EP36 AI - Getting real ROI in your business.

In this podcast episode, Mike and Gaurav Devshamar discuss the practical applications of AI in business, addressing common misconceptions, data privacy concerns, and the accessibility of AI for small and medium-sized businesses (SMBs). They explore how AI empowers generalists, automates tasks, and aids in knowledge retention, ultimately transforming business operations. The conversation emphasizes the importance of adapting to rapid changes in technology and encourages business leaders to start integrating AI into their strategies.

takeaways

• AI is a hot topic and is rapidly evolving.
• Many businesses fear data privacy when considering AI.
• AI solutions are accessible for small and medium-sized businesses.
• AI can empower generalists to perform better in their roles.
• Automation of tedious tasks can save time and resources.
• Knowledge retention is crucial for business continuity.
• AI can help reduce key person dependency in organizations.
• Onboarding processes can be streamlined with AI tools.
• Businesses need to prepare for the rapid changes AI brings.
• Starting with AI can provide a competitive advantage.

titles

• Unlocking AI for Business Success
• Demystifying AI: Practical Insights for SMBs
• AI Misconceptions: What Every Business Owner Should Know
• Empowering Your Team with AI Solutions

Chapters

00:00
Introduction to AI in Business

03:13
Misconceptions About AI

06:10
Addressing Data Privacy Concerns

09:09
AI Accessibility for SMBs

12:14
Empowering Generalists with AI

14:57
Practical AI Applications for Businesses

18:02
Knowledge Retention and AI

20:52
Transforming Key Person Dependency

24:09
AI in Onboarding and Training

26:59
The Future of AI in Business

29:47
Conclusion and Call to Action

Get in touch with Gaurav on gaurav@warpdevelopment.com 

Follow him on: https://www.linkedin.com/in/gauravdevsarmah-763299169/

Find out more about working with me. mike@smbmastery.com.au or https://www.linkedin.com/in/mikeadamscott/

Transcript

Mike Scott (00:01.401)
Hello guys and girls, good to be back. I've been super slack this year. This is the first podcast recording that I've done this year, which is super lame. But I'm going to get back on the wagon now because I miss this and people are asking me for it. So here we go. It shouldn't be a surprise that the focus of today's session is going to be AI, right? But I really want to make this super, super practical. AI is everywhere right now. I think there's potentially fatigue about hearing about it. But for most business owners that I speak to, which is a lot,

It's really hard for a lot of them to know where to start. Today, we're to try and cut through that noise, and we're going to focus on practical ROI driven ways that we can bring AI into our businesses, whether you're big or whether you're small. We want to try and have practical takeaways. Today, my guest is Gaurav Devshamar, who is the head of AI strategy and solutions for the Warp Group. Warp is a 120 person business that has got a very long track record in software development and capability.

and is now turning a lot of its direction into serving customers through AI consultation and implementation together with everything that they're doing. Gaurav is an absolute expert when it comes to AI, which is a hard thing to say in this field because everyone's kind of claiming to be an expert, but this guy's actually done a lot of stuff. What I like about Gaurav particularly low though is that he's got an MBA background and he's a highly technical dude. But what's interesting about pretty much all of the conversations I've had with him is

Yes, you can nerd out and get into the weeds on the tech stuff, but it's always got this what's in it for the business angle. Like what's the business outcome, which is really what we want to be talking about today. I don't want to get into deep dive technical discussions about AI. Full disclosure, I am invested in Warp. It's a business that I'm involved with that's public. But regardless of that, it's still very relevant for this podcast. I wouldn't have gotten involved with them if I didn't think that they were an awesome outfit. So welcome, Gaurav. Good to have you on the show.

Gaurav (01:57.42)
Thanks, Mike. Thanks for the really warm intro. And super excited to actually kick this conversation off. As you said, AI is the hot topic at the moment. And I think it's also one of those topics where you just never run out of things to actually talk about. I guess the, think one thing I will say is it's, unless you're like a top researcher working at Entropic OpenAI.

it's pretty much very hard to be an expert in this space. think the main thing is curiosity and just that willingness to learn all the time because that's what really separates people in this space. It's moving so fast and I think it's that hunger to learn and I think that's probably what I would say is, I guess that leads to the expertise.

Mike Scott (02:40.665)
100%.

Yeah, a hundred percent. mean, just to share with the audience to really put this into perspective, I was having breakfast with Gaurav and Minghao this morning, AI lead developer, and Gaurav was talking about a new model that's come out. I'm like, when did it come out? And he was like, about 50 minutes ago. And like, you know, think that's the speed that we're working with now. Okay. So I want to get straight into the practical kind of conversation here. So I think a question that I want to start with is what are some of the biggest misconceptions that you are hearing from or seeing or observing?

from business owners when they're starting to think about AI.

Gaurav (03:17.952)
Absolutely. And it's a great question. I think there's two at the top of my head that really stand out. One, I think there's massive concerns about data privacy and security. it's like businesses want to play around with AI solutions and actually see how AI matters to them. But they are actually terrified that they're going to kind of...

send out their data to some third party models and what happens if there's sensitive data that gets leaked or hacked. But I think that's where this is definitely a misconception because there's ways you can use cloud based solutions where that clearing fence everything and then everything is protected. So that is probably one that comes up quite often. And the second one is that AI solutions are very expensive and this definitely

relates more to your SMB market because there is this kind of misconception that only the big firms, the larger firms with 1000 plus employees can actually afford AI solutions right now. And that's just not true because AI is like any other solution that you'd use to optimize or like do something in your business. So it really depends on what you're trying to work on and what your

I guess the scope and budget and all those things are and you can actually customize the solutions based on that. So I think that is just it's not true that it's expensive. Yeah.

Mike Scott (04:48.217)
Okay. Yeah. So let me jump to the first one, right? Because I certainly think about that. I hear pretty much every business owner talking about that is like, I don't want my team to put proprietary information into chat GPT because then it's public and there's proprietary information. Going even further, if we think about certain regulated clients like financial planning or any kind of regulated entity, that actually becomes a legal requirement. So without getting too much into the technicality,

just from a non-technical perspective, that observation that you had that most business owners are worried about the privacy, maybe in like a minute or two, no more, without getting technical, how do you put their fears aside? So if you're saying, listen, I know that you're worried about privacy, how do you address that? How do you practically address that with a business owner?

Gaurav (05:42.604)
Yes. I think most businesses are already using some sort of cloud solutions, be it let's say, know, Microsoft Azure or AWS or Google cloud, whatever it is. So the best way I actually speak to them is you're all utilizing this and there's this AI capabilities inside this cloud solutions where you can utilize. And it's just no different than what you are doing with how you're using it at the moment. Like your whole databases could be set up.

in the cloud solutions. And I think all you're doing is you're just implementing another feature they have, is building out AI solutions that you're hosting it in these cloud solutions. And that just means that everything is ring-fenced. Nothing really leaves your instance. And it's no different than what you're doing right now in terms of using those kind cloud-based providers. So it's an easy conversation to have because you're not kind of, I guess,

getting them to use something that's not already there. It's, the only using this most of them. So I think that's probably the, the way I approached his discussions and a lot of them are just not aware of it, which is fine because this space is like moving at a crazy space. And I think, yeah.

Mike Scott (06:53.739)
Yeah. But that's your role, Like that's part of your role is to say, Hey, let me make you aware of what's possible. Let me guide you through this process because you don't need to be expert in this. I want to just make sure I understand this. So basically every business in the world is going to either be on the Microsoft stack using 365, SharePoint, et cetera, or they're going to be on the Google stack using Google apps, Google. Like, I think that's a fair.

It's fair statement. Maybe there's some crazy, outlying, open source Linux-based stuff still. I don't know. But we can make an assumption that the vast majority of business will either be a Microsoft house or a Gmail Google Apps house. Am I understanding you correctly that therefore, if you are on one of those platforms, you've got an instance in which you can build your AI tooling. You're either going to build it within the Microsoft environment, within that hosted cloud environment, or you're going to build it within the Google environment, both of which are

Gaurav (07:19.426)
Mm.

Mike Scott (07:47.051)
secure and ring-fenced and not going to be publicly or open-sourcely, if that's a word, accessed. Is that a fair summary or have I oversimplified it too much?

Gaurav (07:59.328)
No, kind of in terms of, know, your, guess Google and Microsoft are obviously the two biggest providers of B2B enterprise solutions. But I think when you look at the cloud-based providers, we are looking at mainly Microsoft and Amazon because GCP, which is a Google cloud, that comes third and it's, lesser used than the two others. normally I think, so that's the way I think, but it doesn't really matter.

because all of them are exactly the same.

Mike Scott (08:29.625)
Okay, so we're not talking too much within the software of Microsoft House or Google House. We're talking more about are you hosted on an AWS environment or in a Xero environment? Am I hearing that correctly? Okay.

Gaurav (08:41.422)
Exactly. that's, I think that's where we're looking at building these solutions on top of which can bring fence everything.

Mike Scott (08:47.555)
Got it. Got it. Okay. So that's sort of like the first question in terms of AI readiness. Then if I understand it is if your business is either hosted within an AWS environment, Amazon, or an Azure environment, Microsoft, we can build within that framework. That's kind of like then the first step almost of like being AI ready to even start talking about this. Is that a fair comment?

Gaurav (09:09.556)
Absolutely, spot on. the other thing you have, I think a lot of businesses, as you said, have compliance requirements where the data can't live a certain country or things like that. And all of that can be done in those platforms. it's pretty much, it's exactly like how you'd actually use it for storage or any other kind of functions like for those platforms. And AI is just no different.

Mike Scott (09:31.757)
Yeah, I love it. The other thing that you spoke about was that this fundamental misconception that AI is really expensive and it's only for really big organizations. think before I ask you to unpack that, I just want to reflect on an amazing podcast that I listened to recently. Me and you have already spoken about it, but if you haven't checked it out, guys and girls, just Google on your podcast platform, AI or Die. And it's with one of the partners at Sequoia Capital. And there were so many nuggets in that, but one of the things I took away from it, which I found quite cute,

but quite useful was he said like, the definition of SMBs is changing. It's no longer small and mid-sized businesses. Now AI has created this new definition, which is small and mighty businesses. And he goes on to explain this by, the larger your organization is now, the harder it is to move. So if you have thousands of people in head count, it's really, really slow and difficult to move. So actually, where we're going to see

the most gains, the most growth, the most adoption is actually in the smaller, more nimble businesses. So now you've got these five, 10 person businesses in the Valley taking on pretty significant incumbents, right? Which is really like, that's the extreme version of what I Gaurav saying a little minute ago, but let's get into that. So you're a business of, don't know, I don't know what metric to use. I don't know if it's revenue or head count, but let's take head count for now, cause that's quite relevant to AI. You're a business of like 10 to 20 to 30 to 40 to 50 people.

maybe 100 people, but not a multi-thousand person organization. Talk to me a little bit more about what are some easy, accessible AI solutions that you are thinking about, that you are seeing, that you have built, that you have been part of. What is some really practical stuff that you're seeing across different industries in these smaller businesses, not talking about these large corporates?

Gaurav (11:18.542)
Sure. Great question. Very relevant. think one of the ways I think SMBs fit in really well is because AI as a technology allows more people to be generalists in the sense that you can do... Like I have... I'm not a designer, for example. I have no designing.

skills in terms of formal skills, but now I can actually design really good systems and platforms because there's tools out there that allow me to do that. And I can be a more of a generalist. And if you think about how do SMBs work in general for getting AI in SMBs, a lot of people, or majority of people will wear multiple hats compared to a large corporate where it's more specialized. So I think

This is where I think SMB should actually realize that how much power there is in this technology, because it's fitting into what you're already doing. It's making those generalists who are juggling multiple hats way more powerful, because now there's a technology that just allows you to be, it takes you probably more, I don't know, it makes you better, like 50 % better at this generalist kind of task that you're doing. Maybe someone's doing a bit of sales, someone's doing a bit of operations. And I think that's like the power.

There is.

Mike Scott (12:45.987)
So it's almost creating specialist generalists, which is a total oxymoron. But the cool way to think about it is, what are you a specialist at? Well, the thing that I'm a specialist at is I can do a lot of stuff pretty well. Not at the best level, but a lot of stuff pretty well, which is a very useful thing in an SMB, 100%. Unless you're a multi-thousand person organization, you kind of have to be able to do a lot of stuff. We used to call it slashing.

Gaurav (12:49.782)
Yes.

Mike Scott (13:10.541)
You know, I'm the CEO slash the coffee maker slash the marketing person slash the customer service person. Like the slashing roles is very, very relevant in, every business, right? Like I have a deep insight into this because I sit with leadership teams. Zero of them are sitting with people that just do one thing. There's always overlap in, but I also do this and I also do that. that's a great segue into, into my next question that I actually wrote down here, which is let's get specific Nagora of like.

What are some tasks that AI can handle right now that businesses might not be aware of? Automating emails, summarizing meetings, customer support chatbots. Some of these things are quite well known, but can you list off the top of your head maybe five to 10 tasks that AI can handle today that people might not realize?

Gaurav (14:01.294)
Sure. I think I'll go more into what, when I speak to, I guess, people from different industries, what kind of stands out, some of the trends. And one of the biggest ones is that everyone's got, like every business has got a ton of depth in their knowledge. It's like the business IP that they've built over decades or like many decades. And I think the big thing that I've found is how can we get our less experienced employees?

to actually get more info out of, let's say for a manufacturing business, it'll be how can our junior technicians fix a problem or like not even fix a problem, but at least have an understanding of a problem by acquiring all the knowledge that we have accumulated over the many years so that we don't rely as much on someone who's more senior, but then there's limited talent in that senior capacity, which is always the case. Obviously you'll have less people who have...

two decades of experience versus someone who joined the company a couple of years ago. that's, and, and, for example, in a chat bot type interface, if I query something about the business, about, you know, how do I fix this machine? that's doing X, Y, Z, instead of going to a senior, manager who's done it, I can just query the entirety of the organization's knowledge base, and then it just gives me a checklist. So that means.

At least worst case scenario, I'll be 80 % there as a junior technician to understand the problem. I think that's where this massive, I guess, ROI and efficiency gains where you're raising the floor in the business. I, traditionally...

Mike Scott (15:45.889)
Yeah, like juniors, like there will be no such thing as a junior. Like a junior now will have a knowledge base of what a mid-level would have before because the knowledge base is no longer in that person's brain. It's now accessible. Is that what I'm hearing?

Gaurav (16:02.294)
Yes, how and because the thing is every organization has got SOPs and tons of documents, but the honest truth is we don't look at half of the documents. It's just, it's just sitting there. And I think what we have right now is we can search for the right info at the right time. It's a bit like having a 24 seven senior mentor or like senior leader in your pocket all the time. And I think that changes the way

you can actually do business. And this applies for like manufacturing, this applies for retail, hospitality, pretty much honestly, anywhere really, because this is a quite general purpose way to query info.

Mike Scott (16:43.671)
Yeah. I want to jump into that. let's continue with that. Okay. So that's, that's great in theory and I'm going to play devil's advocate, right? So that's great in theory. And I'm sitting here going, my God, you're speaking to me. I've got someone in my business. Maybe it's me, maybe it's my business partner. Maybe it's my first employee, but I've got someone in my business that has got 20 years of business IP and knowledge in his or her brain. And they're terrible at getting it out. They're terrible at sharing it with other people.

but they're exceptionally valuable to the business. I've created this enormous bottleneck. It might be a, if you're like an engineering business, it'll probably be like your founding engineering partner. If you are a, I don't know, sales-based business, it's like the founder that is best at sales but can't train other people up well. So I think everyone gets the concept of a key person dependency because of a centralized or concentrated amount of knowledge in one person's head. But practically, like practically,

How do we do this with AI? Because the problem of it all being in that person's head is still a problem. AI can't solve getting it out of that person's head. We've got to get it out of that person's head into a data source so that AI can query it. Talk to me about practical ways of attacking that problem.

Gaurav (18:02.222)
Sure. And this is a great question, right? Because this is something I've been having a lot of conversations on, is that AI and how good the AI solutions are, depend on your data. That's basically it. If your data is not good, or if you don't have any data, then of course, it'll be hard to build any, I guess, solution for your business that will actually return high ROI.

You can use general purpose tools like, know, chat, GPT or cloud, which are just built. then you can do what everyone else does. But in terms of things that are relevant to your business requirements, you need the data. And if it's just, let's say in someone's head, and that's what I ask my clients to do is we need to somehow actually get that codified and in a way. I think there is a very important

change out here where earlier everyone created documents and SOPs and stuff for humans. Whereas now you have to think about creating this codified knowledge, not for humans, but for your AI, like LLMs. Because then once you have that, then the LLMs can actually do all sorts of, and this capability just keep getting better and better and better. So once you have that codified, which you have to, because if there's no data, there's no solution for that, 100%.

And the other thing you have to understand is...

And this is a mistake I've seen businesses make, is that just because something's been done in a way for the last 20 years, you don't want AI to replicate that exactly as it's been done. And there's the power of AI, because if you codify that workflow or the process or whatever it is, you can actually get LLMs to suggest improvements to how you can do that better.

Gaurav (19:59.63)
and what the new workflows could look like. So you're not just kind of automating what's been done, but you're actually putting those like new intelligence that you previously was just not possible or just, you know, at least not possible for SMBs. So yeah, I think that, okay, not to go on a tangent, but I think the important part here is that, we will need to codify that from someone's head to actually kind of see, okay, this is roughly the workflow. These are the outputs and how we can, yeah.

Mike Scott (20:24.695)
Yeah. But it's, but, just to jump into that, right. It's, like the AI itself has got tools to make that easier. Right. So in the old, in the old days, or not even the old days anymore, like a year ago, that key person dependency and that concentration of IP was in that person's head. And the only way we had of getting it out was through writing very long form documents, like standard operating procedures or what have you. But now we can sit with that key person and we can interview them for two hours a week.

about tell me how you do this, tell me how you do that, tell me how you do the next thing. And we can transcribe that stuff, building this very rich corpus of information for the AI to query. We can take the last 10 years of proposals that have been written. We can take the last 10 years of projects that have been delivered on and do a post-mortem on them and say, went well, what went poorly from this person. And just through conversation, we can now build this very rich data set, which is a super.

practical and easy thing to do. Most of these key people in businesses, if you're listening to this and you're one of them, you enjoy being interviewed on the thing that you know so much about. This is a pleasant process. It's not a chore. Like if you're an expert on, I don't know, a product that your business runs, probably you really enjoy being interviewed on that product. So just talk about it. Use a transcription software, build that corpus of information, and that companies like what Goraver's running sort of

do the work on top of that for you. Would that be a fair, practical approach to this stuff?

Gaurav (21:53.326)
Absolutely. And I think that's actually a really good point that you can use these tools to really quicken the process in terms of rather than sitting and writing a document for months and months to codify everything, you can actually just use these transcription services and then get some intelligence on top of that. And yeah, and I think at times, honestly, I've been surprised by how good these things really work when you hook into the LLMs that you have.

Mike Scott (22:19.245)
Yeah. I mean, just as a use case, guys and girls, mean, in my coaching work, I transcribe all that work and I now I'm in the habit of continually building this database of information to basically query on how to be better every time. So every time I'm having a coaching engagement, every time I'm having facilitation, I've now got this massive context that I can query to say, what should I do better? What should I do less of? What are my blind spots? What am I missing?

to the point that Guarivay, she sent me a prompt, which was quite intense recently, which was like, Hey, give me deep insights about my psyche things that I might not want to hear. And I showed it to my wife and she was like, Oh my God, Mike, this is like, this is so bang on. And the insights it's created was, was insane. Right. But it's because I'd given it a rich, rich data set. I want to come back. I want to ask the same question again though, because what you did there was very useful.

Gaurav (22:55.854)
Yeah.

you

you

Mike Scott (23:13.463)
You went straight into like, it's about the data quality that you have. But I want to ask the same question again, in a very simple, practical way, what are like five to 10 things that AI can do today for your business in terms of automating tasks and taking tasks off people's desks, like practical things that people might not realize? Maybe five to 10 things. Can you list like those tasks that are capable today?

Gaurav (23:40.11)
Sure. think number one, a lot of businesses rely on a ton of research. So you're doing a lot of research based on for a manufacturing firm, might be about machines and parts and things like that. For a consulting firm, would be just your normal service-based consulting, market research and things like that. And all of that can be automated. I think I will mention one thing is that processes like that

The automation should do, let's say, to 80 percent of the work, and then 20 percent of it should be a human review where you're looking at it and judging how good the output is. So that's like one small use case. Second one, a lot of people will just do all sorts of invoicing and billing and just tedious tasks that take up a long time. And these are tasks that are very repeatable and low value.

And I think, so which is why these are low hanging fruits for businesses where you can generate immediate returns and you can save a ton of time without actually, I guess, creating something that's overly complex as an AI solution. So that's one. And funny enough, I think from SMBs, I've also seen, this is something I did not even think was an issue, but one thing I've seen is just inbox management.

because a lot of SMBs will have one email that's shared between different people. For example, sales at xyz.com and everything just comes in there. And then there'd be a manager categorizing it into different things, manually, into different folders and stuff. And then someone will just have a look into it and then, yeah, be it scoping it out or whatever, like coding and things like that. So this is something I did not even think about, because I thought this was just not an issue.

But now I've actually seen that and that can be automated quite easily again with your, you basically it's not just even today's like LLM best AI, but it's also like basic automation that is all the, that's been there for the last, I guess, 20 years. So you hook that into like an LLM and then you get some really powerful solutions. And I think the other one is as we discussed earlier, just querying your knowledge, your business knowledge through like a

Gaurav (26:03.606)
like an internal, almost like a simple chatbot interface and then generating like checklists or whatever. That is, that's super powerful. And I think the main reason is that super powerful is because we discussed about what's it called? Just raising the floor because it's, the capabilities just change in terms of what you can do with a limited workforce. And if we get into, I guess,

bit more complex solution and I think that's knowledge retention in terms of knowledge retention is probably one of the biggest pain points from a 10 % business to like a 10,000 people person business. So it's the same across like pretty much every business and what's interesting is the model changes a bit in terms of how you look at how many people are in a business because

Earlier, if someone left, they left. So it's a plus one, minus one. so if it was 50 % business, remains a 50 % business. If one person leaves, one person jumps in. But now, because you can retain someone's knowledge, and this might be controversial, to be fair, that person kind of, the person might leave, but the knowledge still stays within your business. So I think that that's another use case where you can build different workflows based on what someone's already done.

And you can retain that in the knowledge. So when someone else comes in, it's not a straight plus 1 minus 1. It's just a plus 1 almost. Because that, yeah.

Mike Scott (27:38.169)
You know, there's a very famous sort of line that I love, which is just so interesting now. So it's like a parable, but it's like the CFO says to the CEO, but what if we spent all this money and time and effort on resources on training people and they just leave? And the CEO says back to him, well, what if we don't and they stay? Right. And that's always been a very strong message, right? But this is now changing the game because this is now putting a whole different spin on this

on this challenge that businesses always face around investing very deeply in their people, in growth, in knowledge base, and then effectively having those people move on and using that business as a stepping stone in their career. Now, I'm of the opinion, by the way, that that's fine because I actually think at a philosophical level and a market level, if you build a reputation as a business for growing people, sure, you lose people, but you also attract a lot of people because on the inbound, people know that, going to work for Guarav's company

I know I'm going to grow really fast. And you actually, it's a long-term thinking, but what I'm hearing from you is now you can kind of have your cake and eat it. Now you can grow people very fast for your organization and you can actually wish them well in the world and send them off onto their next career and retain information and knowledge that they've done. can literally have both. So, you know, this, this is beginning to be a bit of a have your cake and eat it scenario, which is, which is

That's a pretty nuanced thing. I never thought of that. That's a really good insight. And it doesn't have to be, like you said, it's controversial. I think it's controversial if you're treating people as fodder to feed your AI. That's not great. But what I like about this is a different lens is, we can actively now anticipate shorter tenure cycles from people in our business, but it's no longer a problem. It's actually factored into our model, which previously had been a problem.

We know that people are staying in jobs for shorter amounts of time. We know that people are not even like lots of fractional type roles. That's what's happening in the labor market. This presents an opportunity to not have that as a problem, which is actually pretty significant. I hadn't thought of that before.

Gaurav (29:47.234)
Yeah. And I think it changes some of the, I guess, dynamics. And also I think it opens up just new opportunities in a lot of different ways. It be onboarding. And that's another use case, by the way, like onboarding new employees. And that's another, like a massive pain point. Whereas now you can actually generate this, I guess, checklist and what you're doing, what new employee kind of, what documents they have to look at.

and what info they want to kind of get in their first week or also have like a kind of simple chatbot that they can kind of speak to us. It's almost like their buddy who speeds them up to what a company does. Because that takes, I think, in my experience, it takes a month, almost to be fair, in some cases, two months to actually get up to speed. Because you don't have a buddy 24-7 with you.

who can actually help you.

Mike Scott (30:47.481)
And it's more than that as well, right? There's a cost. And if you're listening to this, you'll either experience this and you know exactly what I'm talking about, or you wouldn't, and it sounds theoretical. in my previous business, Nona, we generally hired very senior, very experienced engineers. And we actually started the whole typical thing, which is to try and get interns because it's very cheap labor or free labor and sort of upscale them. And we stopped doing it. We stopped doing it completely, right? Firstly, because we thought the model was a bit unethical.

To be honest, that wasn't the main reason. The main reason was the whole model there is take the most senior people in your business and get them to mentor the most junior people in your business. What actually played out though was basically the senior people who are the most valuable in the business in terms of their time were spending most of their time on very junior people. And that just doesn't work out from a commercial perspective. It doesn't work. It doesn't work from a commercial perspective. It also doesn't work from a...

those senior people want to be engaged in difficult, challenging problems, not always just mentoring very junior people. So this is really interesting from that perspective as well, going beyond onboarding. It's kind of like, well, now we can codify a lot of that senior knowledge and have it accessible in the central and growing database. So there's kind of like, I think you used this term earlier, there's almost like a senior or a mentor on tap within a secure framework that is not, there's no privacy issues and is not proprietary data. And it's quite interesting because now I'm just

my mind's sort of sparking up and jumping all over the place. from a valuation perspective, think we might start seeing some really interesting stuff in due diligence. I think there might be a line item that pops up in due diligence on companies from now on is like, what's the quality or value of your data set from an AI perspective? Like we're buying your book, we're buying your typical IP, we're buying your talent, but we're also buying the degree to which...

your AI database, let me just call it that for lack of a better term, we're buying the degree to which your AI database can reduce the key person dependency. That's an interesting concept, right? Is that I think &A advisors are going to start looking at a significant line item is what's the quality of the data that your AIs are querying, which is a whole different thing that hasn't existed before.

Gaurav (32:59.982)
I'm actually so glad you brought this up because this was a conversation I was actually having with my dad the other day where it was about, I think &A is probably the best use case for this, but I think I was having more of a general conversation in terms of how codified data, how valuable that's going to get. It's all devaluable, but how valuable it's going to get more and more because...

The models are getting incredibly powerful right now. But what these models lack is some of this niche vertical data to actually build solutions. For example, let's say if I want to a software that is very CNC manufacturing focused to get something that's the data set, which understands CNC machines, processes, workflows.

it's going to be incredibly difficult or like there's no market kind of for it. But now suddenly businesses that are actually codifying all these things, they could have a new market where they're kind of obviously masking their data so that nothing is being sold, which is private, but at least for the workflows and things like that, where software companies might actually pay them a license fee.

to get access to those workflows. And I think that's a bit different than just mergers and acquisitions. And I think that's something I feel will happen a lot more. And this doesn't just, I guess, go towards businesses, but I think it'll also be for individuals. Because I think that's something that'll... I think we've already seen some of these examples, right? Because initially, for example, OpenAI trained entire to the internet because...

No one really understood what was happening, so they kind of got a free ride, but now they've got a few lawsuits and stuff. I don't think anything will happen on that aspect, but now they actually have deals with Reddit and a few others where they have these deals to actually train their models on the data. So I think that's going to just go beyond, and that's general purpose, right? But I think the vertical use case is the one that's going to be very interesting.

Mike Scott (35:21.465)
It's super interesting. Yeah. Gavin, I won't give your surname out, but Gav, if you're listening to this, you know, from a DDE perspective and the &A space, hopefully your ears are pricking up. That's a bit of an inside joke, but he'll know it. All right, so we've covered a lot today. I want you to shift gears a little bit as we start to wrap up. For business owners or business decision makers or influencers who are listening to this conversation and thinking, this sounds great, but I still need help figuring it out.

What is the best way for them to follow or contact you?

Gaurav (35:51.246)
Thank

Gaurav (35:55.576)
Sure, I can add my LinkedIn, that's one. Obviously you can add it into the podcast. And the other one is you can just reach out to me at gaurav at workdevelopment.com. So that's my contact in terms of my email. And yeah, I'm always up for a chat. I love noting out anything that's AI in business. I love learning about, for me, I think it's all about the business problem. And I know being...

limited in what business like industries I've worked in and stuff. There's so many problems that I can actually work on that SME is no way better than I would, but I then bring in the AI part of things. So I'm always happy to kind of, yeah, if you're in AdLib, grab a coffee or anywhere really, I could be in another state or just reach out to me on my email and LinkedIn and we can definitely carry the chats forward, be, you know, things like how can you get started in AI? Because I know it is an overwhelming space at the moment. And I think what

I really try to push is what this AI mean for your business. think that's the main part. Like where does it fit in for you? It's not about honestly, forget about OpenAI's new models, Anthropic's new models, because yes, the capabilities will matter, but initially those things don't matter. Initially what matters is where are your business bottlenecks? I think there's two questions that I asked. Like where are you bleeding the most in terms of a business leader at the moment? So that's one. And second,

If there's a business who had 50 employees, I get them to think, what could you do if you had 100 employees? Because, and I think that gets them thinking, okay, we can actually do this. We can offer this offering. We can extend our offerings into XYZ. And that's where we, I guess, find the opportunities on what AI could do, because that's like the second step where.

you're not just automating processes, but you're actually adding new value, which is previously just not possible.

Mike Scott (37:54.263)
Yeah, I love it. I think you already answered the question that I wanted to ask you, but just in case you haven't, what is the single core message you want to leave with our audience from today's conversation?

Gaurav (38:06.776)
think one thing is change is happening at a rapid pace. I think it's why I will really suggest that to be prepared for that. And I think take it as a positive rather than like as something that, we're scared about it because it's actually going to be really, really helpful for businesses to adopt this new kind of curve and the new world we are kind of moving into. So be curious, I think learn.

I guess that learning mindset will definitely help and start thinking about things in a way where you, and I think we mentioned some of the examples, it's at how you're preparing a lot of your business knowledge and intelligence for AI rather than just humans. It'd be like documentation. I think that's businesses. Yeah. And businesses who are preparing for that, we'd have such a substantial advantage in two years time compared to businesses who are not.

Mike Scott (38:53.517)
Yeah, I like that fundamental shift.

Gaurav (39:04.482)
And I think that gap is going to be quite big. And I do think some businesses will really, really struggle who haven't adopted because suddenly the gap between two similar competitors will just be massive in two years time. Someone who's adopted AI versus someone who hasn't. And this is where you take baby steps. And I think it's all about thinking about that. Yeah. Just start. Yeah.

Mike Scott (39:26.585)
Yeah. Yeah. Just start, just begin. Right. That's it. I want to ask you one last funny question, but it's not so funny. You guys can't really see Gwarav's physique, but he's totally jacked and he's a machine in the gym. So, various question. How are you using AI to help with your gains in the gym Gwarav?

Gaurav (39:39.79)
you

Gaurav (39:48.558)
Thanks for that first of all. I love the gym. think the gym definitely helped me in other aspects just being disciplined. And initially, it's funny because gym is everything. How I work at the gym is everything that I do in my real life, which is just do the thing rather than know it all or something like that. And what I did was I actually asked Chad GPT, this was ages ago, and he generated me a routine and things like that. And I just followed.

That for it for a bit. I still I've been going to the gym for a while now and I still don't know half of the exercises I actually do I know how to do them. So it's and I think and one of the other things I do is and We we built an app which is a reset AI it's just a fun app and I think you can track your progress and stuff So I recorded all of the things out there and then we have different personalities of coaches in that app, which means that

Sometimes if I need a push I'll leave the coach which is really really harsh and then and the coach just kind of tells me You're slacking off this and that this is absolutely rubbish so I think that kind of adds in extra motivation and you know what's funny is and and see I I use I think this is This is just incredible because I used it in a private manner but in a secure environment, but All these decks are scans that you can do on your body in terms of fat

presentation, things like that. I used this some open source vision models and actually counted mine. First of all, that was like three years ago that I actually did a DEXA scan versus now. And I was actually shocked how good it was at predicting exactly what the DEXA scan showed me three years ago versus obviously I haven't done a DEXA scan now. again, so it showed like how interesting this technology is.

Mike Scott (41:19.939)
Mm-hmm. Yep.

Mike Scott (41:35.469)
wow.

Yeah, that's crazy. Yeah, yeah.

Mike Scott (41:45.377)
Yeah, it's crazy. mean, I was having a bit of fun, but that actually that's turned out to be a really strong metaphor, right? It's just like, get up, do the thing, be consistent. The results will follow. mean, that's like, that's business, right? That is business right there. So, and also like what you said that you don't necessarily know the names of everything or whatever, but you can do the thing. And I think that's a great metaphor actually for what we're talking about here. Just get started. Right. You don't have to be an expert. That's why companies like warp exists with Guaro of leading that AI function.

get someone like Guarev to help you with this stuff, but just get started. Guarev, it's been a super cool conversation. We're out of time. Thank you so much for sharing generously. I'll drop the contact notes for anybody who wants to talk to Guarev or start consulting with him in the business. And yeah, thanks for your time. Have an awesome day.

Gaurav (42:33.752)
Thanks Mike, this was really good. Have a good day.

Mike Scott (42:36.665)
Cool. And to the listeners, you guys know the drill. Don't wait to just think about the stuff. If you heard something that you like, go and share it with somebody. Don't wait. This is very important stuff. You've got to get on top of it, whether it's with a consultant, whether it's by yourself. You've got to start bringing your business and yourselves into this space. Have an awesome day. We'll be back soon.