Garik Tate is a renowned expert in the application of AI for business optimisation. With a passion for simplifying complex concepts, Garik has a knack for making AI meaningful and accessible. His practical insights and real-life examples demonstrate the transformative power of AI in streamlining processes and driving business efficiency. As the go-to advisor for small business owners seeking to enhance their business value through AI, Garik’s expertise is invaluable for those looking to make informed decisions and harness the potential of AI for their businesses.
Diving into the realm of AI, Garik Tate was captivated by the concept of enhancing the intelligence of computers. His journey unfolded as he recognised the profound impact AI could have on small businesses, particularly in optimising processes and driving efficiency. Embracing the intricate nature of AI’s problem-solving methodology, he discovered its potential to revolutionise traditional business operations. Garik Tate’s narrative sheds light on the remarkable synergy between human ingenuity and technological innovation, showcasing the transformative influence of AI in enhancing business value. His unique insights offer a compelling perspective on leveraging AI as a catalyst for growth, inspiring small business owners to embark on a journey of innovation and efficiency.
In this episode, you will be able to:
- Discover how leveraging AI can skyrocket your business value.
- Uncover the power of AI automation for turbocharging business growth.
- Learn to identify constraints and optimise AI for maximum impact.
- Explore how AI enhances recruitment processes for top talent acquisition.
- Uncover the secrets of streamlining operations with AI implementation.
Introduction to AI and its Application in Business
How does the concept of artificial intelligence fit into the business world? Well, think of it as a forward-thinking assistant, always ready with data-driven insights. AI pulls in vast amounts of information, sifting through it all to spot patterns and generate solutions based on those patterns. With this ability, AI can offer businesses a solid foundation for making decisions and strategies. Moreover, AI can automate repetitive tasks, freeing up time and focus for the more creative, human aspects of running a business. In the conversation, Garik introduced AI as a key aspect of advancement in the computer science field. He likened it to the workhorse of this information age, equating its role in handling data and patterns to how tools enhanced physical work in the past. When Darryl compared AI application to chess playing, highlighting human foresight against AI’s pattern-based approach, Garik extended the discussion to business contexts, presenting AI as a way to augment businesses, automate tasks, and drive value.
Letting Go of Tasks and Automating Business Processes
Many business owners find it hard to let go of tasks and delegate, but in the pursuit of growth and innovation, this step is critical. Assigning those monotonous tasks to automated systems like AI can free up a lot of time, leading to enhanced efficiency and productivity and, ultimately, to greater value in the business. During the conversation, Garik emphasised the necessity of systematising and automating processes in businesses. He advised business owners on the art of delegation and discussed the potential of AI in automating work. He floated the idea of AI as a middle step to improve efficiency and gave an example of how AI could blend into existing tools to enhance productivity. Garik’s recommendation of using an internal champion to master and apply system thinking put a practical spin on his AI advocacy, ensuring business owners could glean actionable insights from their chat.
Leveraging AI for Business Growth
When it comes to business growth, AI can be a gamechanger. Despite not being a silver bullet solution, it can unearth insights across industries that can lead to impactful strategies—in other words, AI can help businesses leapfrog their competition. AI’s capacity to process millions of data points quickly and efficiently helps businesses identify limitations and potential areas of improvement, which can then be streamlined, optimised, and scaled effectively. Building on this, Garik spoke about the power of AI in driving business transformation. While engaging with Darryl, he emphasised the importance of identifying operational constraints and business needs. He advocated the use of AI to free up resources, smoothen bottlenecks, and accelerate growth. While discussing the impact of AI, he shared the story of a remote outsourcing company that used AI to refine its recruitment process and consequently experienced significant enhancement in business growth.
Watch the episode here:
Welcome to the podcast that’s dedicated to helping business owners prepare for exit so you can maximise valuation and exit on your terms. This is the Exit Insights podcast presented by Succession Plus. I’m Darryl Bates-Brownsword, and today I’m talking to Garik Tate. Now, Garik is, I guess, someone we really need to be talking to in today’s world, business world world, because he knows all things. My what I’m going to have to do today is get Garik’s complex brain and distill all the information in a way that makes it meaningful and useful to someone like myself.
And hopefully that’ll help you listeners as well understand and how we can take advantage of AI in today’s world. So welcome, Gareth, and thanks for joining us today.
Glad to be here, Darryl. Thank you.
Cool. Now, Garik, why don’t we start by getting an understanding? Because I think AI could be a bit like bitcoin. It’s one of those terms. It’s new and it’s burst onto the scene and taken over our lives, but I just wonder how many of us really understand what it is. So let’s start by getting a bit of an understanding of what AI is, and then that’ll help us identify how we could take advantage of that and make it useful in our own businesses and where it is beneficial and what those benefits are, if that makes sense.
So what is AI?
So AI is a field in computer science, and it’s essentially the study and the engineering of how do you make the dumbest thing in the world, which is a computer, how do you make that smarter in the purest sense? And it’s based off of a lot of things we’ve learned from neuroscience and from philosophy of cognitive mind. But fundamentally, at space, it’s just making computers that are smarter and can think well, honestly, more human, even though we’re finding that there’s a lot of things that aren’t so human about them.
And I guess it’s the appearance of being human, because I think you’ve started on a really important point, is that computers are pretty dumb. They are just a machine. They don’t have a mind. They don’t have a soul or a personality. All they do is have software, and they get inputs and they process that input, and they produce an output, and then they’ll get some feedback and adjust it as required. And the more processing power they have, correct me if I’m wrong, this is what I’ve learned so far.
But the more processing power they have, the more computations they can do, and therefore quicker and therefore appear, or they’re beginning to appear human like, but they still got some catching up to the human brain, as I understand things.
Yeah. The big innovations inside of AI has been definitely from the bottom up approach, which is that it is able to make inferences from a bottom up sort of level, but not so much from a top down level. This is why when you ask it, like, to play chess, it’ll say chess moves, but it forgot what the previous moves on the chessboard were. At least the current generative models are doing that, I believe, called deep blue that fought. Gary Casperov is obviously a very different AI, but a lot of these AIs appear intelligent, but it’s kind of an illusion.
So that’s a really helpful way to explain it, I think. Or just for me. So if I think of a game of chess, I’m thinking, what move am I going to make next? Now, if I make that move, he’ll probably make that move, or he could do that move, or he could do that.
But if I do that move, and then I do that move, and then I do that move, and hopefully I can set it up to do this, this and this, and therefore I’ve got him depending if he plays the way I want him to play. Now, I’m not a great chess player, but that’s the way the mind works when playing chess. And it is saying that a computer can’t have all that. Think of all those steps forward and think of the way that the opponent’s going to react. It just has to look at all of the pieces on the board at the moment and has a look at all the possibilities to move in that exact instance. Is that what you’re saying?
There’s a lot of types of AI out there and a lot of ways to engineer it, but these generative models. So when you say chat GPT, the G stands for generative. The way these generative models work is very much from a perspective that I call the bottom up approach, which is essentially that it reads massive amounts of data and then it finds patterns in that data, and then it can play those patterns out. So when it says, what’s the next word in the sequence? It’s looking at the previous words in the sequence and then going off of that.
So when you’re playing a game of chess, you’re seeing that as a simulation that’s present in front of you. But the way the AI is playing it is it’s just reading the previous moves and it’s making its best guess of something that sounds right for what comes next.
All right. And for the purpose of this podcast, that’s probably enough of a definition, because this is not fair enough, going deep into AI. So again, you’ve set it up really nicely for us to go.
Here’s what AI is now in terms of business. So the people listening to this podcast are business owners who probably got their business from ten to 250ish people. They’re at that stage where they’re thinking, hey, I need to make my business more valuable. I want to increase the valuation of my business in anticipation of an exit in the next two, three, five years, thinking about how do I make my business really attractive so that it can be acquired.
Yes
AI has burst onto the scene.
It’s got all sorts of promises of making our lives easier, being reliable, I guess, removing a whole lot of repetitive tasks and doing those tasks reliably, meaning that we can perhaps redeploy some of our employees to be doing more meaningful work that AI can’t do at this stage and just make sure that their AI tasks are running along smoothly. Is that the sort of way that business owners should be thinking at the moment about AI?
Absolutely. The way that I think about it is if we take a historical perspective here, the human body for the last 10,000 years has been the ultimate work machine for most of human history, caveman days and such. But once we were able to boil down certain very useful tasks into kind of a groove or into a systematised way, we were able to deploy horses, we got horsepower, we’re able to make machines and steam machines and all of these tools and technologies and other creatures in order to do that physical work.
Our body is a platform that’s versatile, but it’s not so specialised in any one thing, luckily. And what AI now coming on the screen is really just a continuation of it. AI is the modern, it’s the workhorse of the information age. So once you are able to groove out, okay, these are the inputs, and we’re going to be talking more about that soon, I’m very sure. Once you boil down, okay, these are the inputs, these are the outputs.
AI, and you have enough data to train it. AI can really take over any repetitive task that you want it to take over and do it exceedingly well depending on how much data you can feed it.
So let’s not mess around. Let’s get into that. We’re exploring around how business owners can exploit AI, and you’ve just alluded to, hey, look, we need to give the system some sort of input.
We need a whole lot of data, a whole lot of pre information, I guess, or pre experience of how we handle that.
Examples.
Yeah, examples of how we’ve handled that input and what the possible outcomes are. So the AI can go, right. Well, here’s the outcome I got in this time.
Let me compare it to all the other outcomes, and therefore I know what to do next. I’m making it sound like it’s thinking for itself, and maybe we’ll end up there one day as well, if that’s what it’s doing.
Have you got some real life examples of how business owners can sort of do that? Let’s bring it to life for people so that they can perhaps stimulate some ideas for them moving forward.
Yeah, absolutely. There was one company we supported that was in the remote, the remote outsourcing industry. They provide executive assistance, and they have a very good USP, they have a good target market, and they were able to really be selling it like hotcakes at a very premium price point.
But in order to justify that premium price point, they have to make sure they’ve hired the very best personal assistants to execute on. And so their big constraint was, how do we find those people and find them quickly? And so this is the very first question that I walk my clients through when they’re asking, how do we apply AI? We always take the theory of constraints seriously. Theory of constraints was introduced by Eli Goldratt in his book the Goal.
But it’s basically saying you have to make sure you’re attacking the right problem. You break down your business into a series of interlocking pieces. I’m sure your mechanical background, you know a lot about that hand systems thinking, and you find the part of the business that is the largest constraint, and you focus on solving that. So in the example businesses case, that was definitely the recruitment. And so we looked at the recruitment as a series of interlocking steps, asking, where are the handoff points? Where are the points that are taking the most amount of time, where the most number of errors are cropping, where the most number of drop balls are occurring, where are those issues? And once we found those, it turned out it was all at the very top of the funnel. You know, once somebody had gone through a couple of interviews, the recruiters handled them very well. The experience was very positive. We were able to get people through the funnel, but the very top, we were making lots of mistakes.
And so luckily, that was a pretty simple part of the funnel as far as the input score is concerned. We got their cover letters, we got their CVs, we got their information from their application form and so we were able to use AI to read all that data. And then first of all, actually just making sure that the AI properly moved them to the next stage of funnel. So replied to their emails, had them fill out the next steps of the process. And as time went on, we also started adding AI in order to read their resumes, read their cover letters and give suggestions if they were a good fit or a not so good fit for those positions.
And after we implemented this, it took about three months. After it was implemented, the recruitment was able to triple the number of people they were able to hire every month, which allowed the business to about double over the course of that year.
So you’ve got my mind racing, Garik, and let’s play with that for a sec. So my understanding of the recruitment industry is that as much as we don’t want it to be, I guess it’s kind of emotional. And what I mean by that is that applicants go to a whole lot of effort in producing their CVs, their resumes and trying to stand out, because they know that for a lot of jobs, hundreds if not thousands of applications get put in and they want to stand out from the crowd. So to do that they need to do something, some sort of emotional self positioning to jump out from the crowd so that their CV gets read by the person and that they get added to the shortlist and they aren’t rejected at stage one. If I’m understanding what you’ve just suggested that you’ve used AI for, is to go, well, let’s do that process. And just instead of reading and having some sort of emotional or feelings attachment to a cv or application that you’ve seen, the AI is conducting that process for us and they’re just purely looking for the facts and the experience and the data without emotion and then just filtering all of the application from that perspective and handing them on and doing all the admin and legwork for the recruiters to then take the applicants that really qualify based on real experience, and it’s kind of hard. Yes or no? Is that what happening here, which is going to change the way people are writing their application letters, I imagine at some point.
So what you’re describing is how do we remove the bells and whistles and get to the real meat of an applicant’s qualifications? I think that AI is very capable of doing that.
These generative, pre-trained models, though, they’re not really emotionless. I like to think of it as hypnotising them you can make them act angry, you can make them act sad, you can make them act happy. In fact, actually with, it’s called jailbreaking them. When you prompt, hack them so that you can make them do, let’s say naughty things or tell you how to build a bomb or whatever. One of the ways that they figured out how to do that is you make it deliriously happy, then it’ll be more likely to give you illegal information.
Like getting something.
Truly, it is exactly like that. And the reason why I think it’s exactly like that is because these models have been trained off of us fundamentally. I think a lot about what we’re doing with AI as a type of mirror. It might be a fun house mirror where it’s kind of warped and it’s doing its own thing, but it really is reflecting back onto us.
It’s our data from the Internet represented back to us. And so I don’t want to oversell that the current models of AI are doing that, what you’re describing, naturally. But the thing is, realistically they absolutely can, because you just make it a multi step process where you have two separate AIs. The first AI extracts from the resume, because of course the humans are still going to sell themselves. So you have the first AI say, hey, no flot, sum, nothing else, just extract the data points, convert it into a bullet point list, and then you have a secondary AI read from that bullet point list.
Of course you’re introducing two points of failure there, but that would be a way that you could easily use AI to make the process more objective. And that’s maybe something we should be doing. In our case, we were mostly comparing to did the human who read the resume think it’s a good idea, and did the AI think it’s a similar idea? And we checked to see when they would disagree and figure out if management agreed with the AI or management agreed with the human. And we use that as a way to calibrate and judge if we were happy with it or not.
But what you’re describing here would have taken a little more time, but maybe is what we should do in round two.
Okay, so maybe it’s a little more evolved that I’m thinking, and I think what’s going on here is I’m allowing my personal interest in the subject to derail us from where we should be heading with the podcast. So why don’t you tell me and share with us what are business owners, how they’re using and looking at AI at the moment. So today, in your experience of what they’re doing out there in the marketplace. What are they getting wrong when they’re playing with AI or trying to apply it or integrate it into their business?
Yeah, I think a lot of people are just being motivated by FOMO. A lot of entrepreneurs and business owners out there, they see it’s a hot new thing, and then they don’t want to be left out, so they’re diving in. And for sure, a lot of these new technologies, new AI tools, are being built every single day. And some of them are very exciting. But I think the excitement of what the technology represents and of the fad of the moment can motivate business owners to be prototyping it and testing in their business without always asking, is this actually solving our core problem?
And so I think that that’s the number one thing that business owners have to ask themselves. What’s the core problem in my business? What’s the constraint that’s holding me back? And then solve that constraint. And then an AI mindset is just opening up a whole new set of tools.
You can solve that by better training, by hiring more people, by restructuring your business. But the AI mindset and the automation mindset is, can we solve these problems with AI? And because these tools are so versatile, the answer is more often yes than we would think in the day of knowledge work.
So is the risk that people today are trying to use AI for the sake of using AI, whereas what I think you’re saying is, let’s get in and have a look at the businesses and go, look, there’s a real problem here. Oh, look, here’s the problem we’re trying to solve.
Oh, and by the way, you could probably use AI to solve that and save a few steps and become a more efficiency. Exactly. Or redeploy someone. This could replace a human doing that, be more accurate, more reliable, a whole lot faster, which now means you can use that person in a more valuable role elsewhere.
Exactly. And I think that when you’re a smaller business, let’s say between three to seven employees, probably your business is going to be a lot more augmented by just the humans using AI to do their work more effectively. You don’t have so many systems and processes that you want to automate. But once you get to this scale of, say, 30 to 100 employees, you’re small enough where there’s still a lot of communication across departments, where a lot of innovation can bubble around, you’re not calcified, and you can move faster than some of the bigger enterprises out there, significantly faster. But on the flip side, you have enough of a systematisation, you have enough of a process that taking a hard look and asking yourself, can AI speed some of this up? Can we redeploy? Resources is going to probably have a lot of low hanging fruit, and that’s where we can’t specialise.
Yeah. And what about automation, Garik? Are the two hand in hand? Are they one and the same. Help me out here.
They’re definitely different disciplines, and I think that it’s a step forward to be seeing AI in the context of automation. So the AI mindset is just looking at AI as a new tool, and in many ways a superior tool in our toolbox, in order to automate more things than before we ever could. Going back to the previous idea of AI is a modern day workhorse of the information age, because before a horse could pull the wagon, but it couldn’t do your taxes well, now AI, there’s enough data, enough examples, enough clear inputs and outputs. AI can. So that’s the idea.
That’s one hell of a horse.
Yes, sir.
So let’s get practical. So, for a small business owner out there who’s in that, I guess, that ten to 30 business bracket that perhaps can spend a little bit of money to work on AI as a solution, what sort of things should they be looking at to go, oh, maybe I should get some help here, and maybe AI can solve this problem or help me perform this task or this process better, or we just really benefit from automating this step using some sort of AI process. And that would help make my business more efficient and therefore more reliable and end up being more valuable.
Yeah. So the first thing is on the constraint of the business, where in the business are we being held back by? And this idea originally came from manufacturing, but I think it applies to all businesses. So it started in manufacturing to say, if you have a series of steps like assembly line, where you’re building a car, and in the 7th step it’s the slowest. If you’ve improved steps one through six, you’ve made your factory worse, not better, because your constraint was step seven.
So as it applied to the remote executive assistant company, or executive assistant and personal assistant company that we supported, if they had more sales, they were actually in deeper shit than if they had less sales at that point because they were already overtaxing the recruitment department. And so they actually kind of had to reengineer their mindset and turn off some of their prospecting because they just had to fulfill. So this is an idea that I think any business owner can respect, even though it came originally from manufacturing. But let’s say that you’ve isolated the correct constraint. The questions I like to ask is one, is, is this an area that we want to free higher capacity?
Or. Let me rephrase that. Once you’ve found the constraint, you can ask yourselves, where are the people spending the most amount of time, where is costing us the most amount of money, and also where is the highest number of errors occurring? And once you get that part, the next step is probably going to be the hardest, which is isolating the inputs and the outputs, because very often the step is inefficient because there’s some messiness. So, for instance, when we were at the top of the recruitment funnel, a big part of the problem is that we were sometimes reading the resumes when the email come in, and sometimes we’re reading the resumes after the application form was filled in.
And sometimes we have to double read it. And sometimes it would be unclear which recruiter would be the one reading it, and it would change from role to role. And so in the first 90 days, setting up the AI was actually remarkably easy. The first thing was making sure that the data flowed in a uniform way. And this is where using tools like notion airtable or a CRM, I think people are very familiar with CRMs HubSpot, and creating good practices around how your data is controlled is going to be the first step.
In many cases, the hardest step. Once that part is established, adding AI on top of it is often kind of the easy part. And you can get a lot of quick wins from it.
And am I hearing the do it once strategy, like, if we can remove a whole lot of rework, what was happening is people were going back because either they didn’t record the information, or it wasn’t captured properly for this step as it was for that step. And if we can just process every piece of information once, well, that’s going to save a whole lot of time.
And let’s do it completely the once and to get it right first time, then we save on doing rework, and that’s always going to benefit every industry. And what’s just come to mind, as you were speaking then, Garik, is one of the things that we do is looking at the capacity of a business. When a business says, hey, look, we generated 5 million pounds of revenue this year. Doesn’t matter if it’s pounds, dollars, whatever it is, we go, okay, so that’s what you did. You generated five mil in revenue.
But given the capacity, the number of trucks, the number of machines, or what have you, if they were all operating at capacity and you sold them at your normal price, and this can apply to services and people as well, what could you have generated? Like what’s the capacity of your business? You did five mil, and that sounds great, but if you had everything operating at ideal efficiency, let’s say 70 or 80%, would you have generated five or would you have generated six, or would you have generated nine? And there’s the capacity management that we do and monitor the capacity. And then the correlation I see with what you did is we go, okay, so let’s match or optimise your business and go.
If the capacity turns out to be ten, let’s make sure the capacity of the sales team is ten. Let’s make sure the capacity of the operations team is ten. Let’s make sure the capacity of the delivery team is ten. Let’s make sure the capacity of the client service engagement team is ten. And then it’s optimised across the whole business and it’s uniform, which is what I think you were kind of saying.
You’ve just given me, I think, a real insight into the application and how we can apply and how business owners can start looking at their existing business and going, where’s the opportunity for using AI in my business? Let me have a look at the various functions across my business. Where am I unbalanced? This one’s significantly lower and could boost up. Can I use AI to boost that up or boost that capacity and match the other functional areas capacity of my business?
And there’s a first step. I don’t know, maybe I’m being a bit.
No. So once you find that department, or that step in your process where the capacity isn’t a ten, let’s say it’s a seven, then ask yourselves, how can we add AI here so long as your data is clean, which is definitely a big if, then AI can more often than not increase the capacity there. And the way that you would go about doing it is, let’s just get ultra practical here is remarkably simple.
There’s tools out there. I think a lot of business owners might have heard of the tool Zapier. And if you haven’t, the long story short is that it takes a trigger from one platform. Let’s say you receive an email and it does something on another platform. So let’s say you receive an email and therefore it puts that email into a spreadsheet.
Kind of a silly example, but it can do anything like that you post a YouTube video, you automatically post a tweet so that the tweet promotes a YouTube video you’re about.
Yeah, there’s a practical example of automation. We’ve got some sort of tool that says if this happens, then automatically do that for me. And it’s a repetitive behavior that a human used to do, and we’re now using a bit of, I don’t know, I guess Zapier is AI. Does that come under the banner of AI?
I would call that automation.
Yeah, Zapier. Is an automation tool which saves hours of time of repetitive behavior. And every time you’ve got repetitive behavior, there’s a chance of making a mistake or something gets missed. And there’s where optimisation comes in place. Thanks. I’m just catching up to you, mate.
That’s perfect. And there’s an additional concept I want to introduce here before I talk about how you add AI into that. But the additional concept is one I got from Tim Ferriss, which is a concept of the compounding effect of free time.
If you’re at like 99% capacity, that’s too much capacity. But if you get to the point where you’re at 98% capacity, well, you have 2% left over to start working to get to 97, and they have 3% and they start going to 96. So basically, the more slack and capacity you have, the more of that can be applied to automating and cleaning things up. And so I, early on in my automation and AI career, very often would find things a little bit too trivial, like, okay, that’s going to save us, what, 15 minutes per week? Is that really worth it?
And the answer was yes, absolutely. And I had the same experience hiring a personal assistant. There’s a dozen five minute tasks that added up to an hour, and I never thought they would really matter. But by golly, freeing up that even just 1 hour across twelve different five minute tasks, I had the capacity to start automating more and more and more. And so this can go too far.
It absolutely can. I don’t want to say it’s an unquestionable virtue, but we shouldn’t underestimate compounding free time.
Absolutely. And it’s, again, another area I look at when I’m working with clients. Business owners can be control freaks and they can get involved in everything. And one of the things we’re looking at is to remember the context of we want the business to be worth more so that they can exit it. And if they’re involved in doing things themselves as control freaks, then it’s not worth as much. So one of the things that we get business owners to look at is we go, let’s have a look at all the tasks you get involved in during the week, and let’s have a look at what you get paid or what you should get paid. And let’s break that down to an hourly rate and then we’ll start matching any of those tasks you’re doing now. And if you could pay someone less than the hourly rate you pay yourself to do them, they’re the first tasks that you should be handing off.
And that is, in hindsight, as simple and as obvious as it sounds, it’s just getting it to that level of granularity and oh, 30 pounds an hour, 100 pounds an hour, whatever the number is, is a trigger. And it just helps the business owner to go, oh, yeah, I should be letting that one go. And then we go, okay, so let’s train the person to do it the way you do it. And once they learn how to do it your way, then we know you’re going to be happy. So that way you can delegate, know it’s going to be done the way you want it to be done.
And then you need some sort of feedback mechanism, weekly report telling you how many times it’s been done and how successful it was or whatever sort of feedback you need to give you comfortable that the process or the task is still in control even though you’re not doing it. It’s those granular steps that we have to take to help business owners let go of tasks and help the business become even better without them, so to speak. And it’s a painful exercise sometimes. And then I’m guessing the next step in that is using some sort of automation tools and going, well, why even get a human to do it? Let’s just delegate the task to automation.
Free up your leadership team, get them to start doing higher level stuff. Yeah,
Exactly. So where does that leave us? You’ve shared some ideas of how you’ve applied this. You’ve given us some concepts of what it is and where the opportunities are to start looking.
How do you decide, Garik if a business owner’s gone? Oh, look, here’s an area of automation that may be a good option for AI to do. How do you make the decision if it’s worth investing, exploring if AI is going to be a good solution in that particular scenario?
So the number one thing is I’m going to answer that in two parts. The first, answering your question of how do we know if it’s a good area? And the second is, then how do you actually do that practically? Let’s say you already have some Zapier set up, and how do you do the next steps? So the question of, is this a good area to use?
AI is in western civilisation, we put a lot of onus on the concept of accountability. And if something has consequences, we very much want a human to be accountable, and that has a lot of virtues attached to it, and I think it’s something that we should respect. Ultimately, we can’t make the AI accountable. Well, maybe that’s taking a little bit too far. Maybe we can, because we can always change it if it makes a mistake.
But we still have to make a human responsible for double checking the AI’s work.
So AI can’t be accountable, which does pose some limitations to the kinds of things that you want to put it in charge of. At the very least, it changes how you transfer the ownership to it. So what we do is if it’s a step of your process that is high risk, but maybe also high reward, the first thing you do is you just have the AI running in the background where you can double check its results and compare it to how a human would handle it. So, for instance, in the CV screening, it’s kind of risky if we say no to the wrong applicants, or we say yes to the wrong applicants, and then we waste people’s time or we lose amasing hires. So we still want to make a human accountable for that.
And so we just log the machine’s pass fail in a place a human can’t see it. And then we just bring up both logs, how often the human and the AI match. And then when they disagreed, we just looked at those one at a time, and we asked ourselves, because the human makes mistakes, too, let’s not pretend like they’re the gold standard. We look at those and say, okay, are we happier with AI or not? And then we train the AI from that point forward, which really just means we adjust a prompt.
I say train like we’re doing something fancy, we’re not. It’s the same thing as doing Chat GPT and just modifying part of your prompt and running it again.
You’re just updating. That’s how we think about the risk.
I remember when I went to Uni and I studied engineering, and I first learned about systems. And what you’re describing to me is basic systems thinking. And I remember someone I worked with in early days.
She said to me, Darryl, everything is a system, including your stomach, which just processes information and rejects it and feeds back and what have you. And I’ve gone, oh, gross. But what you’re explaining to me is just like any system’s thinking. You’ve got an input and then you measure the output. You monitor the output and you go, is it the output we wanted?
Okay, I need an adjustment, and there’s my feedback. So I take a feedback mechanism from the output and I redirect that into the input and I go, okay, so there’s my input. I’ll adjust what’s going into the input, goes through the process again. Does it give the output as expected? Oh no, it’s too high this time.
And it’s exactly how a thermostat works on an air conditioning system or a heating and cooling system. It’s just, is it the temperature we want? Yes. No. Is it higher or lower?
Okay, will I make an adjustment? Okay, through the process? Higher or lower? And I think what you’re explaining or where I’ve arrived at because I’m being a little bit slow today, is that AI is just, well, is a highly sophisticated control system with feedback and adjustments being made.
I think that’s how businesses should think about it. I think AI can be a lot of things in a lot of domains, but I think that is ultimately the most practical sense for most business owners.
Compounding effect of free time.
Yeah, exactly. Good, nice. So let’s think about systemising our business.
And we’re not just using human systems, we’re going, well, since we’re going to be looking at systemising our business, let’s take the opportunity to go, is there an opportunity for us to use AI or any form of automation? If we can use automation, then that’s going to be even more effective, which means we free up the time of the people to be doing more productive, more valuable work, or more valuable tasks for us.
Yep. And to make that just a little bit practical.
Yeah, please.
I just want to give a couple of the steps that gets practical. So on tools like Zapier, like we said, it can trigger information from one platform and then do something on another. Platform. The way that you add AI into that is you just add a middle step. So you get information from one platform and then you trigger something.
With AI, it’s the same thing as going onto the chat GPT and typing something into it. Let’s say you get an email, it triggers, you paste it into chat GPT and you say, hey, here’s the information about my website, here’s the static information about me. Here’s the static information about what we do. Can you draft an email response? And then the output is it puts that email response onto your Garik.
So what if every single Garik you had had an automated response written to that email? Honestly, it might not be that effective because you don’t have all the inputs of every part of every conversation you’ve ever had. But that’s an input problem. It’s fundamentally an input problem. So you can add AI into the middle steps. And I would recommend if people have used Zapier and they find this a little bit unwieldy, I definitely recommend the tool make which does multi step automations a lot more effectively. So if you’re a business owner and you have your team looking into this, I would suggest make as a tool that your team works. And I definitely recommend business owners don’t try to do all this themselves.
They find the right champions internally to master these tools and apply the system thinking.
Brilliant. Garik. I know that this is a really big topic. It’s going to continue to grow. We could probably revisit this episode every six months, or even every three months or so and update it. But given this is our first attempt at exploring how SME owner managed business owners can explore how AI can help make their business more valuable? From this conversation today, what’s the one key thing that you’d really hope that business owners or listeners to the podcast take away from our conversation today?
I think the biggest thing is to see AI not as a silver bullet, but as a very helpful personal assistant. I like to think of it as a very eager, fresh grad that is very well read.
They know a lot of things. They learn everything in college, but they don’t really know how the real world runs. But they’re very well educated and they’re very eager to learn. And if you train them properly, you give them lots of examples, they won’t get tired, they won’t get pissed off. However you treat them, they’re going to reflect that back to you and they can do some crazy cool things in your business, especially if you have dull parts or parts that take just a lot of manual effort in the knowledge space and to get kind of excited by how much time you can free not just of yourself, but also your employees and start getting the compounding effect of free time onto your site.
Garik, that’s brilliant. Thank you. Thanks for sharing your insights with us today. I appreciate it. And I hope everyone else found it as valuable as I did.
Thank you, Darryl. This is a lot of fun. Thank you.
About Garik Tate
Garik Tate is an AI Futurist, Investor, and Automation Expert. His expertise lies at the intersection of AI, IQ, and EQ. He frees up business owners and makes their companies more valuable with AI and automation.
With over a decade of experience as a successful entrepreneur, Garik has founded companies spanning Software Development, Outsourcing, and Publishing. Notably, he has led and nurtured organisations to a team size of 75 employees.
Garik is a sought-after speaker on AI’s future, and its practical application in business. His expertise and passion for the subject have earned him a reputation as a respected voice in the AI community, and he is often sought out by companies looking for advice on how to leverage the power of AI to drive automations, growth and success.
If you would like to learn more about how to start preparing your business, then you can get more information here: It All Begins with Insights.