Privacy Policy

Episode 33 of The Andy Show

Episode 33 of The Andy Show

ANDREW MCLEAN [00:00:35] Good afternoon and welcome to The Andy Show. It is Thurs- is it Thursday? Yes. Thursday, the 4th of June 2020. The time right now is. Well, it’s actually 12:30 on the dot. Now, I have, I’ve been reading about this company and I’m very excited by it. Intelligence has been a very interesting topic analytics, especially at the moment with this lockdown, how people are going to continue and what data can we use in order to best implement that. So it’s my absolute pleasure to be joined by the Chief boffin, the Chief Technology Officer, the Co-Founder of British company Panintelligence, Ken Miller. Ken, welcome.

KEN MILLER [00:01:24] Hi, Andy. How are you doing?

ANDREW MCLEAN [00:01:26] Yeah, I’m not to bad. So you’re obviously a very smart man. So you want to start by telling us a little bit about yourself and what about Panintelligence?

KEN MILLER [00:01:35] Yeah. So Panintelligence as a business, we’ve been going for about 5 years now. We’ve got a team about 45 really dedicated, you know, data geeks, I guess would be a polite way of putting it. But, you know, we are with the people who you show us a spreadsheet to me and we actually get excited. We grew out of another business called Pancredit Systems who worked in the lending and leasing world. And actually, I’ve been doing this quite a while, as you can probably tell by my grey hair, which is getting very long and in lockdown. But, you know, what we try to do is we try to make business intelligence and analytics accessible to all. And what we found is even going back to 10, 15 years, there were lots of tools that people could use, but you had to be an expert to use them. So see, quite often becomes the bottleneck to business. So everyone wants to get a lot of things done, but they have to go to 1 or 2 key individuals to get those things done. And actually, what you find is people find sometimes quite nasty ways around that as well. So if they can’t get that work done, they’ll go and find some way of shoehorning things together. And that’s when you start to get problems. And also it’s when you start to find that your marketing director, who’s a very good marketing director, spends 90% of their time in Excel doing Excel spreadsheets. Thats not a good use of their time, they are smarter than that. So we started to build a product that became Panintelligence. So it was kind of an accidental journey at first. We didn’t really realise, we would started to build a business, but we were quite quickly realised that actually we could work in areas not just in finance. So as we spun out to the business, we did an MBO about 5 years ago. We started to look for other verticals and we’ve been really successful. Obviously, in finance but now in education, in retail, in manufacturing, we have pretty much every marks of markets that you can think about. They all have data. You know, everyone has data. And we have a system which allows real people to crunch the data and extract meaning from it. And that, you know, that’s really the core principle of what we do. We have a strupp line which is enabling better decisions. And really, that’s what data’s for. You know,if you can’t enable decisions with data, then, you know, what’s the point of collecting it in the first place? So that’s what we do, really.

ANDREW MCLEAN [00:03:56] Okay. Excellent. So I would ask you the question. I would ask anybody that’s in the technology sector. So when we speak to people from A.I., there’s obviously tons of A.I.. But we always ask what benefit or what impact does that have on businesses? So let’s look at business intelligence. If I were to phone you up, say this sounds really cool. What impact would business intelligence have on a business?

KEN MILLER [00:04:22] Yeah, you’ve stumbled into a fantastic area of there are you know what what actually is it A.I Machine Learning. You know, all of these things become a bit of a marketing buzz word. One of the things that we’re really careful about is we try and debunk some of the myths. And the best way we do that really is by telling stories or explaining and case studies of how we actually use and how we implement that. So I’ll give you an example in education, we work with our partners called Tribal, and they run the software so 6 halls, universities in the U.K. and we looked at a problem that we wanted to solve for them. So one example of this is we looked at retention of 6 former students. Now, for the colleges, it’s quite an expensive process to onboard those 6 former students. And actually it’s very common for students not to make it very far in those courses. So a lot of them will drop out at sort of 6 weak points. So what we were doing was actually building models and we took the same models that we used in the world of finance. Where we’d look at underwriting alone so who’s going to repay a loan and who isn’t? We used exactly the same sort of modelling, to look at who was going to be retained and he was not going to be retained as a student. Now, what that quickly allows us to do is identify the characteristics that are important in that. Now because we’re not using really complex tools like Python and our programming tools, really. We using tools that end users can actually use and get benefit from themselves. They suddenly bring into that the domain expertise. So when you’re talking about, you know, distance from college or you’re talking about number of clubs, they attend, these characteristics that we know are significant in terms of, you know, the outcome of the students. Actually, they’ve got all this wealth of knowledge. So they hold all this information around that, the context of it. And that’s actually what’s quite often missing from the A.I. ML builf journey, is the context and the meaning. And these are the people who you have that in spades. So we bring them into the conversation and we show them tangibly what they can do with this. And actually, A.I. Machine Learning is a journey. No one is going to go into a system, switch it and say okay, we’re going to let the bot work all this out for us now. It’s not going to happen. It has to be a progression to that. I promise a lot of people I don’t talk about Teslas very often, but every now and then. And it’s the same with the self-driving cars. You know, so no one is going to switch cars over to self-drive tomorrow, you know, it just can’t happen. But it’s a gradual step by step process where when machinery or decisioning can do more and more for us. But in the first instance, it tells us something that we may not have known. Or validate something that we do know. So in that education market, a really simple example is when we noticed students who were travelling more than 10 miles to college were very likely to drop out. Now, this was at a college that we were using the sample college, “sorry”. And instantly those people were the domain expertise said are but it will depend whether they getting the bus or the train. The train is a really quick, simple and consistent travel. Whereas actually the traffic on a bus might be really inconsistent. And they showed us that actually by segregating that characteristic really quickly. We were able to bring a lot more power to the model itself and therefore the outcomes that we’re able to predict. And that and that’s what’s really important, is that self-service bringing people in and not having to have a couple of experts sitting in a black room removed, far removed from the decisions. But bringing the people almost on the front line into those decisions. That means that actually we have far greater success in implementing the Machine Learning algorithms that we actually build. And that’s the biggest problem in the world of A.I. in Machine Learning is there’s lots of really clever stuff out there but the adoption is still really, really low. And that’s because they don’t get the right people on board and they don’t get the right level of trust and buy it. Whereas actually if you bring those people in from the beginning you do.

ANDREW MCLEAN [00:08:32] That’s uhm… I mean, that picked my interest because my background before this used to be psychology and it’s a very interesting, there’s a thing in psychology called, third person effect and with the third person effect, if this happens, the other people will do this. You ask the same person, will you do this? They’ll go, no. So I would ask you the obvious question, because you mentioned that there regarding the buses. When you actually implement datat, does it often complement or contradict assumptions of human beings have made about business decisions?

KEN MILLER [00:09:07] It will do, they all do one of both of those things. Now, cards on the table, I studied psychology as well. So how we both accidentally got to where we are today, you know.

ANDREW MCLEAN [00:09:22] A long journey, yes

KEN MILLER [00:09:24] Yes, exactly. But yeah. So when we’re doing this, one or two things will happen. So either the ring fall silent when we show you the first analytics to people. And that’s usually because you’ve shown them something that they hadn’t considered before. Or you start a fairly robust arguments. And that’s usually when you’ve touched a nerve because it’s something that someone new and you’ll get someone in the corner of the room saying, you know, I’ve been saying this for years, but they haven’t had the empirical evidence to start to engage in a conversation about it. And very quickly, you get the yes, but, this, and yes but this and actually what they’re doing is they’re giving you new characteristics that you could analyse that no one had written on the bit of paper at the beginning of the exercise. And it is why it’s such a recursive journey. So you build a model and the first model is rubbish because it tells you a little bit, but not much. But actually what it does is it eeks that knowledge out of those people. The stuff that’s in the head, but they don’t want on paper. So they’ll say, and the classic is example of the education, they starting to tell us it’s whether they come on a bus or a train. We didn’t have a clue about that. It happened at the college we’re looking at a train station 2 minutes away and the bus journeys were horrific. Now they knew that, but they hadn’t documented that anywhere. And actually also at that same time, they they spent quite a lot of time capturing information about parental log ons to the parental portal. See if, you know, if it’s parental engagement that’s significant. And we really quickly found out it wasn’t, actually that challenged one of the myths that they did hold. But again, once you’ve challenged it, it’s kind of it’s not telling you why. Those are the answers that have to come from the domain experts. But they quite quickly through conversation with each of us to start to drive things. So you kind of, as a data analyst, sit there with a bit of paper writing as fast as you can, because everything they’re saying is like gold. Because they’re saying yes but we know the reason that student did well is we called their parents in for this kind of meeting and this, that’s not been documented. Okay, so how many other children have had that same interaction that could be effective? And, you know, all the while, you’re building up this model which allows people to make really, really smart decisions. And it’s kind of there’s a phrase called the Kobol that we came across a few years ago and really, really like it. Which is actually the first thing is the model is kind of like that person on your shoulder getting a little bit more information. It’s not doing a job for you, but it’s just helping nudge you and, you know, nudge you in the right direction. So, yeah, it’s all about psychology. Everything’s about psychology.

ANDREW MCLEAN [00:12:04] Well, I’ve got to be briefly dip into this then, because that’s kind of weird. But a little bit like a linear regression that you would do is that if you- how does your system differentiate between something causing an effect on a business and a correlation that just happens to be A + B = C…

KEN MILLER [00:12:28] Yeah, I so. So there’s no mathematical determinant to to decide what’s cause and effect. You know, the classic example in maths, if you look at correlation between things, speed and engine size are correlated. Now as a human, we understand quite clearly that if you increase the capacity of your engine, it will go faster. But actually, if you look at it from a mathematical point of view, making the car goes faster, increases the capacity of the engine. So actually that’s where you do need that domain knowledge to be able to understand that. Now we all know that the sun doesn’t rise because the rooster crows. Actually, we say we all know that if you go back to 3000 years, the civilisations that didn’t quite have the same beliefs, you know, they made sure that they kept the rooster “lesson”, the sun would rise. So, yeah, there is no mathematical way of determining that. That’s what you need from the domain experts. To get really when you say this is likely to happen. And actually, all the maths we use is linear regression. That’s that’s what we do. And we have a little bit of Fridays which is, you know, calm down, it’s not A.I.

ANDREW MCLEAN [00:13:44] I love it. All right, let’s take it over the boffin theoritical, let’s go back to the practical world. As much as I hate to, I could talk about this all day, but let’s go back to the practical world where we’re currently on lockdown. There’s a lot companies talking about what they’re going to do after lockdown. There’s things, for example, like social distancing and all sorts of measures that have to be in place. Business intelligence, how can that help companies pivot, especially during this kind of time?

KEN MILLER [00:14:14] Yeah. And there’s been phases for us as we’ve gone into this. So we got very, very busy as lockdown was approaching, as lockdown started. Because I think despite the fact that we all had early warnings from what was going on in China. I think to be fair, we were all sideswiped by this as businesses, as well. So a lot of it was scenario planning. So one of the one of things you can do in our tool is a thing called What-If Analysis, and that’s where you can feed in different parameters. So you know, so if the furlough period goes on at the time we were talking about maybe 2 months, will it be 3 months? Now we’re pretty sure it goes to October. But at the time, we didn’t know. So, you know, by squeezing different parameters. Now what does that mean in terms of cash flow? If we reduce marketing budgets by X, what will that do to our sales funnels? What will that do to long term? So actually, it’s the short term dip still worth it. So those are the sorts things that we worked very closely with a lot of our partners and businesses. And actually, the first thing we did was we did that for ourselves. You know, we sat down, bill, ourselves. What-If Analysis calculated to look exactly how the business was going. So that was kind of like the first few weeks of the furlaud. The second sort of phase was it was it was genuinely looking at how engaged staff members were. So a lot of businesses suddenly they got through the first phase, they got the technology working. They, you know, they started to get the finance teams slightly happier. My FD uses the phrase cautiously optimistic all the time which it’s a pretty hefty phrase. So we got them to the point where they were cautiously optimistic or had a plan of what they were going to do. But then people started to think about, you know, staff’s well-being. How well are they still connected in with the business? And that’s where, again, we started to use the software to monitor people’s engagement levels very carefully. You know, how how many phone calls are making, how many emails and rather than just being a big brother from a psychological point of view. It was checking in that actually they weren’t were remaining engaged. And where you could see low engagement from staff members, you could see what and how was going on so you’d have really sensible conversations with people. So that was that was kind of like phase two. And that’s an ongoing phase for us with a lot of our partners at moments. I think where we go to now is people are starting to sort of pivot their businesses and decide, you know, okay, we used to do this. The market’s gone or or not available right now. So what do we do instead? I think I was talking to you earlier when we were chatting about an example in retail where we have partners as an example here, do monitoring, footfall monitoring with cameras to see who’s coming into store, who was going out. And that used to be for marketeers, you know. Is it people aged 25 to 30 coming in? If so, where do they go in the store? So we actually we position goods. They flip that very quickly and we’ve got a couple of partners working in very similar direction here into social distance monitoring. So they’re now measuring any point how many people are in the store? Are they starting to look at things like age distribution as well? So they have a lot of young people in the store and a lot of old people in the store at a particular time. So those different types of groups actually we go forward over the next 6, 9 months. You know, however long this this goes on for. We’ll have to start to look at segregating people. So there is a real “design” that something’s gonna happen. Where actually, you know, young people. I’m gonna to start to mix a bit more and older people will have to look after themselves a bit more, so making sure that stores know that they actually maybe have times for younger populations and times for older populations and how they monitor that. And we’ve got to get past the guy with a high sitting on the door with a check out, you know. Yeah, that was an emergency measure that can’t go on forever and technology is the answer to that. So looking at how our partners can pivot and actually because we’re linkedin in with them. They can do it without writing very much code at all. So they can literally create a dashboard which has the new information in there. And they were able to go to market and test ideas in hours, if more, you know, a couple of days. And that’s the speed that things are having to move right now.

ANDREW MCLEAN [00:18:29] That that’s really interesting, because I was going to bring in my next question, which was always the thorny issue around summer at historically around analytics, around intelligence was people who already have established systems. And it’s how I think you just mentioned that to deal with that no code or low code. How can people can integrate them with the current systems they already have?

KEN MILLER [00:18:55] I mean, low code, no code means a million things. I mean, let’s be honest. It’s the new new big data, it’s the new big puzzle? I think a lot of people will have to work out not just what the definitions of no code, low code are, but what they mean to them. Now, what it means to us is our partners are able to create demonstratable products, very very quickly that they can test the market because everything’s been turned on its head, you know. And it’s not new, we saw this, you know, back in the in the millennium bug era. You know, we saw this in the last recession. Things do turn on the heads and it’s the people who can pivot and pivot quickly. Who will seek out and find the new markets opportunities. But they have to do that very, very quickly. And what they can’t lock into is a traditional dev cycle for that. So what’s becoming more important, especially in the data days, sort of Cloud services in Amazon, in Azure, in Google on those sorts of platforms. They’re effectively a big sweetshop of products that you can start to bolt together. And we’re just a part of that. And, you know, quite an important part of that. But we are part of that. And actually more and more people are starting to build. You know, I wouldn’t call them products, but they are actually building solutions out of things that are already available rather than going back, it is a case of, you know, there’s a phrase for everything isn’t it? The case is not reinventing the wheel, if this exists here than actually your speed to market can be reduced by using that bit rather than trying to do it. And also, you tend to find that it allows people to focus on stuff. So really, where we’re talking about the no code, low code at the moment, it feels more like we’re still having the discussion about build versus buy. So your dev team wanna build everything from the ground up, you know, as what developers want to do, that’s always. Having been a developer for many years as well. I understand that, too. But, you know, if you want to move quickly, then actually you’ve got to look strategically at what you build versus what how you partner and what you bring in and what you use. You look at that that shelf of tools that readily available and pick them up and put them together. And that’s when tools which interface with each other nice and easily become really important.

ANDREW MCLEAN [00:21:17] I remember reading about I think it was Facebook many, many years ago. And as Zuckerberg had this thing where he said almost testing things in the wild so he would get, as developers develop something put out if it weren’t brilliant, if it didn’t work. And I think that’s what we’re seeing here around fast deployment, faster release, get it out. We can tweak it. We can… What were your thoughts on that? What are your thoughts on the pros and cons, particularly around things like longevity of a particular app or something like that?

KEN MILLER [00:21:47] Yeah. So app development is always a journey. So there’s the phrase minimum viable product. You know, one third is bounded around. Sometimes it’s good and sometimes it’s bad, you know? But, you know, you’ve got to get things into people’s hands to really test them and understand them. So, you know, let’s throw it back to the psychology here. You can’t answer these questions for people and how they use them. And actually, I think people are prepared, certainly at times like this, which is why I think, you know, new enterprises are going to be the winner through this COVID period. A lot of people looking for technical solutions. They don’t necessarily know what they’re looking for, but they are willing to give things a try. You know, we’re seeing that in retail. People are willing to try a bit of software out in the wild to see if it improves and if it makes a difference then the adoption will be incredibly quick. So, you know, people need to really think about how they’re able to scale these things up again and actually again. If you’re taking things that belong to a community like AWS or Azure then, you know, scalability is built into those platforms themselves. You’d have to worry about that. So if you’ve got an idea that takes off, it will be move as fast as you possibly can move. And that might be really, really quick of its software adoption. You know, if you trial with, you know, a supermarket and you’re trialling in 2 or 3 sites and they decide that they’re gonna roll out worldwide, you know, you can go from from nought to a million very, very, very quickly. And, you know, that’s it’s not just, you know, getting the prototype ready, it’s been up to scale up, and scale up elastically as those things go. And I think people are more willing these days to to sort of try those things, see if it works. If it does work, then start to throw some weight behind it.

ANDREW MCLEAN [00:23:36] Let’s go to the let’s go back to the core here, Panintelligence themselves, the stuff you’ve been working on during this- I heard my researchers gave me some stories that you’d gathered information and data useful for the actual COVID crisis. Can you tell us anything about that?

KEN MILLER [00:23:57] Yeah. So we’re still monitoring it. And at the beginning of the process, I think I was actually sitting at home and my sister-in-law, she’s a nursing practitioner, she runs quite a few of the nursing trusts up in the Northeast. And we were looking at provision of beds, you know what was going to happen? And actually, it was the point where down in London has been phases of the virus and in the U.K. as well. So, yeah, the Northeast was a couple of weeks behind London. So taking a look at what happened with intense care beds, but also looking at some of our other partners like Rota Gate and people like that, who actually scheduling off people. So it was about, you know, is this going to scale? What would it look like? What’s the provision at the moment for beds? And also, you know what the implications of that on hospitals, because they’re going to have to create lockdown wards that were COVID wards. So, you know, they had to work in isolation as well. And just trying to sort of pull that data together now. It was interesting in terms of what data was available and what data wasn’t available. And really, in a way, that was a sort of throwing a hat into the ring, thinking, now, how can we help here? There’s a lot of businesses, obviously, at the beginning of this wanting to to see if we could help in any way. So the data was interesting. We now we keep monitoring that as we’re going on. So we’re watching some of the COVID data as we go through and seeing how we compare. And we’ve kind of built a model which compares us to Italy. We sort of lag Italy now by about 2 weeks. As we go through this and just comparing certainly as we come out of lockdown now. That’s going to be really, really interesting in terms of whether that’s effective or not. So we continue to monitor those things on a daily basis. I think as the COVID period , some moved on, we obviously have to turn things into things that were commercially viable as well as, you know, of interest. And that’s where we sort of pivoted with our partners more than anything. And the sociall distancing became a bigger project for us, from that point of view. We still keep track of it. We were also, we’ve got a dashboard internally that keeps track of all of our employees, how much time they’re saving, not driving into work. What their carbon emissions are? You, know, I’m a bit of a carbon freak. So, you know, I try to get them all to drive electric cars as much as I possibly can. But, you know, we’ve put together a dashboard looking at carbon and fuel savings for everyone in the business as well. So there’s lots you can do. You know, data’s always fascinating, really.

ANDREW MCLEAN [00:26:45] Certainly is, I’d love to talk to you again about this in a few months and see what your results are. I guess, it’s absolutely fascinating, particularly looking at these things and in terms of data. I’m running ouf time, unfortunately. I’m gonna ask you one other thing around Panintelligence, that has to do with human decision making. Go back to the psychology. Again, my research has taught me story about human centred A.I. something to do with tennis. Could you elaborate?

KEN MILLER [00:27:15] So this is whay we talk about the domain experts and the adoption of Machine Learning. In that, you know, I think, quote it is that only 20% of Machine Learning practises actually get implemented in the world. So 80% of things fall by the wayside. And for me, it’s all about trust. I’ve got a bit of a presentation. I won’t give it to you now, but this is about, you know, where technology is words and how it’s built the trust of the people. So I’ll talk about Hawkeye in tennis. You know, the video representations for me in Hawkeye where they brought the participants into it. So the players, the crowd, the umpires were brought into that decision where you saw the animation of the ball go across. So my kids are a little bit older now, but when they were younger, if they were playing badminton in the back garden and they would be complaining, you know, you play that fabulous game where there’s no lines on the turf outside. So they’re trying to decide who is in and who was out. And it’s fairly arbitrary to them because there are no lines. But I’d get a little shuttlecock and replay that the whole guy moment of it going du du du du du du du and is it in or out? And actually, it is where you find that Machine Learning can bring in a certain degree of fairness, you know. So, all human decisions have a bias in them. And you know, where we talk about bias in A.I. usually is replicating the bias of the humans who’ve led it, the data ringed to, to feed into the models they’ve created the A.I. decision. Whereas actually, if you can make technology fair and I did a fabulous… I grew up with Max Roach shouting the umpires, that was wonderful, and we missed a little bit, to be completely honest. But who can you argue with when the machine has made the decision? And as long as you feel that the machine is making a fair decision and actually it can better, but it’s all about trust. And I do drive a Tesla, and Tesla’s exactly the same. It’s got the screen on the side of it and it’s showing you how it’s making decisions and it makes some terrible decisions and you do have to grab this and it will still at moments, but it will get there. You know, when you do, you don’t need to be over. And that’s what we talk about, a human centred A.I. is A.I. you’ll only be adopted if people trust it. But it also needs to be monitored. And that’s where big, big, big challenge for BMI comes in, is making sure that you’re monitoring, so people stop doing the rudimentary jobs, then start to monitor the machines that are doing those jobs. But unless then if they’re not able to look into that window and find out what’s going on, then, you know, again, the trust is broken down or it does start to run awry and make some really bad decisions. So it’s all about gaining trust, you know, bringing the domain experts. Use their skills, you know, the whole cobot let it sit alongside humans. Nowhere in history of a sort of mechanising things have humans ever been replaced. They’ve become the supervisors of the machines. And actually, that’s the role the A.I. has to play in Machine Learning algorithms that are implemented. Someone needs to be able to sit there and sample the data. And effectively it comes down to it, being a simple as was that a good decision or not? And if you look at 100 sample decisions and you look at 99 of them, they were good decisions, ones that you would have made. Then you’re happy. If you look at it and it’s dropped to 75% of those decisions feel like good decisions. And actually, you know it’s time to intervene and look at the model again.

ANDREW MCLEAN [00:30:50] It’s genuinely fascinating stuff, Ken. Okay Ken, so thank you for that, the interview itself is over. But I’m going to just take a quick peek behind the curtain to find out the background of Mr. Ken Miller. So I’m gonna ask you a couple of questions. Number one, what’s going on in the roof there?

KEN MILLER [00:31:09] So I’m a bit of a restless individual. And what I found when I came home was I normally have a 3 hour commute to and from Leeds from where I live. And I didn’t have that anymore. So I was working a lot more than I should. And actually, that was okay for me. But I was putting pressure on others, asking them questions out of hours. So I hadn’t made an airfix model for a very long time. I had one sitting on the shelf. I made an ethics model and I kind of kept going. So this was about 27 on the ceiling. You can see a few of them was even more. So I just let’s keep building airfix models at the minute. So that’s just to stop me annoying people.

ANDREW MCLEAN [00:31:49] I love it. I love it. And I also hear a mention of a robot.

KEN MILLER [00:31:53] Yeah. So I’ve got various robots around the corner here. So this was, I got told by H.R. to bring this home. This was a social distance monitoring. Before it was, I got inspired by this. When we had a fire alarm call and we all went outside and they went through the list than I was. Well, no one’s got any clue who who’s here and who’s not here. So I had to use a Raspberry Pi in there with an arm, no body on a robot. And it walked around the office and it used facial recognition. Just use an open CV. So stuff I’d hacked in and pulled down from the Internet, recognised it so effectively, took a registry. But it does look a bit like a nasty spider when it crawled around.

ANDREW MCLEAN [00:32:32] Yeah. So that might look great on film from the 80s with the spiders attacked, that was great.

KEN MILLER [00:32:38] Also so Ho aren’t you happy about me recognising people in the office…

ANDREW MCLEAN [00:32:43] Ohh, there are no fun. There are no fun at all.

KEN MILLER [00:32:46] And I’ve got another one. It’s a whiteboard robot, some guys up in Scotland called JJ Robots. A four of them, it’s hacked in. It uses our scheduler. So Panintelligence rolls runs over. It looks at support tickets. If a support ticket is come in and no one’s touched it for 30 minutes during working hours, it writes on my whiteboard here. And, you know, I need to to make sure we’re looking after the customers a bit better so…

ANDREW MCLEAN [00:33:12] The machine that grassing us up now Fantastic. That is fantastic. Ken, unfortunately we have run out of time. Thank you so much for your insights. Thanks so much for explaining about business intelligence and the work you’ve been doing around just in the past, the present and what you’re looking at be doing in the future. And thank you for juicing us the robots. It’s a… I’ll keep my tongue on that. That’s fantastic. Thank you so much, Ken.

KEN MILLER [00:33:41] Brilliant. Thank you, Andrew.

ANDREW MCLEAN [00:33:42] Thank you. That was Ken Miller. The Chief Technology Officer and Co-Founder of Panintelligence talking to us about business intelligence and analytics. No code, and things such to be doing around social distancing in the retail sector. So much, so much interesting things, particularly when you tie in the human psychology and technology together. Fascinating things. And we’ll be keeping an eye on Panintelligence for the next few months to see how they’re getting on. But the time now is coming up to 4 minutes past 1. You have been watching The Andy Show on Thursday, the 4th June 2020, and until tomorrow. I’ll see you soon.