by diving into the deep end. 10 years ago, most businesses didn't have access to AI tools for operations. Can you talk a little bit about how AI tools have been transforming business operations over the last few years?

Garik Tate 1:58
Yeah, AI tools at this point are really gaining speed. You know, last year, a lot of companies were experimenting with it, testing it out. But now I think a lot of people are realizing just how powerful it is. And the key reasons now AI can be used not just to classify data and predict outcomes at the end of the day, I think that's really still what it's doing, it's just doing it on a more micro level with a transformer architecture, but now it's actually being used also to generate content and generate data. And by generating that data, it turns out that you know, a lot of what we humans do, when it's boiled down to its inputs and outputs, which is not always easy, but when it's boiled down to its inputs and outputs, pretty much any step of any process can be automated now with AI, which is really, really incredible.

Dave Erickson 2:52
We're looking at kind of, two aspects of AI. One is improving operational efficiency, and the other is kind of growing the business. And although business growth can be tied to operational efficiency, maybe we can kind of divide it a little bit and talk about it in two sections. So let's talk about operational efficiency. When you are looking at a business that you are thinking about investing in or working with, from an operation standpoint, what are some of the things you see and look for in a business that you're going to basically come in and invest in?

Garik Tate 3:31
Absolutely. So the, there's one side is just saying, is this business a good one to invest in period? Do they have good growth opportunities? Do they have a solid plan, the founders, the right founders for the next stage of evolution, you know, as a company culture, et cetera, et cetera. But the level they go in a little bit deeper than than most investors and consultants is also, you know, does this business obviously have problems that I can solve? And when I'm looking at it from that point of view, I'm looking for bottlenecks inside of the business's operations that are hindering its growth. And specifically when I'm looking at those, those bottlenecks, I'm asking myself, is this a bottleneck where the AI both has access to the inputs and outputs, and be the inputs and outputs are clear. And if those two things are correct, then you can remove almost any any obstacle, or I should say any bottleneck to the business's operations, which for some businesses can really be it can double their business overnight. I have I have an example of that. Our sister company, which is a 2x U. It's an executive assistant company that provides executive systems and personal assistance to high net worth individuals. What was and is, is growing quite rapidly, they were able to find clients for that business. But we weren't able to find the talent on the recruitment side, we had trouble finding the right people rapidly for, for those, those positions. Really great IP, really great process once we onboard, folks, but it was, was being, difficult to find them. And we were getting plenty of applicants. But we had a bottleneck, we weren't able to process them fast enough. And so we had to go through the process of finding the inputs and outputs and then getting that into a consistent uniform data layer. And then once AI gets plugged into that, we're able to triple and I can talk a little more about exactly what that process was. But at the end of it, we were able to triple the number of hires we were able to do per month, which was able to double the business pretty rapidly by just removing that one constraint.

Dave Erickson 5:59
What are some of the operational tools that you find have been the most helpful when it comes to AI automation?

Garik Tate 6:06
The number one thing is really make.com. So if you know Zapier then make.com, is going to be just the next level better. But it's,

Dave Erickson 6:16
I use make.com. I prefer it over Zapier mostly because it's, it's more cost effective. But also I like the way it integrates all the different automations whereas Zapier’s interface I don't think is clean. That's just, but that's just my personal opinion.

Garik Tate 6:34
Zapier definitely has two two competitive advantages. One, it has a wider app selection. So I think that make will have 5000. I think Zapier has some like 20,000, or something like that. But the reality is that both have the 20 of the apps that you actually want to use. The other thing that Zapier I think is better at is that the UI is a little more step by step for someone who has no idea what they're doing. So if you've never done any integration, you might want to test out, get a free trial on Zapier and then jump over to make.com make.com is a little bit harder if you have no idea what you're looking at. But once you grok it, it's way more intuitive, significantly easier to debug. And it also allows you to do multi step integrations, which is really something that's necessary if you want to automate complex complex processes.

Dave Erickson 7:26
Yeah. Yeah. becoming an expert in make.com is going to be its own art form soon.

Garik Tate 7:33
I really do think so. And the, you know, the way that, you know, for those who are listening who haven't used these integration software's, essentially what they do is if a certain event is triggered on one application, like Gmail, or Google sheet or a CRM, it then will automate another action is taken on another platform. So maybe every time you receive an email, you want to log them in a spreadsheet, or every time an action is taken in CRM, you want to send an email report, you can integrate all these different applications together. And the cool thing about that is that you can plug ChatGPT directly into that. At this point, I'm recording this in March 2024. Just, Just recently, in early this year, it looks like some of the other large language models are starting to surpass GPT 4. For a long time they weren't, I mostly just kind of chalked it all up to noise. They know what GPT 4 is good enough. I still think it's good enough. But at this point, it might be I'm gonna have to do at another level. Another round of exploration to see which one's best cloud looks really, really interesting right now, though. But whatever large language model you choose, and I would recommend ChatGPT for the ease of its API, it's very easy to understand API, you can go on to the ChatGPT playground, if you just Google that. And you'll see an awesome interface for non developers to explore, look around, kind of Intuit it. And once you Intuit it, you can plug it in to make.com. And then you can automate prompts. So what if you receive an email, and every time you receive an email, you want to ChatGPT to draft a response? Well, hey, that might be a terrible idea. Because every, you know, it doesn't have all the context and that's when it's going to come back to inputs and outputs. But you could build something like that and you know, maybe every time we receive a an unrecognized email and you write in a certain way, or if you don't receive an email from from Microsoft, you know, draft in certain contexts, but you can you can automate drafting responses, you can automate, read anything with with this tool stack,

Dave Erickson 9:30
You know, determining, kind of sorting and if statement logic to emails to you know, respond to the easy ones that are easy to respond to and ones that don't fit into a certain pattern or have keywords that may require something you know, where it's kicked up and a human has to look at it. You know, those kinds of things are possible in the automation for sure.

Garik Tate 10:00
I love automating these, these steps when the inputs and outputs are clear, and I think that's a really important mindset to have when approaching these. So the way that I think about it is that everything that we love, humans to do, are, are things where the inputs outputs are still fuzzy. So, for instance, if you have an accountant, you that accountant, when it's giving you, when they are giving you advice is not just giving you advice, because they read your p&l statement. That's one input. But there are other inputs are, you know, their conversations with you, understanding where you are at inside of your inside of your lifecycle, understanding where your business is, at, understand your hopes, and dreams and fears and goals, and understanding all those points. And if you could tell a machine, all of those points give a lot of examples. And sometimes you don't have a lot of examples. Sometimes you're and you're the only, the only example out there, you know, one of zero, then AI is going to really struggle. But if it's a repeated process, where you do have lots of examples and data points, and you can extract those inputs outputs, then, you know, what you can do is really remarkable.

Dave Erickson 11:19
So besides make.com, what are some of the other AI tools that you find are useful for operations?

Garik Tate 11:26
So a lot of folks right now, actually in 2023, if they were building tools with AI would, would primarily be focused on chatbots. And that's not something that really interests us at all. And I don't think it's going to add a lot of power to, to business, sometimes it can, for sure there's definitely some amazing use cases out there. But for most non tech companies are, you know, are looking to,

Dave Erickson 11:57
Yeah. chatbots are, yeah, chatbots are great for like, customer service inquiries. But there's a lot of companies I think, don't do customer service chatbots very well. The companies that do do some chat bots with customer service very well and make it so that the people who are interfacing with those chat bots, find solutions quickly. It seems to be a real benefit. But I've seen some implementations of it where it was awful. And I'm trying to communicate to their customer service through their Chatbot. And it's a horrible experience. So it can be very good. And it can be very bad, right? Marketing chat bots on websites and stuff. You know, they have a use kind of. If you look at them, it's just providing some basic information that is already kind of on your website. But some people like to just ask the question and get an answer, then that may be fine. But I think, there was, for a period of time, particularly two or three years ago, more chatbots than anybody needed. Right?

Garik Tate 13:26
I think a lot of the more recent explosion of chatbots is a lot of just FOMO. It's a lot like, you know, I saw ChatGPT, and you know, I'm supposed to use AI in my business to stay ahead. So let me now use that. And it's from, from a phone point of view, as opposed to solving a core problem, too. But to answer your question, if you're looking to build a chatbot, I think the best tool out there is voice flow. The, what I would ask my clients, when they're, when they're looking into this is, you know, is this removing a constraint though? Do you have a point in your business where you have an, either an overwhelm of, of too many uses is an area where quality assurance is dipping, or consistency is dipping. And if those two answers are correct, then it might be the right tool for the job if it is involved the back and forth. Actually, you one thing that we did use it for though, was build internal HR chat bot, so users that, not users, team members that didn't understand our policies or need to get reminded on something we could chat our internal chat bot. And that was a fun thing that, that also was very, very fast to build up and to get the whole team to be, you know, thinking a little more in terms of AI and it was a good cultural moment for us.

Dave Erickson 14:30
So when you're looking at a business from a business growth standpoint, and say you come across a business where their operations are somewhat reasonable and scalable and they don't need a lot of fixing, but they're struggling to grow. What are some of the AI tools that you start bringing in or start thinking about using to help them grow their business?

Garik Tate 14:52
I'm a massive fan of Alex Hormozi. I'm sure a lot of your listeners have heard of him. He's right now, a pretty big member and a pretty big guru in the, in the business world. And I like his way of simplifying all marketing down to a set number of, of channels. There's earned media, paid media, that's, you know, where you basically build a social following versus paying advertising to get access to a, to a following. Then there's cold outreach, and then there's warm outreach. And there used to be two others of affiliate marketing, and I forget what the other one was, but he, he started trying to simplify it down. But there was those four as let's call those the four basic pillars. AI can be used in, with a very different mindset across those, those four types. So you know, AI has always had a huge piece of paid advertising, which is, you know, CPAs, and making sure that your advertisements are the right people. And that really hasn't changed too much. On the earned media side, AI has had a big impact in how much content we couldn't create. And that if you know how to use it well, and have a big increase in the quality all of a sudden, you can, you don't have to pay your large artists studios or built or paid large graphic design companies or other things, you just create great content. But the flip side is now if anyone can do that, well, the quality dips precipitously, so you, so we're creating new social norms and creating new social contracts on what isn't is not acceptable. And that, that also just boils down, I think you'll just whatever it gives the most value to the end user. And so if you can keep that quality high, then you're going to do very, very well. On the cold outreach side, this is actually something that we are building right now we're building a chatbot, not chatbot, now an automation system that scrapes data out there online on key prospects and scrapes data, enough to then customize better messaging to them. There's not a lot of great tools out there on the market, I would just suggest, so that's why we're building our own. But if, if you're interested, I think that LEM list and instantly right now are probably the two that are gonna crack this nut for most people, we wanted to build something that had more data points that collected then either LEM list or instantly would use but those are two good good apps for for cold outreach. And then warm outreach is a little bit of the same amount of content side and on the cold outreach side for scraping data. So little bit limit similar.

Dave Erickson 17:44
I'm curious, you know, a lot of the methodologies for warm and cold outreach is to supply the potential or the target with content. Sometimes that can be very short content in the form of questions or statements and trying to initiate a flow that results in a meeting or some business. But I think what's happening is a little bit, there's so much of it now flowing to people, and people have so little time to read, we're generating more content and people can read. Because obviously generative AI is helping and making it easy to generate a lot of content and people are just getting swamped with content. So what do you think that's going to create a problem, you know, pendulum swing back and forth and business. People are starting to get tired of reading all this content. Some of it is really bad to begin with, but some of it is reasonable. Where do you think the pendulum is gonna swing next? And how do you think outreach is going to happen if you know text and emails and that stuff starts becoming less effective?

Garik Tate 19:01
We've been here before, there's been multiple evolutions, especially as the work world globalized and, and you got access to different forms of different labor markets and different tools out there, where the marginal cost of creating content has continued to go down significantly, and now it's lower now than it certainly ever has been has been before. And so the way that I see that evolved, when this has happened previously is you know, society develops. And you can think of it almost like, like an immune system response, you know, society creates different either tools or norms to then filter out all that just the illusion of content that comes through. And the stuff that we lit through is the stuff that we want. So, you know, what do we want we either A. want to be entertained, we want a little dopamine hit inside of that moment, or B. we want something that will actually solve one of our short to mid to long term problems. And we, as business owners, as content producers, just have to up our game to understand exactly what are those problems that we're supposed to be solving, and then solve them more effectively for, nowadays, often it's really about solving the problem for free, you know, basically giving away all your secrets. So then people can really see you, understand you, and then they'll work with you or buy your product service if they don't want to do it themselves and that's really been, been our philosophy. And I think that's been a big shift in the last few years. With people like Neil Patel and Alex Hormozi are doing it.

Dave Erickson 20:46
Yeah, there's a lot of YouTube tutorials out there. And I think the role that those YouTube tutorials play is, they let people know how much work it actually is and then if they choose to do it themselves, they get to struggle through the learning curve, and try to figure out all the things that actually weren't said in the videos, which are almost as important as what is said. Or they decide, okay, this guy knows what he's talking about, I'm just going to hire them. Right?

Garik Tate 21:14
Exactly. And I do see another shift going on in marketing right now, on top of that, which is a lot more, what previously would have been value based content is now shifting over to more comedic based content or entertainment based content, which is just another form of giving value. You know, if people are on a social platform, because they want to be entertained in that moment, can you get into that that entertainment set so then just, you know, make sure your brand awareness is higher, but that that, you know, has a longer game and I don't think this will get you guys as serious customers. But it definitely is picking up steam and who knows might be, might be the next big wave in marketing?

Dave Erickson 21:39
What are some of the kind of the misconceptions or the myths surrounding AI tools for business growth? They're not a magic elixir, they all take some kind of work, but what are some of the things that you know, people come to you with, with misconceptions that they have about what AI tools can do for growing their business?

Garik Tate 22:18
You had, you had another guest on your podcast a little while ago, it was on AI, using AI for hiring. Do you remember his name? Chris, Chris Lin. Lim, he talked about I think very well on his episode, to say that, you know, AI is a very well educated, very well meaning intern. And so it's somebody who you need to give very, very clear instructions to, but they're quite clever, and they know a lot. So you can think of them as like, just, you know, came out of college, and they, they have an incredible memory. But you got to get them the level of, of instruction, you would to an intern and provide the same level of, of quality assurance on their output. And I see a lot of business owners, you know, hoping that it can stop them from thinking as hard, to essentially, it's just, you know, no one, no one wants to do work. And while AI does allow us to move faster, it has just shifted, the work hasn't made the work significantly easier, it just now, instead of the work being the difficult part, when you're making the content, and now is just about your taste. And I do think that the most, two most valuable skills moving in the future will not be our ability to generate things, it will be our ability to articulate them, articulate what we're looking for, and then the ability of our eyes to have the taste of what's good and what's bad. So it's gonna be our eyes and our voice that are, that are going to be the two biggest skill, it's not gonna be our hands anymore. That's how I think about it.

Dave Erickson 24:02
Yeah, I think another skill that's going to be important is the ability to detect AI bullshit, and the ability to read what isn't really actually true. Because AI can't really figure out if something is false or true. It's just taking a lot of content and stringing it together in a way that answers a question or talks about a subject, right? And sometimes it just happens to pick up stuff that you know, isn't real or isn't true, or doesn't fit the context or you know, and so, that ability to see what AI is doing and know Oh, that's actually it's not correct, or it's going in the wrong direction, or, you know, that editing ability I think will be really important in the future as well.

Garik Tate 24:51
That reminds me of the book Thinking Fast and Slow by Daniel Kahneman. In that book, and he's talking about system one, system two And the way they think about that as a system two the analytical, slower moving thing, often it's met its masteries, over top down thinking where it thinks logically and objectively, while system one is more bottom up, it thinks, empirically, you know, what have I seen before. And it's faster at making creative connections. Ai at this point is pretty much all system one. And, you know, you can think of system one is as a little more like, unconscious or more subconscious, you know, faster processing, and your subconscious, anyone who, you know, read self development, or Tony Robbins knows that your self, your subconscious is just going to accept whatever you give it as true. And it's your conscious mind, it's your analytical mind that has to act a little more like a watchdog at the gate, to allow certain pieces of information in order to reject certain pieces of information, so that you can be training your brain in the way that you want it to be to be trained. And it's the same with, with AI, it's all system one. So it's, of course, it's going to accept whatever we tell it, it's in a black box. So just what you said, we have to improve our ability to fact check or check primary sources or to be able to discern bullshit.

Dave Erickson 26:14
What do you think is going to be kind of the future trend in AI tools? How, how do you think AI tools are going to start evolving over the next year or two?

Garik Tate 26:25
I think one of the biggest things that we're right on the precipice of is much better multimodal usage. So translating an image to text and vice versa. And that's going to extend also to diagrams, spreadsheets, tables, there's already a lot of that, but I think that, that GPT five, and the next round of, of large language models are just going to take that to a whole new level. One thing that has, has happened, that is a little more still, in the research labs, it hasn't been as, as large as distributed the public is, AI is ability to take in very, very large data sets. So we're talking about like, the entire works of Harry Potter. And that is extremely helpful. Because, you know, as a consultant, if you take our accountant example before where they know everything about you, that's probably you know, gigabytes upon gigabytes of data, just every interview, also, every facial expression you've ever made every, you know, pause that was there, that's a that's a ton of data. And so if you wanted to, the AI to be even more effective, and you have to feed all that context in not just the explicit stuff for the implicit stuff. So I think that has had a big impact on our abilities to, to work with these machines. Other than that, I definitely was really surprised in 2023, by how good the open source models have become. I, at the very beginning when, when ChatGPT first came out, I very much saw like a lot of people that the billion dollar price tag and saw that while a few companies will definitely be competing and playing in this game. It's not as available to the regular to the regular folks. But some, some innovations and the release of the llama model from, from Facebook, the either the release or the leakage depend on how conspiratorial you are. But at this point, it's, it's pretty. It's definitely, definitely more just releasing. The release of the llama model and some, some innovations with I believe it was like the chinchilla white paper and the Laura white paper. I believe that that proved how, how much cheaper these models can be fine tune. So the open source community has been making leaps and bounds. And I think the big players always will have access to resources that will keep their stuff on, on the cutting edge. But for most businesses, if you're worried about how you know how private your data is, if you're just wanting to be in control of your data, or make sure that your model is very, very stable and will change based on new releases. I think that the open source models out there and the ones you can run on your own servers are significantly better than I would have predicted at the beginning of 2023 and will continue to be improving very fast.

Dave Erickson 29:27
Yeah, and obviously a lot of people are understanding this, which is why the server market out of Taiwan and China are doing very well because everyone's trying to buy AI servers, right, and that's why Nvidia stock is so strong. And you're gonna probably see a lot of AI servers that private companies and you know, SMBs are going to buy and use to run their, you know, AI and LLMs, right.

Garik Tate 27:57
One of the biggest concerns I do have in the future of AI, aim is that our, our current supply chains for, for these chips is very brittle. And it's very, very globalized. And there's a lot of links in the chain where only one company based out of one country can do it, and they only have one customer. And if any of those companies go away, like those five engineers in that one company like ours are technically irreplaceable, and everything will break down very rapidly. And I do see that with, that kind of system is not robust and not something we want to build a big future on. So when I think Sam Altman is seeing it the same way, which is why I think it wasn't just a clickbait title. But you know, $7 trillion, I think was a quoted price that he's trying to raise to create a new, a new supply chain, and I think you can, you can do it for a bit cheaper than 7 trillion, but certainly is a trillion dollar problem, for sure. And so I do think that

Dave Erickson 31:05
Well he needs 7 trillion to ensure that whatever he builds will have enough longevity, to not create the problem he's trying to avoid, which is a brittleness in the, the supply chain. Yeah, he's got a need to develop a product, and you gotta need enough money there. So that he can ensure that product will be around for 30-40 years, right? And matter what the economic situation is, right? So he does need a big money buffer. Right.

Garik Tate 31:37
I hope, I hope he succeeds on that one. I'm still still not convinced seven, seven is the number I think you could do maybe half a trillion, but but but that's, that's come on my ass. But,.

Dave Erickson 31:51
You know, those people, they love throwing around money for expensive engineers and buildings, and nobody's gonna want to go to.

Garik Tate 31:59
So if you're if you're in college right now and looking dive in the software world, you know, I think that the heart, the more physical engineering sciences are going to probably be growing just as fast if not faster than the Digital hard sciences. And that's probably a good thing for just civilization advancing because wherever we build them, software has to run somewhere. So I think more smart minds attacking that problem is definitely a good thing.

Dave Erickson 32:26
Yeah, and I think countries are now starting to realize that that's a resource that they want, may want to keep in, in house or in their country, they may want to establish their manufacturing to be able to at least supply themselves. And

Garik Tate 32:41
we're going through a phase of like deglobalization

Dave Erickson 32:43
I think so, I think the pandemic woke a lot of people up to the fact that very long supply chains are not necessarily the best thing. They, they are susceptible to pandemics and natural disasters and, and all that and, you know, we went for decades, in the 60s and 70s, having relatively short supply chains. And I think now they're trying to move back to that. Companies are starting to realize, okay, you can't centralize all your manufacturing in one place, it's better to spread it out. And you're seeing that in Asia, you know, India, Vietnam, Thailand, they're all becoming much stronger in manufacturing. So it's not just centralized in China. But I think you'll find it in other places Africa is making actually pretty big strides in, in developing their manufacturing in IT and technical prowess so I think you'll see it kind of moving around and supply chain shortening up in the future.

Garik Tate 33:43
Yeah, if your listeners are interested in learning any more about this I definitely recommend the videos from Peter Zeihan if you if you type that into the YouTube he does some some interviews and podcasts, but he also does a lot of short form videos where just talks about some different subjects. I definitely don't agree with everything he says. He's definitely a heterogeneous thinker. He's not a he's not going to give you the same views as a lot of peer strategic thinkers, but his stuff is very interesting, I think it does. Often predicts things before they get into the rest of popular media. So if you want to be on kind of cutting edge of some things that you don't hear from too many other sources, I think he's an interesting guy to check out.

Dave Erickson 34:24
Well, Garik, this has been a wonderful interview. And for our listeners who would like to know more about Garrett, please look at the description. We'll have links in there, and you can contact him if you want to discuss any more about these subjects. Anyways, Garik. Thank you very much.

Garik Tate 34:42
Thank you, David. It's been a pleasure.

Dave Erickson 34:50
Thank you very much for taking this journey with us. Join us for our next exciting exploration of technology and business in the first week of every month. Please help us by subscribing, liking and following us on whichever platform you're listening to or watching us on. Weibo. If you enjoyed this podcast and please let us know any subjects or topics you'd like us to discuss in our next podcast by leaving a message for us in the comment sections or sending us a Twitter DM till next month, please stay happy and healthy.