Episode 2 Transcript

00:00:10:07 - 00:00:41:17

Callum: Welcome back to Tech will Save us Podcast. We're lucky enough this week to be joined by owner and founder of Ontaro, Tony Paskin and the ever present Head of Data and Strategy, Patrick Murray. 

Patrick: Hello. 

Callum: Now let's start with you then, Tony. Can you provide a brief sort of introduction to Ontaro? What does it do? 

Tony: Yes, Ontaro was developed, which is an online child protection to help parents monitor their children's online activity through their mobile device.

00:00:41:18 - 00:01:12:22

Callum: Excellent. So how does it work and what's it, what's it actually doing?

Tony: Well, we've used AI to actually monitor the language in the conversation of what children are having online. We're using apps and well over a certain amount of categories, it will alert a parent of any dangers that are apparent and so they then can talk to their child about it.

00:01:13:00 - 00:01:45:16

Callum: So it's really a conversation starter between parent and child, it enables that conversation. 

Tony: That's right. Yes, it is. Yes. 

Callum: And what was the motivation behind developing this app? 

Tony: Unfortunately, I had the experience of of someone actually committing suicide. And when we started looking through their social activity and all that online activity, we could see that they'd been self-harming, going on self-harming sites and then ultimately committing suicide. And I just felt we were never in the, in the loop with a child's activity and life. They were so busy and as an adult we’re so busy, that we needed some kind of early warning system and so Ontaro was born from that really. 

Callum: Fantastic. Obviously the hope that this is a preventative measure for future things such as that then?

00:02:11:01 - 00:02:34:21

Tony: I think Ontaro’s not a substitute for openness and communication with the parent. It's there to encourage that, you're not spying on the child. You tell the child you’re putting it on their phone. I mean, a lot of parents give the child the phone and tell them they want to take that phone back off them and check it and look through it. Well, as a child of 13, that's a breach of their privacy rights because a child of 13 has the same privacy rights as an adult. And so you're not encouraging the child, you're pushing the child away. Whereas Ontaro, you put this on the child's phone and you can tell the child, I don't need your phone. So, so it creates an openness and brings you closer and encourages you to talk to your child.

00:03:02:17 - 00:03:37:13

Callum: So it’s striking a balance between protecting the child online and respecting their privacy rights as well. 

Tony: That's right. Yes. Yes. 

Callum: Very good. So in your opinion, what is the role of education in protecting children online? 

Tony: Well, I think education is the point. They've got the access to children with schools to educate them on everything that is about online and the dangers. I think the online bill is is trying to encourage things, but I don't know if it goes far enough. It's just information and knowledge and children are so apt at learning, they are programmed to learn. So we just need some time to encourage them. I know they're trying to encourage schools a curriculum about suicide and mental health, which is good, but I think it needs to go further. Children today, we've never been in such a position. We learn from our parents, but this has never been, you know, social media and the online activity and everything is so big. It's a first. 

Patrick: It must be difficult for parents. I mean, we I guess we come from a generation where social media was just starting out when we were in school or when I remember having, you know, a MySpace account, which felt a lot different to the kind of social media you get nowadays. But to some extent, we were sort of, we were still raised using that kind of tech, but I don't think it was quite as pervasive and into every part of our life as maybe it is nowadays. And so one of the things I like about learning about Ontaro is a lot of parents who didn't have, you know, this is quite a new world to them as well. And they maybe don't even know, even if they did have that insight themselves, would they know what to look out for? Do they know what kind of sites their kids can go on, what kind of things they can access here by having an Instagram account and, you know, tools like Ontaro and AI tools in general. You know, it's almost a way of bottling that experience and putting it in an app. And, you know, not only are you not prying, but you can be confident that there's something there that does know. 

Tony: Yeah, I think that can help. Sort of. You're right. The physical, the physical dangers as a parent, was your child going outside and being exposed to the world or to the locality of a predator or some nasty person or something. But what we've done is we've given them exposure and access to the world in their pocket. Yeah. And there's, there's all those people, all the negative and nasty sides of it. The child may not see or be exposed because it's like, uh, I think it's the word anonymity, people don't, don't take responsibility for their actions. And it's scary. The online is scary. And parents, because we just can't leave it for the government and people to take action to protect. You've got to do it yourself. And then this is one thing that Ontaro can give parents a tool to do that. 

Callum: Fantastic. But just going back to something that you mentioned there, Patrick and I know it's something that the viewers will want to know. MySpace account, is that still active or not?

Patrick: Well, I actually went looking for it recently. I want, I wanted to kind of relive some of my old hairdos and see what it was like. But, you know, MySpace back then and social media in general. I mean, what was a MySpace account? You messed around with a bit of HTML. You put a song that you thought was really cool on.

00:06:44:12 - 00:07:02:17

Patrick: That was all nice as far as it went. And it feels like now, in the days of Facebook, where it became a lot more photo conscious, everyone was judging everyone. It became about likes. It became about commenting on people. I feel like we saw that, but thankfully, I'm glad Facebook came in a little bit later, I think for us.

00:07:02:19 - 00:07:30:13

Callum: So. Well, so we'll drop the link in the comment section. 

Patrick: Yeah. Yeah, for subscribing to just for yourself. 

Tony: So just stepping in on the social media side of things. It's, it's the difference when, when you're younger and you say you fell over at school or, or did a funny face when you were yawning, the only people that saw it were in your vicinity. Now, somebody catches it on the phone and it's all over Instagram, Twitter, Facebook, and you really can be ridiculed for that. And so you can't, children just can't get away from that, it's not localised anymore. And that's that that's the difference, I think, as well. 

Callum: Yeah. So how can parents and guardians best make use of these functions on the app to increase its effectiveness of safeguarding? 

00:07:56:02 - 00:08:20:21

Tony: Well, Ontaro will, will alert you of any concerns through the different categories that we have and you will, they will  go onto their dashboard…they have a dashboard and it will give them the alerts and it'll give them a snippet of the of the word and what alert and concern and it will give them, um, a percentage or a degree of how serious we think it is. And then they can take the action of whether it is serious or whether it's not. And then they can help us, they can help Ontaro learn and grow as well, by choosing whether it's, it is or isn't serious. We also provide access to charities where they can get help and look, so we don't just leave them, where we give them the work. They know that there's a concern and that's it. We also direct them where they can get advice and help to start conversation with the child. And so that also helps, we don't just leave them in the lurch with that. We encourage them to look at these charity sites and get the help that they need to talk about it.

00:09:01:11 - 00:09:03:11

Callum: And where can people find it and you mentioned it’s on the android?

00:09:03:17 - 00:09:26:15

Tony: Yeah it's,  if you go online, it’s www.ontaro.co.uk, you'll find that there, there's a lot of information that you can read to help you and then you can click on the, um, the link to open it and that will take you to a page where you can subscribe. You get 14 days free trial. Um, so there's no, no cost to try it. Um, well, yeah, it's, it's something that's there to be able to try for free.

00:09:37:20 - 00:10:00:05

Callum: How did Razor help you on your journey then towards building Ontaro?

Tony: Well, from my first idea of what I wanted, it took me quite a long time to even find anybody that could actually understand what I was wanting to achieve. I found Razor on my doorstep virtually. And from my first contact, two weeks later I got such a fantastic, um, reply, a report of what I didn't expect which gave me so much confidence that I’d found the right people there and gave me enough to move forward with them and to the next stage. And so yeah, I feel quite lucky that actually, although it took me a while, lucky to find Razor and the steps that we went through to get where we are today, it's been, I've been lucky I think, to have found Razor, and such a company that has achieved everything we've wanted, we've set out to achieve and gone further really. And when we were talking about AI, I think data is a very important thing and I've learned a lot about that, data mining and scraping and all that, so it's just been fantastic. Razor has really stepped up to the mark and blown it out of the water. Yeah.

00:11:18:08 - 00:11:34:00

Callum: Thank you very much for sharing. What advice would you give any sort of new businesses or current businesses who are looking to adopt AI? 

Tony: I would, I would encourage you to embrace it. It's going to be here. Um, if you can identify that it will help you, then, um, yes, definitely. Um, I think, I think it's, it's, it's improving all the time and, um, yeah, AI, it will be a fantastic aid as long as we, you know, we use it like that and I think, yeah, I think the governments are or you know, realise it and I think there's a lot of scaremongering about how serious it can be and what kind of, I think it can be. It's great that they are looking at, um, controlling it, or looking at how serious it can be because they neglected that when they first provided internet and online. And that's why we're in the position we are with online because it wasn't policed or implemented correctly whereas they are doing it with AI. But AI, I think it’s fantastic. 

Patrick: I think it’s good in a lot of ways how big it is in the news at the minute. I think people are becoming, and it's important for the public and for people to be aware about what is possible with it. Have an awareness that especially with things like generative AI. It helps you create things, with things like images. It’s changing, we saw not that long ago, those pictures are going, when Donald, when those rumours were going round about Donald Trump getting arrested and there are all those pictures going around. It's important that people are aware that you almost can't trust everything like that, you know you need to question it. I think people are becoming more aware of the value of their own data off the back of it. You know a lot of these algorithms that we see, you know, AI has been around for decades, but we're seeing it, you know companies like Netflix, Amazon that we talked about in previous podcasts, using that data to recommend things to you. That's not done by accident, that’s by collecting people's data. I think people are becoming more and more aware about value. It's going to be an interesting, interesting few decades, I think, coming up for that. 

Callum: And so do you think organisations are ready for AI?

00:13:39:12 - 00:14:04:06

Patrick: Yeah, I think it depends how you frame AI, I think um, one of the great things about tools like ChatGPT that we're seeing and, and even with image recognition, there are tools that you see through, through Microsoft, like through their cognitive services, where actually you can get a lot of functionality out of the box. You don't always have to start with just your own, with your own dataset. And when you're using things like language models, there's more and more opportunity for organisations to start from, you know, build on the shoulders of giants, as it were. I think we've seen with data in general to, to do AI for, say, your own, your own purposes to train something from scratch. Generally you need quite a lot of data. And I think it's coming at a point now where companies are more aware of that in general and are likely to be sat on more data  and use it more. But I think I'm in that regard, companies need to walk before they run with AI and focus on like, how can we use data, how can we use and if we get data in the right people's hands to empower them and maybe they AI can come next and make that even better is the next stage. So I think there's a lot of opportunities for organisations to start using AI, but I think starting with the data that enables more and more things. 

Callum: So you talk about walking before you run, is that the logical first step? Then? 

Patrick: Yeah, that's a massive sort of step. I mean, so in the we, we talk about when the services that we offer as is our Edge to Cloud offering I think which which is particularly in manufacturers is where they start collecting data. They may be hung up on keeping that on premise and using that for not letting that leave their site, but I think more and more manufacturers, when we're out there talking to them, are starting to be aware of what if we move this into a secure cloud. It actually looks at a lot of possibilities to use things like AI to look into that next level of what trends can we uncover and how we can build this into the applications that we use to streamline the manufacturing processes. And I think that's a step that a lot of organisations are going to have to take through a lot of this kind of technology. 

Tony: Do you think AI is used as an umbrella…? 

Patrick: So I mean, AI has been around for decades. You know, AI is anything where a computer is seeming to take some intelligence data.

Tony: So it’s data?

00:16:09:04 - 00:16:28:07

Patrick:  Yes, based on some data. As information or data comes in, there appears to be a logical step. And then something happens, and that could be an if statement. AI could be an if statement. But that is about as narrow as you can get for AI. Ask one very particular question taking some data and effects. If this, then that. That is AI.

00:16:28:07 - 00:16:54:11

Patrick: Then you know, that is they are what we're seeing now there is this sort of next level. You know, when we see things like ChatGPT and these huge language models that seem to be doing a lot more, a lot more complicated actions and making much more complicated decisions, but really that is still quite narrow. You know, ChatGPT isn’t going to drive a car and, you know, there are, you know, self-driving car technology that is becoming more and more common. You know, it's possible. You know, Tesla's can pretty much do that now. Yeah. What's holding it up is, is laws, basically. And I think this is something that's going to hold up AI. It's a bit of a blocker at the minute, where AI is almost held to a different standard to humans. Yes. If a self-driving car takes a million journeys and crashes on one of them, yeah, people will say, oh, it's not safe.

00:17:22:03 - 00:17:51:04

Patrick: Self-driving cars are not safe. Humans crash all the time. All the time. And is there an argument to say, you know, is there a threshold that isn't zero when we say actually AI is making us safer? Of course it's got difficulties. When you hear a lot, people talk on social media around self-driving cars of, if it was crashing and you know, there was an old lady over here or a group of kids over here, what would it pick to drive into? Humans aren’t held to that standard. And so I think there's an argument to be had there of when do we say AI is safe enough?  We were having this discussion recently with one of our and one of the products that we're developing around using ChatGPT to extract information from documents. So I used to work in the Nuclear Industry before working at Razor and that is an industry with a lot of documentation… but there's an application with that, where in an environment where there's a huge amount of documentation in the regulatory world, using tools, using language models to interrogate those documents, answer questions. So what becomes going through page after page becomes I'm asking a question.

Patrick: I wouldn't have AI right now just deciding what is safe for nuclear power plant. However, that kind of tool in the hands of someone who already has experience and knows can make them more effective. Yeah. Yeah. I think that's the point that we're out now with AI, is making people more effective through building these tools to help them, not just putting AI out in the driving seat. 

Callum: So it’s not necessarily a replacement for people, their interactions and their work. It's something to enhance their work. 

Patrick: It's something to enhance people. their work. I mean, don't get me wrong, you know, over the next century, more and more things are going to be automated. That's always been the case. You know, when the combine harvester came along that automated and replaced people. And I think as a society, there's a lot of questions that need to be asked about, you know, what does that economy look like in 50 years, 100 years time. But right now where we are is, it’s there to make people better. You can't always replace that human element. It's what can computers do better than humans, putting that into the power of humans doing the stuff that they do better. 

Tony: And so information and technology always when you once you gain that information, so it's empowering. And then I think this makes it more efficient to find that information. In my day it would have been the library. Yeah. Yeah…And now the information is in your hand. And you hit  it on the thing about crashes, you know the car was a great invention but they very quickly learnt that crashes happen and people died and so they started looking at ways to improve it, which was seatbelts, airbags and lane system. And the same with Ontaro, when children go online they will be, we need to make it safer and so really a child goes online that's an online journey well put a seatbelt on it with Ontaro.  

Callum: Brilliant. So how far is this going to go? When we hear things like computer vision. Can you sort of give us a brief summary of what that means?

00:20:51:11 - 00:21:11:15

Patrick: So computer vision is, computers are very, very good at looking for patterns, looking at trends. And in all cases, when you look at an image, yeah, we do that as Humans we’re identifying something. When did that thing occur? Is that thing different to how it normally is. That pattern understanding is a great computer application that's based on computer vision. It can take many different forms. So obviously we do a lot of work in the manufacturing world where, things like identifying defects, if there's a, if someone's produced a, you know, steel component and you're looking for where where does that not, where are the issues that the last 2000 pieces didn't have, that looking for that kind of thing.

00:21:34:06 - 00:22:17:21

Patrick: And we have a lot of conversations with things like the textile industry. Obviously, that's a big industry in Yorkshire and where in a lot of cases when materials coming in through is still very manual or very human focused to look for, okay, where does that, imagine after a six hour shift and you’ve got material coming past you, you’re not seeing those, all those defects. Computer vision and the products that we develop for computer vision, are for identifying those and automatically doing that. I think that's something again, particularly the manufacturing world where an industry where we rely on sensors, I mean just things like temperature, vibration, easy, easy peasy, but now where it comes down to, you might have seen a post we put on LinkedIn quite recently around counting, you know, when is it when there's an action that’s happening over and over again. You can just point a camera at it and count something happening. You can see things going on that otherwise it would be very difficult to monitor with a sensor. And I think that's something that's going to be really, is becoming a huge thing in manufacturing and again is enabled by getting on the cloud or getting data on the cloud and using that and computing power on tap and to adopt that more and more. That's something that we're seeing a lot and which is very exciting. 

Callum: And how can, you touched on a few things, how could speech and gesture recognition also be used as a tool in the same sort of way? 

Patrick: Well, yes. I mean, I mean, if you see anything that we put out, I mean anything to build a Razor, one of the first things that we're always considering is that human interaction with technology. Now just the nature of the things that we build, typically that will be a user interface on the screen has to be designed for usability. What we're finding is, as we're working in these industries where actually the people using our software are operating something else, they’re operating a piece of machinery. You know, they need their concentration somewhere else. If you were driving a car, it's very difficult to drop that, to go over here to interact with, you know, a device. So that's where we are using things like voice recognition. Gestures, where people can continue what they're doing, but if they interact with technology using their voice, and they can still get that benefit that enables them to do their job better without having to stop the thing that requires their attention. And that's a big application. 

Tony: I use Alexa for that. 

Patrick: Well, Alexa was a great one. Yeah, and I think a lot of us, who  particularly have a northern accent, remember a time where voice recognition didn't really work very well for us. You know, and we all struggle with that. And that was something that was really improved in recent years and became a lot more accessible. And I think that speaks to a lot of the ethical issues that you get with AI and making sure it does work for everyone.

00:24:27:12 - 00:24:46:16

Callum: I still struggle to understand some of the things that you say, but I think I caught the most of that… So… pitch shift it. Yeah, that's good. So yeah, just touching on something you said earlier about generative AI. How can that be used across industries and what are the benefits of that?

00:24:46:17 - 00:25:07:04

Patrick:  Well, I'm not going to try and say, you did a good job of saying that, but it's designed to say the very least, it's naturally one of those applications of AI that's probably become the biggest in the media, that's been picked up a lot. You know, we saw things like DALL-E, if you remember that time when images were getting generated. We touched on the thing with the Donald Trump photos and creating that kind of artwork. It's opened up a bit of a minefield for copyright. You know, who owns what? What if something was trained on an artist's images and produces another one who owns that and that kind of thing. Where I think it is a little bit safer at the minute and where we're seeing applications of industry, are things like, yeah, writing the hundredth version of a document that you've already written. I mean, we write a lot of documents and, and using things like text generation to, to, to I guess fill in that manual effort where the 800th document is the same as every time, every other time and let you focus on all the interesting bits. Again, link linking back to applications sort of the nuclear industry. You know, we worked with a few organisations who have a body of work documents that they've created over years. It's very hard to know what's in them. Having that AI that is trained on that can not just write new documents but can answer those questions. It can interact with them in a very natural language way that's a little bit more intuitive than the typical search box that we are used to. And so again, taking away that waste of people's time trawling through things when they should be doing other things is an exciting one.

00:26:29:22 - 00:27:15:18

Patrick: …and you know, replacing all designers as well. 

Tony: But I think if it's implemented helpfully, then maybe it can help us to reduce our working week. Hmm. But you know, not just reduce the salary but encourage enough so that it makes us more efficient, but it helps humans to reduce, you know, so we don't have to work as much. 

Patrick: I mean, I recently read a book called Utopia for Realists, if you look , I can’t remember the name of the author…

Callum: We’ll put it in the comments section…

00:27:15:21 - 00:27:38:18

Patrick: …which talks a lot about that, the ethics of AI itself. It's open to all these discussions of things like universal universal basic income. And I mentioned before about things like combine harvesters, there's a huge percentage of the people in this country who worked in farming before automation came in. But what happened is that different industries popped up and people moved to that. So actually generally, the output, the production in this country and jobs has pretty much gone hand-in-hand until recently. And we're starting to see more and more as things get automated and as AI comes in, we're starting to see that decoupling of those two things where production is continuing to go up in a lot of industries, but the jobs are starting to reduce and I think that's where we start to get into these conversations about short working weeks. Things like universal basic income. Yes. Because what I want, I would hate to see happen with AI is that we have all these tools now for generating art work. Yeah, that's what we should be freeing humans to do, we should be freeing humans to create art work, creating music. And you know, we've all seen your SoundCloud… and some things on that…

00:28:30:13 - 00:28:51:09

Patrick: So it's opened up a lot of interesting conversations in that space. So Utopia for Realists, I really recommend that book for that kind of thinking. 

Callum: I suppose it’s all talking about making our lives easier and making the human super human and really sort of creating a more efficient working structure. Speaking of which, I told ChatGPT that you were coming in today.

00:28:51:12 - 00:29:16:03

Patrick: Oh yeah, yeah. 

Callum: So to save me, you know, think up questions and stuff I just asked, what would you ask Patrick Murray, Head of Strategy and Data. This is the question, what are some of the most interesting insights or discoveries you've encountered whilst working with data or AI?

Patrick: Oh, good question. 

Callum: I can't take credit for like from that.

00:29:16:03 - 00:29:33:22

Patrick: I, I think one of the great things that I've seen with working with data, is it's often not, it doesn't always have to be some huge breakthrough when you collect the data, it's not always this huge ‘aha’ moment. It won't change things completely. It doesn't always have to be. We were working with a steel manufacturer in Sheffield who started picking up, you know, using their own data and collecting more of it to look for insight. One of the things that they found was on their, when they’re melting their steel and they’re closing the door of where they are heating it up. What they found was through measuring the temperature that they weren't quite closing the door the whole way. They were seeing where there was a big temperature difference, a huge drop off down right down by the door, just through putting these simple temperature sensors ot thermometers…but what that meant was a very simple change of making sure that you shut the door every time, because you can imagine the amount of electricity that goes into that kind of application, the amount of energy they have to pay for.

00:30:25:21 - 00:30:46:05

Patrick: They have been able to knock a few percentage points off that over the course of a year, actually resulted in a huge impact in the money they were saving. So while it's not always a huge exciting difference, those small things can actually make a very big impact even though they seem quite boring…

00:30:46:09 - 00:31:10:04

Tony: No I don’t think you can underestimate that number. It could be, it could change the actual structure of the metal, the heating up of the metal. Yeah. Yeah. So it's, yeah, it's quite interesting that. 

Callum: And it's the only thing that we can do in that space where we're getting these insights, we're able to take action and we are actually able to optimise, you know, processes for, for certain companies.

00:31:10:09 - 00:31:41:08

Patrick: Yeah, absolutely, and that's what we did. And the first step is always, it's very hard to improve things that you’re not measuring. And that's one of the great things with getting started with IOT and that's one of the things that we are going with our Edge to Cloud offering with Razor edge. It's just getting people starting to measure these things, starting to look at them, starting to make it part of their workflow when they're planning things and seeing what drops out because it often, you know, you can imagine with that application that I talked about, it didn't take very long. It didn't take many times making sure that door was shut for that little project to pay for itself many, many times over. And a lot of the conversations that I’m having in industry where, you know, we're going to events and we speak to manufacturers, you talk digital transformation and they go, Oh I’ve got other things we need to spend our money on, particularly the moment, that's a huge thing now. That’s something that we will work on over the next ten years. But it doesn't always have to be that huge thing. Projects like that, a perfect place to start, start small. Start with a, look for a high impact place to begin and then you can start moving different things. And that's where everyone needs to be starting. Don't always worry about a huge digital transformation thing. Think what are the problems that we are having and how can we put these solutions in to start improving them? 

Callum: Patrick, thank you very much. 

Patrick: Thank you. Thank you, Thank you. ChatGPT for  that question. 

Callum: There’s a few more if you want to… How can data be used to optimise processes for organisations?

Patrick: It's one of the obviously, areas that we work in, in logistics and we work with a big freight company in the UK. Um, it’s like what I said before about looking for different, rather than trying to do an all in one digital transformation, look for opportunities to find efficiencies. Obviously data driven tools was a big part of that whether it was looking for ways to combine different trains. But we actually built an optimiser using the data that they had on the different routes and the schedule. We managed to identify, you could reduce, we could reduce the whole fleet of trains by about 9% because there were so many inefficiencies that we found in it. And by collecting that data and combining it from different sources. So one of things that we build is a data warehouse, where we can pull in data from different sources and build a more complete picture of what's going on in your organisation to find those different inefficiencies and find opportunities where maybe those two bits of data wouldn't have come together otherwise. And so there’s a lot of opportunities there for organisations. 

00:33:53:16 - 00:33:56:04

Callum: Well, Tony, Patrick, it's been great to have you on.

00:33:56:07 - 00:33:59:20

Tony: Thanks very much. Thank you. 

Callum: And thanks for watching. See you next time. 

00:34:00:01 - 00:34:10:15