← Embracing Digital Transformation
Loading summary
Transcript44 lines
- [00:00]
A
Usually you have lots of posts where, where you'll say, with no coding experience, no experience and no skills, you too can pick up this thing for $5 and revolutionize your business. And I think the expectation that this stuff is easy needs to kind of go away a little bit. You have to put work in to get something out.
- [00:21]
B
Welcome to Embracing Digital Transformation, where we explore how people process policy and technology drive effective change. This is Dr. Darren, Chief Enterprise architect, educator, author, and most importantly, your host on this episode, the Myth of Easy AI. What leaders keep getting wrong with Dr. Ashwin Mehta, founder and CEO of Metaology. Ashwin, welcome to the show.
- [00:55]
A
Thank you for having me. It's, it's a pleasure and hope it's going to be a really interesting conversation.
- [01:00]
B
Oh, I'm sure, I'm sure it will, because we had an interesting conversation a couple weeks ago when. Yeah, when I first talked to you, I said, oh, we have to have Ashwin on the show. Great background, great insight. But for my, my viewers, everyone knows of my listeners that I only have superheroes on the show. Every superhero has a background story. So, Ashwin, what's your background story?
- [01:24]
A
Yeah, so intro, of course, is I'm the founder and CEO of a company called Methodology, which is an AI and tech consulting business. But my, my story started in the midst of time many, many years ago. I, I started my career in chemistry, and that was many, many years ago. It was a very interesting time for the chemical industry. Moved through consulting, moved, did an MBA and a PhD in statistical modeling of human behavior, started down the path of data science and machine learning from there, did some work in a big consulting firm, big four consulting firm, and also then moved into a global pharma company. So I've had a wide variety of experiences across quite a few different disciplines that companies usually have. And most of the stuff that I do now is, Takes that multidisciplinary approach.
- [02:20]
B
I was gonna say you're like a Renaissance man.
- [02:22]
A
I mean.
- [02:24]
B
And I also noticed the guitars in the back, something you didn't mention. You didn't even mention it.
- [02:31]
A
No, it's, you know, there's, there's, there's this dichotomy of the things that you do for fun, the things that you do for, for a living. But yeah, I played these glorious guitars that you see behind me. I've played these for probably 35 years, and I think each of these has got a nice little story behind it. But this one in particular is, it was a rare find. I started about maybe 5, 10 years ago maximum. Looking at these vintage RGs. Now we call them vintage nowadays, but we call. We're talking about the mid-90s, the kind of golden age of, of these RG series. Ibanez and I found this one on RG 550. And fortunately it was one of these which was factory scalloped according to the guy I was buying it from. And scalloping means that. Let me see if I can show this to you guys. But if you can see this in between each of the.
- [03:33]
B
Yeah, look at that.
- [03:34]
A
The frets, it's kind of cut out, which means you get a very different look and feel. It comes to playing a little bit like sitar. So I don't know if you know what that is, but I also play sitar and you get that feeling when you press down that you're not actually touching anything. So the, the play experience is very different and allows you to be a little bit faster. So that's, that's my custom guitar, new pickups and everything is just customized for my, for my playing my fingers. So that's the baby, right?
- [04:06]
B
That's. That is so awesome. And we talk a little about music. Music's really important in, in our society. And I, I'm sure it, it guided you through, through most of your career as well because it just, it's for, for me it's, it's very important too. So I, I love that we have a little bit in common.
- [04:29]
A
There's that creative side which, which is important, but the other side of it is very interestingly for me that as. As I was growing up and I was in bands and you had to do all of this stuff like, oh, we're in a band, we have to have a website, we have to go cut an album, we have to do. And it starts to take you down this path of interacting with tech in a very different way. And probably that experience has really helped what I do now.
- [04:56]
B
That's. That's awesome. That is really awesome. Let's move from music. I thought we could spend a whole hour on music, but let's move from music to, to AI. Generative AI. You're at the throws of this because you're out there consulting with companies, trying to do something with AI. All the CEOs. There's great memes out there that a CEO comes in and says, we need AI and what for? I don't know, but I know that we need it. Are you seeing the same thing? They're coming to you saying, ashwin, help us with AI. But they don't know what to do.
- [05:32]
A
Yeah. So I think the key signals that start conversations for me, the clients, are we want to do something with AI, but we don't really know what. Or we have this massively sprawling landscape that is evolving day by day, and we don't really know how to make sense of it. Or what is AI. It's very scary for us. Is there anything that we need to know? So there's probably those three signals that start a conversation with a client. This thing about we want to do something, but we don't know what it is. It's a very interesting meme because it's so true. But also, you have to kind of unpick a little bit. Why. Why are people curious about this in the first place? And sometimes it's a driver in the business, sometimes it's a leader who said something, and, you know, everybody needs to go off and do AI, figure it out. Or maybe it's a competitor, you know, some other business in their sector is doing something, and it leads them to want to be curious. But it's absolutely true. It's a. It's funny because it's true. Right?
- [06:34]
B
Yeah. So. So you think there's a lot of fear then or. Or fear of missing out or what? What is it that's driving everyone in such. I. I haven't seen anything like this since probably the 90s.
- [06:49]
A
Yeah, I think it's a little bit of fear of missing out, but I think there's the rumor mill. And the rumor mill, I think with many businesses is they have somehow cottoned onto the fact, and it's a fact that there could be cost savings if they think about the world of AI and businesses are under pressure, budgets on the scrutiny. Cash flow isn't what it used to be in many businesses. So we have this kind of budget imperative. Can we get efficiencies? And while that is a potential thing that companies can look for in the AI space, it's not everything. You know, we could think about optimization, we could think about different outcomes with tech that we have. But one thing I do find out in the wide world of industry is that there's an expectation of efficiencies, but there's also an expectation that things are easy. And if we look back to the wall behind me. Right. You know, we will. If we think about the instruments that we play, we. We go off. I play guitar, you play piano. Or if we think about the sports that we play, you know, I used to do martial arts, and some people maybe did football, soccer, or, you know, American football or something like that. When you first touch a football, when you first pick up a guitar, you're not automatically in a stadium playing to thousands of people. Note, perfect. Getting it all right, this expectation that things are easy, that if I just pick this up, if I just ask one or two questions, somebody will say, there's this tool over there that's just going to solve all of your problems. And this, this kind of expectation is not helped by social media, which usually you have lots of posts where you'll say, with no coding experience, no experience and no skills, you too can pick up this thing for $5 and revolutionize your business. And I think the expectation that this stuff is easy needs to kind of go away a little bit. You have to put work in to get something out.
- [09:17]
B
Do you think that is starting to go away? I mean, because what. MIT just came out with a report that 95% of gen projects are failing. Do you think people are starting to catch it?
- [09:28]
A
I mean, let's talk about that for a second. I think I read that paper and I was surprised in two different ways. So one is I was surprised at the bold statement of. It's pretty bold, yeah, it's a bold statement. And the other is I was surprised by the quality of the research. I thought it was pretty low. And methodologically speaking, I would have expected more from mit. The bold statement, of course, if I think about my experience with companies, many big companies decided to go down this journey of November 22nd. ChatGPT comes out January the next year. Companies are saying, right, we're going to have our own version of this thing without having the technical experience, without having the, the ability to implement and maintain its scale without really having a purpose. Just saying we're going to do it because we're expected to do it. And it comes back to the strategic use case kind of idea. If you have something that you want to do and you mobilize people and technology and all of these things against that purpose, you have things that you can measure. You can expect some value, generally speaking, or you can expect the experiments of fail in a measurable way. So we have something we're going to measure. And I think a lot of companies just went off on this little odyssey without necessarily having a purpose, measurable success criteria, all of the, all the skills to maintain something. So you get sprawling projects that go on and on and on forever. And why would we be surprised that these things didn't generate value? I think it seems obvious, right? But the other side of it is. I was reading the method section of the paper and the bold statements around we're not getting workforce adoption of AI. And the method says, what we did is we asked a bunch of CEOs if they thought that workforce adoption was happening without asking the workforce in the slightest. And you know, I think that there are limitations, of course, but I, I wouldn't have thought that MIT have never done a workforce survey. It doesn't seem insurmountable to ask folks. So I thought there were, there were various aspects of this paper that need unpicking if we're gonna, if we're gonna dig in.
- [11:51]
B
You know, I actually think it leads to what we're seeing in the industry as a whole. It's this, I need speed and I cut corners to get to that speed because if I don't do it, someone else will. Because AI is speeding everything up where there's a lot of work slop that's going on because of it. Right. A lot of half done research, a lot of I just need, I just need some product out because it'll, it'll work itself out. I think we're seeing a lot of that right now.
- [12:22]
A
Yeah, the kind of, in the software space, the idea of perpetual beta. So nobody's ever going to release anything. They're just going to put it out for, effectively beta.
- [12:31]
B
For beta or alpha.
- [12:33]
A
Yeah. And just see where it goes. Yeah. So, yeah, it's a strong statement, this idea of 95%. But I think companies need to kind of ask themselves, well, how many of your initiatives before AI led to lots of value? Was it all of them? Is the, is this kind of an unrealistic expectation? So these things need to unpick. But I do think that even in the, especially in the AI space, we have a dilution of terminology, we have a dilution of understanding when it comes to some of the basics. And how many companies were doing fairly good data science, fairly good ML, they had data pipelines, they were doing this stuff on pattern recognition and they were able to get something out which was valuable before gen AI. And now all of those projects also being lumped into this kind of broad idea of did we deploy Copilot, did we deploy Gemini, did we deploy ChatGPT? And we deployed it to a bunch of people with no use case. We didn't do anything in the way of literacy training and we just said, what do you think? Go off and play with it. And people made a bunch of cat pictures, they made a few cartoons out of selfies, and maybe they wrote A few emails, and that was it. So I think. I think we need to be realistic about what companies should do and what they're actually doing, and where those two things marry up. We can expect value. Where there's a big gap, I think we can expect a miss, a mismatch in value.
- [14:07]
B
So we already do see some value, right? I mean, you already mentioned that. Dancing cat videos. I mean, those are. Those are great value, right? No, or. Or the email. People use it for a lot of email. But you're talking, like, fundamental business advantage, business efficiency, and things like that. So how do I get started then? Do I just take my list of things I was already going to work on and now try and apply AI to it? What's. What's your approach to it?
- [14:39]
A
So I'm gonna. I'm going to invoke the social, cultural phenomenon that is Dragon Ball Z at this point. I don't know.
- [14:47]
B
Okay. All right. Yeah. Oh, yeah.
- [14:49]
A
But in Dragon Ball Z, there is a transformation that happens, and I want to. I want to align that to business for a second. So the transformation that happens is this guy. If you've never watched Dragon Ball Z and you're watching this podcast, I urge you to go and watch Dragon Ball Z at some point in your life. But there is a transformation. The guy who needs to do a thing, you probably don't need to know any more than that. He needs to do a thing and he surpasses himself. He transforms and he says, this power comes because there is a need. And if we think about that statement, that is a statement that applies generally to business. Do you have a need? Do you have a pain point? Do you have backend operations where you know the efficiencies are not there? Do you have customer cycles or sales cycles where you're not able to meet the needs, you're not able to respond to queries, you have customer service issues. Are there marketing challenges that you're having in your business? Are there operational or manufacturing challenges that you're having in your space where you're not able to match things like incoming materials to outgoing production? And there's a gap in those things? So there are lots of potential use cases across businesses where you could think, we need to do something differently. That's really the starting point for all of this stuff. And this is rooted in management consulting. The second step, of course, is we need to do something. And that solution could potentially be a technology solution. So it could be an AI solution. Now it's fine to go and play with things of Course, but there needs to be a solution, problem, solution, fit, otherwise we can't really expect value. And that requires a little bit of knowledge of the market. It requires probably a little bit of knowledge of your own company to figure out what it is that you're going to do. Now in the old days, the old days, we used to have rpa Robotic Process Automation. Yeah, yeah.
- [16:47]
B
Five years ago. The old days.
- [16:48]
A
Yeah, yeah. It seems like a distant memory now, but that stuff is still very applicable today. And base the basics of automation are probably going to be very valuable because nowadays the barrier to entry is very low. There are so many platforms out there, nearly every large tech provider, the, the saps of this world workday and all of these folks, Salesforce, you know, a lot of them are doing things in the automation space within their platforms and there are dedicated automation platforms as well. So in that kind of space, businesses that are trying to look for a tech solution and they're trying to look for efficiencies, start there, start with your processes, start with some broad based automation and see if you need AI, because not everything needs to be AI. Right, right. So I think that would be a really good starting point. Understand yourself, respond to a need and then map detect that need and see if it's an AI problem.
- [18:00]
B
Do you think that the AI aspect of this is so new that people don't know how to use it? I, I don't even think we've even tapped into even a partial, the partial power of what gen generative AI, specifically natural language processing and context aware Natural language processing is, is actually, can actually bring to process improvement. Because what you're really talking about is process automation and process improvement. Do you think that we're just on the tip of that or do you think it's well understood enough where I can start leveraging it to actually get stuff done?
- [18:42]
A
So I think the tech is capable enough that people can leverage it and get things done. But I would start off in the simple space rather than trying to start off with something very complicated. And if I just partition this ever so slightly into potential use cases. So if we have a potential use case where for example you might have an inbox where customers can request the product brochure, an email comes in, there'll be something that says request brochure in there or request something about a product and then you want to respond with an email that says thank you for your inquiry and we want to potentially include a PDF or something like that and maybe a diary link so they can book a meeting with the salesperson, that's a fairly typical use case that's going to be broadly useful. Now we can break that down into something that is basically an automation that doesn't really need any artificial intelligence in it other than scanning for keywords and that that simply email comes in if includes keyword, et cetera. If we take that a step further and you want to analyze your entire inbox for 25 different types of query, we start to get into the space where we could build complex automations, or we could use something that's a little bit agentic, that has the capability to scan, triage, make some decisions and then send things off across multiple workflows. We get a little bit more complicated in that space and we say, right, well, what we want to do is we want to have some nuance in nearly everything that comes into this inbox and we want to have the capability to answer complex queries, respond to FAQs, give folks the answers that might be on our database. So we might go down a little bit of a retrieval kind of odyssey to say we need to have retrieval capability and we need some kind of agent that's going to make decisions based on inputs and retrieval. Already in a few sentences we see how things have become quite complicated from that first piece and the platforms that are out there at the moment. You have the capability to daisy chain multiple LLMs or LLM instances, to do different tasks, to think, to partition out a problem into various components and then take action on components. And the reason I explain it like that is the tech is there right now to allow us to be able to do that. I won't necessarily go into the world of super agents. You know, we've got something in the browser. Here's my credit card details. Just go on, go and book me an itinerary. Yeah, I think I would hesitate to go down that road right now, but I think businesses that are able to understand their own processes can be somewhere on this journey that I've just articulated with the tech that we have now.
- [21:41]
B
So, so you can take a lot of the things that we learned during the RPA revolution and now apply it. And now it, it almost seems like I just have another tool that I can add to it that can actually accelerate the automation or the replacement of, of processes. I know I've done this with, with some of our customers where I came in and just completely wiped out some processes that they were there. Primarily historical, ad hoc. How do I convert human stuff to computer language and computer language back to human stuff? I Can get rid of a lot of just processes in general by looking at them differently and understanding how different tools can be used. So we can take a lot of the process re engineering for no better word that we've already started in the RPA process and now apply it even, even more effectively. Is that what I'm hearing?
- [22:43]
A
Yeah. And we can apply it not only more effectively, but I think the key thing for me is that the barrier to entry is now very, very low and that comes from a handful of things. So the first thing is of course we have very capable platforms now which are easy to understand, maybe low code platforms which always get better if you have the ability to code, but if you don't, you can start somewhere. I think the second thing is we have a proliferation in the market of readily available APIs. So most third party tools now have APIs where 5, 10 years ago this was a difficult thing to be able to create. And the third thing broadly is integrations across a landscape. So you have a number of different types of integration now you can call whether it's APIs or model contacts, protocol or agent to agent. If you want to do complex things, you can do that. If you want to add code, you can go and ask any of these agents to generate some code and it might be okay, depending on what you're asking. So we have a lowering of the barrier to entry in many different domains all at once. And that's very interesting and very exciting. But absolutely, yes, I think businesses can go and take action on the traditional automation stuff in a way that's very easy now.
- [24:15]
B
So, so does that mean that I got to hire experts to come and do this stuff? Because that's a lot to learn. I mean, it really is, it's a, it's a lot of so. And where do I find these experts? How do I know that they're real experts when they say they have five years experience in this? They don't. Right. I mean, because. So where, where do I go if I'm a, if I'm an executive or even if I'm a, you know, a small company that I'm trying to get things going?
- [24:43]
A
Yeah. So I mean, the first caveat is that it's never been easier to learn this stuff. We have information absolutely. Everywhere you go on YouTube, you ask any of the agents that you're working with, the chances are you can get to something that's workable pretty quickly, but it requires you to have the time and the aptitude and, you know, not to be scared off by Certain things. The second question is slightly trickier and that's where do you find experts? There are self proclaimed experts. Yeah, yeah. And it's sometimes difficult to scrutinize the person who's popular on LinkedIn versus the person that can do the thing for you. And it's difficult. I'm not going to sugarcoat that. Many companies will learn the hard way by engaging a self proclaimed experts on LinkedIn versus going to folks that have maybe some experience. I would say the categories of experience are probably telling. So there are companies out there who have maybe a consulting spin on things. Now you may, if you're watching this, you may have opinions on consultants. I'm sure you do, but there are consultants and consultants. And I think any, any folks who've been into the tech consulting space probably have a methodical approach. The same with management consulting. Probably there's a methodical approach which you can scrutinize. So when you get a proposal from a vendor, you can look at their approach and say, does this systematically seem to make sense? So for me, I would say if I'm going to work with a client, I'm going to try and identify systematically use cases against an approach. So we might say, well, we have 25 use cases in this business. How do we prioritize those things? We have some that come from different sectors. We want to go down a method that will help us to prioritize against timescale budget skills, data availability, tech availability, you know, many factors that would lead you to be able to rank some of these use cases. And then potentially you want to go into some kind of rapid prototyping with those use cases, depending on, depending on your tech landscape. So it doesn't have to be something that's out there. Many companies do have a tech landscape. They do have maybe a Microsoft or Google system, maybe they have a handful of other key things that they're bringing in. Maybe they have some point solutions. You don't always have to build everything. Maybe something in your stack already delivers the capability that you want. Yeah.
- [27:31]
B
You may not even know it.
- [27:32]
A
Yeah. And I think a consulting approach that is systematic in this way is more likely to produce value than somebody who has a strong opinion about AI and maybe talks about the fluffier side of AI without knowing the technical details. So for example, a strategist might talk about, we don't want to replace people, we want to support people. Great. It's a talking point. It's not a strategy, it's a talking point. We might say, well, what about Ethics. Scrutinize the word. If they're talking about ethics, do they mean ethics or do they mean bias or do they mean governance or do they mean data security?
- [28:14]
B
Right.
- [28:14]
A
And then scrutinize the approach. Are we actually going to get something prototyped by this particular vendor or are we not? And these kinds of questions I think probably sift out most of the, most of the opinion market from the delivery market.
- [28:32]
B
I love how you, how you differentiate opinion versus delivery because there's a lot of opinion out there, not a lot of delivery. As, as we're already seeing, whether we believe the MIT report or not, we all know that successes are not as high as, as we want. Right? But you know how that goes. Hey, Ashwin, if people want to learn more about you and your business or want to reach out to you, how do they go about doing that and where can they learn more? Because I really love your approach. I think it's great. Ashwin is not an opinion guy. He is a, you know, let's get down and deliver something. So where do they find out more about you, Ashwin?
- [29:21]
A
Absolutely. So the social channels. If you find me on LinkedIn, if you want to follow me on LinkedIn, I post regularly podcasts, articles and how to kind of posts. You can also contact us via our website methodology.com spelled like my surname. And their white papers are. There are actually quite a few white papers on the website which are frameworks of guidance for, for things that you can go and do yourself. And then if you need further support, of course you can contact me. Those are probably the main channels. Follow me on Instagram I don't have many followers on Instagram, but there are regular posts that come out around, you know, podcasts and things like that. But any of these channels you can contact me for not only strategy and vision, but also the things that make it more tactical. So governance, tech, landscaping and data.
- [30:17]
B
That's awesome, Ashwin. This has been, this has been very insightful. I appreciate it because you're, you're really helping us. Let's get to some bra. Let's make something real out of this instead of a whole bunch of fluffiness. I love how you called it that. So that's awesome. Thanks. Thanks for coming on the show, Ashton.
- [30:35]
A
Thank you for having me.
- [30:39]
B
Thanks for listening to Embracing Digital Transformation. If you enjoyed today's conversation, give us five stars on your favorite podcasting app or on YouTube. It really helps others discover the show. If you want to go deeper, join our exclusive community@patreon.com embracingdigital where we share bonus content and you can always connect with other change makers like yourself. You can always find more resources@embracingdigital.org until next time, keep embracing the digital transformation.