Levelling the playing field – Data centric design – [FULL INTERVIEW]

0:02
Hi, everyone. I’m here with Hemant Patel, who’s the CEO and founder of anumana. And they’re in the data analytics space. So Hemant, thanks so much for joining me, I really appreciate it. Thanks, Chris, for inviting me. So, as I say, you’re in the data analytics space. And that space just seems to be burgeoning. Over the last little five years with data, expanding hugely analytical techniques we’ve been put to play in ever more detail, what are some of the trends that that you’ve been seeing? I mean, no doubt, you’ve been keeping pretty busy. What are some of the trends you’ve been seeing in terms like this sort of transformations happen over the last like five years or so? It’s really interesting, because the technology, and the techniques have been around for a long time, got industries, like financial industries, particularly investment banking, that will well ahead of the game, then that purchase was probably a leader when it comes to data and analytics compared to other industries. Now, as I’ve moved out of the debt production industry into other verticals, have seen sort of a quiet lag in the capabilities, particularly sort of across the commerce space. And we’ve been in utilities. But what we’re finding is because technology in itself, the infrastructure has advanced so quickly, over the last few years, with Chad GPT, with mid journey with the generous of AI modules and applications that are coming out, is really bringing into the mainstream, what technology can now do to enable advanced analytics, which is where the conversation is becoming more prevalent. What’s what’s changed, what you’re saying is, I agree, so some of the analytical techniques have been around since the 70s, to a certain extent leave mathematically right. It’s almost like the is it just that the cheapness of processing power, and then that gives us just wider accessibility, so almost like the finance groups had the money to be able to invest in it because it was decision making and as a return there, but now it’s just the low cost, the infrastructure means it just accessible to a much wider kind of group, and we’re getting fintax, basically developing new techniques, new kind of software to help us use it exactly. I think the actual analytical techniques that are being applied aren’t as innovative as the technology that it sits upon, that the technology has brought it a lot closer to home, it’s accessible. Irrespective of what your own machine is, just with cloud infrastructure, the ability to use cutting edge processing power cutting edge machines, is available at fingertips individuals, which then means that it’s a far less or far more reduced cost to to really get to the cutting edge, obviously, 1015 20 years ago, we need to invest in big machines, the infrastructure itself was the limiting factor, that now with engineering techniques, and with different processing power with cloud technology, it’s all

3:05
there to enable cutting edge analytics. And now we can really move the dial more what does that mean, with the limitations? What’s been the impact of things like? Remember, the Internet of Things that used to be that there was really around gathering data or was like, a much more granular level, what’s been the impact of that, and I suppose, and just our digital infrastructure in terms of the measuring much more, because it also feels, yes, the infrastructure has changed. But also, the amount of data we have access to, because we’re recording things digitally, seems to be way bigger than it ever seen than ever was before. It’s almost like it’s built on top of all of these different technologies to get us where we are today. Absolutely concepts of big data for a long time now that we’re now in a phase where

3:49
you don’t have to be a real expert. When it comes to decision science and data analytics to interrogate large realms of data or even broad varieties of data, we’re now looking at utilising the technology itself, rather than the techniques to be able to really go into deep data mining techniques. And that gives a lot more access to businesses and individuals to be able to really compile

4:23
big datasets for the purpose of generating value for businesses. And I suppose from a data science point of view, and I believe large language models to one side for now, we’re just talking about regular sort of machine learning AI and the different kind of like modelling techniques that arise it. I know there’s different versions you can use, but it’s still there must be like that human element involved in terms of spending a lot of your time around what is the business problem and trying to solve for and then getting the data to solve for that? Because that constitutes a lot of like the day job still, for most data scientists haven’t sent for me as a built data and analytics teams over a long time frame the difference between

5:00
Good day, and analysts integrate data analysts was almost an art rather than science. You could have someone who’s highly technically capable. But if they couldn’t understand the language of the business and understand how to interpret and ask the right sorts of questions, that we’re never going to come up with the great data analytics outputs. So that still remains the same, that’s still important. And now more than ever, because you’ve got that quicker access to the end product, which is now available. So

5:31
with every project we do, it all starts at the same point, what are we trying to solve for? What is the business value? Data and Analytics is a tool, it’s not a product, it’s not an output, it’s a tool to get you to a product and output. At the end of the day, you’ve got to understand what is it in for the business what you’re trying to achieve? But in terms of efficiency around data science, the you’ve got the what is the business problem, you’ve also got cleansing the data and making sure the data then it becomes applicable and those kinds of things. Because that data cleanse, probably the area where there’s probably more efficiencies to do it automatically versus because it still sounds like there still needs to be that human kind of interaction, that piece is going to be difficult to automate or push away. But there’s probably efficiencies to be gained around the data interpretation piece, which can then give you the outputs. Yeah, I think

6:24
from the start to the finish, there is an overarching framework that you apply, and Eve parts of those, that framework, there is an opportunity for increasing the efficiency for automating, I don’t think there are any parts of the framework that are that you can that are protected from automation. But there is always going to be an element of human nature to extract that extra element of value. I think what’s becoming really important now is the concept that data and analytics is a symbiotic process would previously wouldn’t have been an organization’s data and analytics app separately, there are silos and a lot of businesses are starting to understand that the role of the CDO is one that covers a larger space, you can’t have analytics, reporting and finance and data reporting into it. And technology, you need to have someone at the centre managing the whole process. And that enables you to understand how engineering can enable analytics and analytics can create the requirements for engineering, and they’ve got to work seamlessly together. Yeah, I was on a call the other day, one of the conferences the other day, and we’re talking a little bit about what does the process look like? So we used to think very much in terms of this is the process and the moving linearly, A to B to C to D, whatever it is, and, and as long as like the data would then just fall out, or that we would just observe the data through the process on the outside. At what point are we going to turn it around and say, Actually, it is about the data, and you design the process around gathering the data throughout the process. And then the data allows you to then get better. So rather than it being in our mind, in the real world, we think about the process or the report or the output rather than how do we design the data that can then give us the output? Do you think we’re ever going to switch that round and become way more data centric, even in terms of like process design, even in terms of agent conversations example is like, what data are we gathering? And what does that tell us? Then we can use that for then a lot of other things, and it really is about the data, not the process to send it to them?

8:29
Absolutely. And we’re seeing that now with some organisations that are

8:35
that aren’t data and analytics mature, from a capabilities perspective, but from a cultural and a, from a cultural perspective, they are mature, they are thinking about that end objective, they’re not just thinking about how do I get a report? Or how do I answer a specific question. They’re thinking about, like, how do I generate business value from data? And a lot of that comes from, or what data do we have? And what data do we need? And how do we go about understanding what that end objective is so that we can start collating data earlier on, and giving ourselves that richness in the future? Because it isn’t an overnight processes. It is. It’s an investment of time as well as money. And the sooner you start, the more value you’re going to get.

9:29
Other some industries that further ahead than others. You mentioned financial services versus retail as an example. But even within that, and subsections within that who’s doing this stuff really well versus I suppose less to a certain extent, but who would you point is, these industries are really doing really well on these areas of industries are really doing? Well. I think I saw a stats about the percentage of footsie 500 companies that have a CDO in place, and I think that was a perfect metric to demonstrate which industries are doing it

10:00
in banking and insurance CEOs were in 50% plus of industries in retail, it was below 20%. I think that’s perfect metric there to demonstrate exactly which industries are out of the game that were chartered, but doesn’t mean to say that within those industries, there are businesses that are really advanced and pushing on, really the periphery of analytical and technological capabilities ahead of other verticals. But certainly as a collective, there are, I’d say, retail utilities are our industry industries that are coming into that space now. And what about large language models? Are they really new? Is this something new that’s coming on? What’s your kind of view on them? When I use them? As a consumer? They seem like magic. But I mean, as a data scientist, what’s your kind of professional opinion? Personally, if I’m honest, I wouldn’t call myself a data scientist, there are a lot of other people that are much more capable than me with data science in particular. But from my experience, I think this is really where technology has enabled the mainstream use of large language models, the techniques, the techniques, I’ve been there for a long time, it’s just as the stock has been able to

11:15
bring that to the consumer, that’s a really been the last few years that move the dial forward. And now,

11:25
innovation is just dry for the moment, there’s so much out there, God knows what’s going to come out in the next week, let alone the next couple of years, is great to be in a position where we really got to push the boundaries of innovation now, because the technology isn’t, it’s becoming less of a limitation for us. But yeah, I think it is now the techniques aren’t new, the technologies are enabling it. But now the techniques are becoming more prevalent. For for individuals.

11:59
I mean, what I find particularly interesting is how you can almost take almost like weaker signals, but a lot more signals to then combine that to look at patterns to come out with an output bit like we do with language, right? So you have or even if you observe something, you got all these like inputs coming in, you got a massive inputs. And it’s the how do you observe that to then come out with something? So for example, one of the examples we’ve been talking about, can you use these techniques to look at brain science as an example, because there’s so much that we can’t comprehend it, understand what’s really going on? But maybe we can use the computer to help us to understand that. Do you think there’s, do you think there’s applicability for those kinds of techniques in retail, as an example, or even financial services? But yeah, I think, and I guess this comes back to the animal and a business model that is predicated on a belief that have that businesses now have the opportunity at a much smaller level, to develop their own proprietary AI solution. And when you have that ability to develop your own proprietary AI solution, you can solve your specific business problem, which gives you the opportunity of solving problems that were previously impossible or, or thought to be improbable. And that opens up a real

13:16
expansive opportunity, that for me, I don’t think

13:22
I don’t really see a business problem as being the limited factor in whether you can apply AI. I think it’s really understanding

13:32
the data space and the value in investing in AI to solve a specific business problem. So if the scale is there, then we’re going to be seeing machine learning applied to many different business problems that previously weren’t.

13:49
And it’s interesting, how much of a threat Do you think it is? to large businesses, the fact that small businesses will be able to compete almost like on a much more of a level playing field, and certainly in terms of cost efficiencies are one of the interesting things you can use, you can use it to generate cost efficiencies, because it you don’t have to employ an army of people to do it, you can do it in a much smaller scale. And so that that cost efficiency size of business goes away to settings, which makes small businesses maybe more competitive, do you think is that? Do you think that’s a threat to a much larger businesses?

14:22
Yes, and no, I think yes, in the fact that it does give smaller businesses the opportunity to scale incredibly quick, a lot quicker than previously, resource. Human capital is no longer necessary a limitation for growing business, which has always been the case. However, with larger businesses, as long as they remain ahead of the game and remain, they’ve got the capital to start off with a theory they should have the capital to invest in, in innovation in analytics and keep that competitive advantage. So I think

15:01
For me, I think it allows smaller businesses to become successful a lot quicker. I think it allows them to establish themselves and create market presence, whether they then can compete with other organisations and larger organisations, the end of the day, the biggest challenge to small business is going to be the wealth of data available to them. That’s something that they unless they’re acquisitive, unless they find alternative data sources, that something they really can’t compete with local organisations on that they can be agile, they can be nimble, and they can be innovative, which could give us

15:42
I suppose, I said that sort of thinking about businesses in financial services should tend to be very data driven rather than retail as example, we’ve actually got, you might have actually retail stores as an example of a physical plant, right, which is, that still doesn’t go away. Right. So as much as we talk about data, and these conceptual business models, the physicality of businesses sometimes just doesn’t go away into, like, transport logistics, those kinds of things. Yeah, there’s so much technology available that actually there is we’re looking at computer vision as a technique that’s enabling small businesses to understand how their

16:20
house how consumers are moving around the store to be able to understand the desktop opportunity of placement, product placement that provide that’s available at a really small scale, as well as a large scale. I think, in my experience, at the moment, I’m seeing believe that over 90% of businesses are data analytics, illiterate and

16:41
naive. So I think there is a huge space for creating competitive advantage by just starting early adoption. What about software programmes for doing some of the analytics, right? So that’s been a bit of a burgeoning market that SAS was one they used to be way back, there’s the RS come in, I know that you do stuff with Python as well, which is feels like that’s kind of like taking over to a certain extent, what’s your view on packages for doing some of the analysis? And what works? What doesn’t? And where do you think it’s gonna go?

17:10
I think I mean, there’s,

17:13
I think people can get obsessed by learning multiple different programming languages and become, it becomes a tick box exercise, either I know our Python and JSON, language, etc. And one point, I think, what’s the purpose of it?

17:32
At the end of the day, for me, I think Python is pythons most popular language in the world. I personally haven’t learned to yet rely on moment scheme to utilise it. But for individuals looking to enter the data analytics market, I think it gives enough accessibility to elements of engineering elements of software programming or elements of, of analytics, that it’s a really good entry point. And you can do pretty much anything that you really need to do within it. There will always be new programmes, new languages coming to market, obviously, there are people that are trying to come with no code platforms to make programming more accessible to technologically illiterate people. So there’s lots of different options. But I think for me,

18:23
creating solid foundation and one core programming language and creating

18:31
a market that is encouraging more people to start utilising it, even if it’s a small end of the spectrum, even if you’re not a data analytics person, I think is going to become important. And I think, personally is my belief, then next five to 10 years, we’ll see a programming language probably pricing becoming an option for children alongside

18:53
a, an actual spoken language. Yeah, and I know, we’re gonna chat a bit about the Code Academy that you’ve got, as well, which sort of helps, I think, disadvantaged children around like helping them code. And in particular, particularly in Python, I think it is Wait, you’re kind of at the moment, but I suppose just generally, in terms of making sure you got coding skills, because it’s going to be important in the future, right. Yeah. 100%. For us, there’s a lot there’s a lot around data literacy. At that larger scale and organisations, businesses are starting to understand that they need non data professionals to be data literate, whether that’s including programming element or not. But I think there’s a massive cultural shift towards

19:35
entire organisations becoming a lot more data proficient. So that data driven decision making is adopted a lot more. I think, one of the biggest challenges a data and analytics professional usually has in their career is that handover between themselves and operations and a lot of that comes down to a lack of data literacy, to understand what is it that

20:00
The data analytics professionals are doing. So creating that literacy at the starting point is going to really help these businesses and transform. I was quite surprised as you’re looking through job boards the other week, and just like, how much it’s changed since since I was doing out like lasting even I think it’s probably like five years ago, in terms of requirements for the job, and how many more are now asking for, for coding skills as part of some of the roles so it’s becoming evermore prevalent. And he’s used to seeing it, see it change over time, where it’s becoming part of the roles. So it’s the sofa, particularly for the young kids now going through, they’ve got to understand they got to understand how to do it to be really competitive for the jobs, that thing? Yeah, yeah, absolutely. Yeah. 100% agree with that. And I suppose just on the data side of things, going back to the top of the conversation is like, is we can all think of it in terms of what a process looks like. And we do process mapping, but it’s almost like how do you do data mapping? And I’m thinking in terms of data, because that allows you to then do the data decisioning, doesn’t it? So it gets I suppose it’s just trying to change that thinking round a little bit? And have you found the Code Academy in terms like rollout and adoption? Yeah, it’s been really popular with, with all the schools that we initially engaged with, I think the initial surprise was the fact that they’re wondering what the where the cost was going to come into it. But because it’s a free programme that quickly came on board. And it’s been phenomenal to see how much interest we’ve had the first call that we’re in, when we rolled out third cohort, we had 67 applicants for 2015 spaces. So it’s really it’s not just the children, it’s the parents as well that are starting to take wind of it, encourage their children to go and apply for these courses. So I really think it’s going to be an important part of curriculum. Yeah, maybe just give a brief, how many schools you in now, how long has it been going? What’s the I suppose? What’s the scope of it where we’ve been so far? Yes, I guess to take it a step back even further, just to the reason I started of the Code Academy was, there’s three reasons in fact, two that are commonly talked about, and one that I believe isn’t, as commonly talked about, one, there’s a data skills gap, as government released paper back in 2021, articulating how many jobs they believed were required in the market, how many jobs will be vacant, because there aren’t enough professionals in the market, to there’s lack of diversity, predominantly gender diversity, but also ethnic and disability diversity and within roles. And then the third one, which I feel isn’t talked about is that I believe that there’s a scenario where data science and analytical roles can become pay to win. And by that I mean that because the salaries are booming, because there’s so much interest in those roles. And because it’s only really a skill set that’s required, that skill set can effectively be taught and learn. And you can put yourself on courses and hearing about phenomenal amounts of money being spent by people going on data analytics courses. And ultimately, that can give you that that advantage. But obviously, if you don’t come from, if you don’t have the means to put yourself through courses be put on courses, then that can give you disadvantage in that environment. So that was the core principles behind the date and amount of Code Academy. So we now are in schools, in we’re in five schools in Manchester, focusing on cohorts of 15 children,

23:36
at least half needs to be Pupil Premium identified, which is government classified disadvantaged,

23:42
that has scaled from one school back in September. So we’ve, we’ve done our initial phase of scaling, we’ve got partnership with the University of Manchester. And the intent now is to really push it and scale beyond the five, we’ve got an aspiration of 20 over the next year. And then beyond that our vision is or at least step one of our vision is 1000, disadvantaged children taught Python each year, across Greater Manchester. And if we can achieve that, then beyond that, it becomes national and hopefully global, it’s fantastic. And as you can see the changing requirements of the workforce, and how we need to think about things differently, as it certainly plays into all that in terms of the future for that for the future generations. Right. I think that’s it, I think the only way forward with businesses is towards data and analytics towards technical proficiency. And if there’s currently a gap in skill set now then that’s only going to become resolved. If you start really early. We need to blood in the next generation and really get them interested. And I think some of the classes that I’ve been to it’s interesting to see the shift in children’s mindsets now when we first

25:00
just rolled out the programme a year and a half ago, just a couple of years ago, when I asked the children, what do they want to do if they wanted to go into use technology? Why were they learning Python while learning programming? Most of the children said computer games, they wanted to be against design. And that’s fair enough. And as a 1415 year old, I’m not surprised by that answer. Fast forward to a year, the last six months, asked that same question to a different cohort, same age group. And now they’re saying, web design, the same software engineering, this includes data science, and it’s phenomenal how much awareness there is of the opportunities available for that, and the breadth of opportunities. And the purpose behind going into programming is very different. And that’s over the course of a year, let alone anything further. So I think it’s really positive for for, for the economy, for the local community, as well. It’s interesting how things change. And so is one of my regrets is that back in the late 80s, I didn’t go into computer game design, because I think that would have that was probably getting the timing right. But but looking forward to the next sort of five years out from here, what are the areas you think are going to be important? And what do you where do you think we develop from here, because it’s also about the things that are smaller now will be the things will be much bigger in five years time, rather than the things that are big now, and we’re talking about large language models. Now everyone’s talking about that. But it’s what are the things that you think are going to be important? And we’ve got to learn today, in order to get to in order to get to the future? Where are we going? I think, to a certain extent, for me, what’s important is having the accessibility of technology becoming more available for a broader audience. So really,

26:53
ignoring some of the more advanced elements, that’s quite obviously, there’s a lot of excitement there. But being in a position where we’re not saying that 90% of businesses are data analytics literate, we’re saying that 90% of businesses are data and analytics literate, I think creating that base level of understanding and that and

27:17
when businesses set up when individual businesses, you know, what’s important, we need to get an accountant, we probably need a lawyer at some point needs from HR professional as you’re going along. I think having the data professional, as one of the early adopters within an organisation is something that needs to become a common theme. So I think really getting that foundation established, so that it’s common practice for data and analytics to be a necessity within business rather than a luxury for after down the line. And then I can’t predict the next 612 months in terms of the analytical advancements, let alone five years who would have thought a year ago, we would be talking about Chuck GPT now and generative AI and it’s coming along phenomenally, just in the space of a year, I think what’s going to be important, and I guess, probably too boring answers to a really interesting question. But for me, I think what’s going to be important is really getting the governance right around analytics, and really making sure that we do it right. There’s a lot of fear around AI and fear, what are the consequences of it? Obviously, there’s that paper that was signed by all the large organisations six months ago. So we need to pause until it’s felt. But

28:38
I think that’s, I understand why that happened. And there’s probably some commercial element in hindsight, probably, they will have some commercial element in their consideration. But I think rather than pausing advancement, we need to embrace it, but understand how we embrace it in the right way. So I think there’s going to be a real importance with between developing policy, governance and structure to and bring that into the regulatory environment and having robust regulation around the use of analytics. And I think that’s where we need to get to in the next two to five years naval advancement. You can see the bunfight starting about data can’t even this week over who’s got access to what data there’s lawsuits going left and centre and so like, it’s again, it all comes down to the data really doesn’t. Who owns what versus not so it’s gonna be interesting to watch that, that that kind of play out. And when you’re talking about the chief data officer, what I one thing I had, in my mind is my photo collection. It was like, we all have these photo collections on our phones, where it’s like, everything’s everywhere. And honestly, if only at the start of it, I’d actually organised it and I knew where everything was, and I’ve done a little bit by little bit. It will be just fantastic. I could find everything and it’s almost like that was almost like my analogy for the data office. So if only my data was structured in the right way, and I knew where everything was and I can actually use it, then I’d be it’d be transferred

30:00
bought it exactly that benefit of hindsight. I think there’s a character called Captain hindsight within that’s probably the right sort of roll for the name for the CDO.

30:13
Hey, man, thanks very much for making the time as it’s fascinating to hear about what’s going on in the in the data world. And that says decision science and the data science elements of it. Because because often it’s we hear a lot of media talk around at some of these exciting things. Like that was like the magic of the end product. But actually, it’s the, as you say, it’s the data that governments are making sure you get it right, actually, underneath the hood is important because that’s where it’s really going on. So I really appreciate the insight. Now, it’s great to talk to you, Chris. Thanks very much. Thanks a lot.