The Progress in Practice Podcast: Episode 7 Chetan Joshi V2

Welcome to Progress in Practice, where we explore what it really takes to build and sustain a professional service business. The hard lessons, the smart moves, the uncomfortable truths, because success leaves clues if you're willing to look for them. I'm Charlie Reading. Let's dive in. So Cheetan, welcome to the Progress in In Practice podcast. I am really looking forward to chatting to you today because I know that we're going to end up spending a lot of time talking about the world of AI, which for anyone that listens to this or follows what I do, I'm incredibly passionate about. So really looking forward to talking to you today. So to kick things off, the first question I always want to ask is like, what do you do? Who do you serve? And what makes you different? So how do you help people? And who do you help and why?

Sure. Thank you for that, Charlie. Before I answer that question, yeah, thanks for the invite. This is something I really like to do is educate and learn key questions that the audience, your audience have. So thank you for the invite. So, you know, who do I, who do I serve? Who do I help? I predominantly work with the financial services and banking industry and helping them solve complex problems using AI. And the AI that I focus on is called vertical AI. And I'll go into that a bit more around what vertical AI is. But I also help entrepreneurs and business owners. So everyone wants to do something in AI, but they're struggling to figure out where do they start and how do they do it. So I help them move from kind of experimentation to execution and then helping them build what I call vertical AI agents. And so it's designed for their industry, their workflow, and it's just one outcome which allows them to get the accuracy that they desire.

So just for those people who aren't absolutely clear on what you mean by vertical AI, explain that terminology as if I was 5.

Yeah, sure. So let's take two, two babies born at the same time. Now the fir— the first baby, we train the baby mathematics, English, science, human emotions, and it grows to become a mini Jack of all trades. It's a very, very artistic, and it's, and it's always thinking of generating an answer. It's been trained that way. Now the second baby, we only train it mathematics, mathematics only, pure maths. Now, That baby becomes very scientific. It understands right from wrong, and it always is subservient to the human. So if it doesn't know the answer, always check with, with the human. Now, the first baby is generative AI, and the second baby is vertical AI. Now, both of them aren't competitors. They can be used in different use cases. So when we look at generative AI, which hit our living rooms in end of 2022 and early '23 with ChatGPT, that was an amazing engineering feat which allowed us to understand how AI can be used in our day-to-day lives around summarization, around brainstorming, MVPs. But then quietly in the background, banks have been developing vertical AI solutions that are really targeted into their industry, that understands their domain, their context and convention. So it's a— it's— there's two types of intelligence. Everyone always talks about GenAI, which is horizontal AI, and I talk about vertical AI, which is narrow AI.

So it's, it's like having the specialist, I suppose, is it? It's like saying this particular AI is built to do this task really well. It's not gonna do all the other stuff. You're not gonna have a conversation with it and brainstorm your, your best man speech with it, but it's gonna do this job and it's gonna automate this process. Is that, is that what you mean?

Yeah, exactly. And the way it works is that you feed it the data. So, so it's only your data that goes into that model. So you are training this AI model the way you want to be trained. So it's like a sponge. It soaks up all your input, all your in-house rules, all your in-house data. And then there's the guardrails to say that if it doesn't know the answer, it always comes back to you to say, is this right? And then you have that feedback loop with the, with the AI, and it learns. So the next time it doesn't ask you that question. So you have more control around your, your data. And what that does, it increases the, the accuracy level to 100%. Whereas if you have an AI that just thinks on its own, you don't know where the data is coming in. There could be some consequences around, you know, data going out incorrectly or payments going the door. So it's control mechanism around that.

Brilliant. Brilliant. Love that. And so what story best explains why you do what you do?

Yeah, sure. So I started my career in banking in '97, and I spent lots of time standing in front of this big white box called a fax machine. Some of your readers might not know what a fax machine is, but I was qualified in university, had a degree in corporate finance, but I was spending hours and days just spending time processing data. And I, at that, when I reflect back on my journey, my story, I was a human robot in these big corporations, you know, very smart in what I was doing, but I was doing a job that a machine can do. But at that time, the machines weren't able to do those tasks for us. So it, you know, there was no real judgment involved in those processes of clicking buttons, reconciling the information, and then pushing it down to the next part of the workflow. So that's where that kind of inefficiency came and that risk came around that. So, you know, we were acting like robots. And now what I realize is now with AI, we can replace all those humans doing those kind of repetitive mundane tasks with AI. And then we shift them now to more high value add, more kind of using their judgment and their expertise, which is a win-win for everyone.

Brilliant. I couldn't agree with you more. I think the term I found myself saying, and I'm not even sure whether I heard it somewhere else or whether I made it up. Is to automate the repetitive, humanize the exceptional. And I think it's just, you know, AI is not about replacing humans. It's about uplifting them, isn't it? It's about freeing them up of the repetitive, boring stuff so that they can do the stuff that involves empathy and creativity and leadership and all of that kind of piece. So I think that's absolutely brilliant. And I think there's so many jobs within a business, you know, particularly working in financial services that could be automated and done far better by by a robot as you describe. So if you could go back and send your younger self, you know, maybe that younger self that stood in front of the fax machine, if you could go back and sort of give them, you know, send them a memo, give them some advice, knowing what you know now, what would you say and why?

You know, it's a great question. So for me, it would be about stop trying to be indispensable. So when you're starting out in your career, you, you want to, especially in the banking, financial services, you want to put the hours in, you want to be first in, last out. And you're not thinking in a way that you, you should be building systems that can replace what you do. So if you're not there, the systems will continue to work. So in that way, that dependency of that human goes away. So I spent a lot of time sacrificing myself and working with these systems with friends and family, when actually if I started thinking about systems and and how I can automate everything. And that's what's kind of been developed in my career is now whenever I take on a job, I think, how can I automate this? How can I control it? And how can my management team get intelligence from what I've done? So that was the kind of thought I had around that. And systems allow us to do that, to step away from just burning calories on tasks that can be automated.

I think this is brilliant. And I think I've found myself saying to business owners in the past, The way to create a saleable and scalable business is to make yourself redundant.

Correct.

So what you are saying is exactly this. It's like, how can you consist— now you take your experience and your wisdom and you turn it into an automated process that happens whether you are there or not, because that's where you build a saleable and scalable business. And can you give an example of like how you are, you've done that for a business using AI, for example?

Yeah, sure. So for a business that had a reconciliation process, every business has a reconciliation process. You know, they have to compare sets of data and then make sure that information is accurate. And then there's an action on the basis of that. This task was a very boring task for that manager. And when that person left, they found it really hard to hire someone to do this task around reconciliation. So what we did is we sat down with them and said to them, okay, what are all the inputs? And what are the process that the human has to do around this. I call it IPO, input, process, output. So now we understand the input, we can understand what the human does around this. There's data that you need to extract. When you extract that data, I need to reconcile, I need to match, I need to force match, and then I need to then pay someone or make a movement into the system. And that's the kind of output. So we're able to understand all those steps, build it into an AI agent that now sits on their mailbox. It reads all those emails as those emails come in. So now that inbox goes from, you know, 200 emails on a daily basis to zero within 15 minutes, which is absolute game changer for that team. But it also takes that data and then now pushes it into the next part of the workflow. So we call it text-to-action, T2A. Now the systems are working automatically with no human involved in that kind of process. And then there's what we call like a Q2, which says to the human, now you need to look at these exceptions. So what you're doing, and I try to do this for many decades with multiple million pound budgets in banks to try to get to true exception management. But with AI, we were able to do that now because now you're shifting the human from looking at emails to now only looking at true exceptions. And that was a process that took 3 to 6 months. Now, there were a lot of lessons learned there because when you introduce AI to people for the first time, especially in the corporate world, they get really excited and they throw everything at the AI agents, say, can it do this? Can it do that? And it can, but it needs time to be trained. People forget that the training component is so important. They, they see it as ChatGPT or Claude or whatever they use in their day-to-day lives. But when it goes into the corporate world, you have to continuously train the model because there's different edge cases. You know, your, your audience changes the way they send you data and you can't control that. So there has to be this continuous training process. So it's working through those different steps and making sure that the human is only looking at the exceptions and being part of that kind of self-learning. and that saved that team, you know, STP, straight through processing. So they were going from like 20% STP to now 70% STP, which is really powerful for that manager. They didn't need to hire more people because the AI works, you know, Monday to Saturday, doesn't take a holiday or sick day. So it kind of ticked a lot of boxes for that manager. So I was able to put that advice around the steps of how to build an AI agent from scratch and then get the benefits from a return investment.

Brilliant. And of course, this is amazing for that company because that company is now doing something quicker than it's ever done it before. It's doing it more profitably because it hasn't got to pay somebody else to do it. There'll be certain people listening to this that say, yeah, but somebody's now without a job that previously would've had a job. So that kind of segues perfectly into my next question, which is, if you could pass on, I have no idea whether you've got children or not, but if you could leave no money to your next generation, just some wisdom or some advice, what would that be and why, based on the world as it's changing through AI adoption?

So, like, there's again another great question. So it's building skills that compound. And what I mean by that is don't just stay in one lane. Look, continuously learn. You know, when I look at my career, I started out in operations. Then I learned about technology. I moved into technology and then I moved into trading to help the front office. I'm constantly learning, and I believe that is something that, you know, from the next generation point of view, we're in this world where AI is moving so fast, and I worry about this knowledge gap, you know, in the next 5, 10 years, you know, when these machines don't work, how do you solve those problems? You need to be able to pick up more skills, new skills, because the markets will change, the technology will move on. But you need to think more about building that kind of knowledge as opposed to kind of focusing on status and money. You know, money will always come and go, but the knowledge you need to build is so important. And now is the time to keep on learning and developing and thinking how you can take things to the next level. Brilliant.

I think that's fantastic advice. And I think it reminds me of, so Jim Rowan, I think I'm crediting the right person, but Jim Rowan said, I'll give you a guarantee. And bearing in mind, this was long before we were thinking about AI. This would've probably been in the '70s or '80s. But he said, I'll give you a guaranteed strategy to be wealthy beyond your dreams. And that is read a book a week. Yeah. So book a week is this constant learning that you're talking about. And then regardless of what's going on in the world, whether it's AI, whether it's mental, you know, health, those, that constant inquisitive learning allows you to deal with what's being thrown at you at that time and find where the opportunities are. So I think that's brilliant, brilliant advice.

Yeah. This is a bit also that the more you learn, the more you earn. Yes. So it's, yeah, it's just constantly running and running is also great advice as well. When you're running, you're thinking, you're talking to yourself. That helps as well control the internal thoughts.

And really, and that's a kind of beautiful thought, isn't it? Because Really what we are saying is that if we can take away a lot of those boring repetitive tasks that, you know, the human was doing previously, then their job becomes, and this is really the transition that an entrepreneur needs to make to build a saleable and scalable business anyway, is to spend more time thinking about the business, working on the business as opposed to in the business. And, and you're absolutely right. The best way, one of the best ways to think about your business or to think about anything is by going for a run. So when you're out running, so, so the task that the human does now becomes way more empowering, but also free of being stood at a desk or sat at a desk or whatever. Because the AI is doing that part of the job, probably.

Yeah, exactly. And with AI saving you time, you can now have a lunch break. So you can go for a run at lunchtime. You can read a book, you can take, you know, take that walk, get fresh, you know, fresh air. You know, the AI is kind of giving you that time back. But then you've gotta be smart on how you invest that time on health, on wealth, and, and, and, you know, reading and running.

I think this con— so Bill Gates and Steve Jobs were both famous for having a think week every, every, like, even, even in Microsoft's peak busiest time, Bill Gates was taking 2 weeks a year and just going off, taking a load of books and papers and, and reading them on his own. And that kind of ability to take time out and think is really empowering. So, If AI frees us up to be able to go, well, look, you can go off to the Lake District or the Cornish coastline or wherever you want to go. And all your job is to now is to read books and listen to books and consume, you know, because the AI's doing the boring stuff. I think that's an incredible opportunity. So what you've got, you've got this amazing business that's helping both banks and entrepreneurs adopt AI and automate the repetitive. What's the biggest obstacle that's blocking you from hitting your goals at the moment?

Sure. So it's, it's, it's not the technology. It's more to do with the hesitation. So too, too many organizations are paralyzed by, by choice now. So, you know, they're, they're constantly being bombarded by organizations saying that they can do AI. Now, firstly, what type of AI is it? If you remember the beginning of the call, we're talking about You know, Gen AI, you know, horizontal AI versus vertical AI, narrow AI. Also, is that AI a wrapper, which means that they don't fundamentally own the AI model? So it's, you know, for example, if you use something like ChatGPT, you know, etc., that's a wrapper. You can't get into the black box, you can't change the parameters, the guardrails of the data, you can't optimize it that much that you have that ability. So technology teams become an obstacle, and, and rightly so, because they're, they're also learning how to use AI in the, in their, in their world. And they, they will, and they will, they will get there. But there is this challenge where they're now saying that we'll do it ourselves, and, and they will be able to do it, but they need to partner. And I, I constantly have this conversation with, with my clients, and I, I always go back to Apple. Apple is obviously one of the biggest technology firms on the planet, completely disrupted everything they do. But when you look at what their AI journey was, you know, they did spend billions and billions of dollars on that kind of AI space. But why did they partner with OpenAI? You know, they're going to partner with OpenAI because they're going to, in parallel, build their own internal tech to have their own kind of Apple intelligence on its own. But they decided to partner with OpenAI because they were the market leader. They could learn from them. And we know from history what they did with IBM. You know, they dropped IBM from a chip manufacturer to create their own. Silicon chips. So we know at some point they're going to drop OpenAI. So they're the kind of challenges and obstacles I have is teaching those technology owners and technology people are responsible technology that, you know, it's not about build versus buy now. It's about build versus buy versus partner. And that's the new future that we're in now because there are amazing firms out there building these AI solutions. And I think you should just partner with them, learn, and then see strategically if you grow with them or you, you know, in parallel build your own internal tech.

Brilliant. And there'll be a lot of entrepreneurs, particularly people running professional practices listening to this, like maybe they run an accountancy practice or financial advice practice or firm solicitors, and they're looking at this going, where on earth do I start? What would your advice be on where do they start on their AI journey?

Yeah, sure. So I've developed a framework and when I started getting into this into a lot of detail the last 2 years, I was trying to find things out there that I could read, but there's nothing really out there around the kind of vertical AI space. I built this framework which is called the Autobot, the Vertical AI Autobot, and it goes through kind of 7 steps. The first step is all around— in Autobot is A is around what do I automate? So look at the tasks that are burning the most amount of calories that take 30 minutes, 20 minutes to do. There should be the kind of low-hanging fruit that you should go after to look to automate. Can the machine look at that? Then it's looking at the Unravel U, which is what data do you have to do that process? You know, do you understand? Is it unstructured? Is it structured data? And be able to kind of break down that data into kind of chunks that can then be fed into the machine. Then it's the kind of training, it's what I call the T. So training the model, how do you, you know, what, how many data points do you take, what sample you take, and then you look to optimize the model. How do I integrate it into my systems? Am I using Salesforce? Am I using Google Mail? What kind of interfaces do I need to optimize my AI agent? So that's the kind of 4 steps, it's called the auto. Which is the autopilot. That gets you to a pilot. That pilot then allows you to go out and tender for vendors to pitch to provide you an investment cost to develop your AI agent. And then the BOT, the bot is building it now. So how do you build it? You know, what do you kind of put it on the cloud privately or do you have it on-prem? You know, do you outsource it? And then once you've built the AI agent, how do you kind of operate it? You know, what's the kind of the guardrails around the operations, the straight-through processing? And last but not least is how do you then transform to the next set of AI agents. So that kind of Autobot framework helps a business owner go from zero to hero in AI with low cost, low impact, because that first 4 steps is them just putting their idea on paper. And once you put it down on paper, we can then break it down into kind of the chunks that is needed for the AI agents.

Brilliant. And so when I did— and I can see that working very, very well for sort of medium-sized and bigger businesses. I can also sort of hear a one-man band thinking, oh my God, that sounds like it's still really daunting. When I put together the AI Masterclass, which is like an AI course that we put out, I suggested for sort of one-man bands, maybe the starting point was just to get a good CRM system. So for example, like in financial services, the CRM systems have been pretty horrible for years. And yet there's a new system called Ningyi, which has really gone in on AI and it's got everything, you know, the AI sort of weaves into the whole process. And that to me was an easy way of a financial advisor just going, well, look, if I adopt their software, and I suppose it's partnering in the sense that you maybe, maybe you said essentially if I adopt their software, then they're going to do a lot of this research and build for me. Do you think that's a sensible way of going about it?

100%. 100%. Because yeah, like you said, those one-man-band entrepreneurs, those solopreneurs, they having that kind of partnership with, with a firm that can solve the, like you said, like the CRM problem, and that's a big challenge. Also sort of landing pages on websites. Again, that's very easy to do with AI now. So you can just sort of drive, drive traffic, you know, into, into your, into your sales funnel or so those are kind of really easy, easy kind of accelerators. I totally agree with that, what you do in your AI masterclass. Brilliant. It's just that people want to just know where to start. And I think that's the right step. Is look at that kind of low-hanging fruit and then go for it.

And if, and if you're partnering with a software partner, do you partner with one that's embracing AI as opposed to one that's resisting it? And you'll kind of ride on their coattails, won't you? Yes. So I always ask everybody what book or framework or mentor has really helped them on their journey. Obviously you've shared your framework, but what books or other resources have you found that have helped you see the light in this, in this aspect or something else?

Yeah, so for me, you know, I work with a company called Intellimation.ai, and so they, on their website, they have a lot of, you know, use cases, a lot of frameworks. And they've taught me, you know, the founder, Harpalak Badur, he's taught me ACI. And ACI is Automation, Control, and Intelligence, and how you can use AI to kind of tackle those three key areas. Because that's what predominantly you're trying to get with these AI agents. How do I automate something? How do I then improve controls and to make sure that nothing's going out that shouldn't go out? And then how now do I try to get intelligence from that? So that's a framework that I've learned through Intellimation.ai for the last couple of years. And then, you know, there are other books out there that, you know, I sort of follow, you know, and read and stuff. So again, it's, you know, I like the kind of Rich Dad Poor Dad book, which you've got to now start going back to early questions around you. You've got to think about how you can sort of make money work for you. And now in this time we're in, AI can predominantly do that for you, be able to create those solutions to help people and sort of coach and train them. Brilliant.

Love it. And then finally, to finish off, for people that want to find out more about you, find out more about your companies, where's the best place for them to go? And obviously we'll put these links in the show notes, but where, where's the best place to, for people to find you?

Yeah, sure. Couple of places. So website-wise, it's coglateral.ai. So that's where you can see my thinking, the frameworks I'm doing, sort of real, real-world applications. You can also follow me on, on LinkedIn. LinkedIn, but where I'm sort of reflecting and sharing what I'm doing in the kind of the corporate and non-corporate world. So they're, they're two good places to learn more about myself and my company.

Amazing. Uh, Cheetan, as always, it has been a pleasure interviewing somebody on the podcast, but especially this one because we've been talking about a subject that is so close to my heart. I, I can see us continuing this AI conversation for, for quite some time, but it's been an absolute pleasure chatting to you. Thank you very much for, uh, for coming on and for sharing this wisdom.

Thank you so much. I really appreciate it. I wish your readers and followers all the best on their AI journey. Thank you.

Brilliant. Well, thank you. And I think, well, we'll obviously put all the links in the show notes. I think if you're trying to think about how you take your AI journey forward, Chetan is definitely somebody that you need to be tracking down and following. So another amazing interview on the Progress in Practice podcast, which as a reminder is brought to you by the trusted team that helps people build a saleable and scalable business. Until the next interview, thank you very much. If this conversation resonated, you'll find more support, structure, and community inside The Trusted Team, where ambitious service-based founders come to build businesses that are scalable, saleable, and sustainable. You can find out more at thetrusted.team and If you're enjoying Progress in Practice, please make sure you follow the show and leave a rating and review. It really helps more founders find these conversations and to bring you better and better guests. So until the next time, keep progressing.