March 26, 2024
Adapting to the Ever-Changing Landscape: Insights from the Frontlines of Tech Innovation
Ryan Purvis delves into the dynamic world of tech innovation, uncovering valuable insights from the frontlines. From navigating the challenges of an evolving landscape to exploring the future of technology, this episode offers a firsthand perspective on adapting and thriving in the ever-changing tech industry.
Show links:
Follow us on Twitter: @thedwwpodcast
Follow us on LinkedIn: @theDWWPodcast
YouTube channel: TheDWWPodcast
Email us: podcast@digitalworkspace.works
Visit us: www.digitalworkspace.works
★ Support this podcast on Patreon ★Transcript
Adapting to the Ever-Changing Landscape: Insights from the Frontlines of Tech Innovation
===
Ryan Purvis: [00:00:00] Hello, and welcome to the Digital Workspace Works podcast. I'm Ryan Purvis, your host, supported by our producer, Heather Bicknell. In this series, you'll hear stories and opinions from experts in the field, stories from the front lines, the problems they face and how they solve them, the areas they're focused on from technology, people, and processes to the approaches they took that will help you to get to the scripts for the Digital Workspace inner workings.
How you doing?
Heather Bicknell: I'm good. It's you know, it's Friday. I'm going to see Dune 2 tonight. So good.
Ryan Purvis: Yeah. Did you watch Dune 1 again before you watched Dune 2 or not?
Heather Bicknell: Yeah. Kind of just installments throughout the week. But, no, it's, Dune 1. It's still good. It holds up. I'm excited to see all of the events that unfolds, what happens in the second half of that book.
Ryan Purvis: Cause it's not, it's it's not really a sequel. It's really the second part of the first movie.
Heather Bicknell: Yes, yeah, and they're really the first movie is [00:01:00] sort of just the initial setup of them, you know on Arrakis, so Yeah, lots of lots more like magical sort of things to happen in the second half.
Ryan Purvis: Hmm. Yeah No, I think it's gonna be really good. We might go watch it tomorrow night if we can get tickets.
My wife's never seen Dune, never read the books So it'll be interesting for her to be exposed to it.
Heather Bicknell: How far have you, I've read through Dune Messiah, so I read the first book and the second book. How far did you get in the series? There's so many and I don't know if it's worth it.
Ryan Purvis: Yeah, I'll be honest. I think I stopped after a while.
I think I read two or three of the books, and then I stopped. And I think I stopped just because I was struggling with the language a bit. And I started reading. There's another like other authors that I like to read. Their stuff came out and I was like, I'll just read those books because I know those are good books to read, et cetera.
And that's probably what happened is I just refocused and then I haven't gone back to them. But I have been thinking about getting, the audio books. I mean, a lot of audio books recently. And I find [00:02:00] that's actually a good way with some of these sort of like, Wheels of Time and, The Lord of the Rings and all those sorts of books, which are very long reads.
But you can audio book them and do them in one and a half times speed. And then it's not, it's not as bad, you know, so, so that's what I mean. That's what I've been thinking of doing.
Heather Bicknell: Yeah. Well, if it's a good narrator, let me know. Maybe I'll do that too.
Ryan Purvis: Yeah. It's, it's very good.
Like, Alex Hamazi, I've been listening to a lot of his stuff. He's got three books out of very much sales, marketing stuff. So I like to read his books and listen to the audio book separately. And that's really nice because you can listen to like, I listened to the audio book first and you kind of get a sense of what's coming and what's interesting.
And then you go read the book, but you then you just basically go straight ahead to where you want to look at. And I think there's actually a way with whisper sync that you can flag stuff. And say, this is interesting. So when you read the book, it actually synchronizes it to the book. I haven't figured out how to do that yet, but I think it can do that.
Heather Bicknell: Is WhisperSync, is that the name of an app or is that a feature? I haven't heard that term before.
Ryan Purvis: That's part of, the Kindle, Audible.
Heather Bicknell: Okay. [00:03:00] I don't have Kindle, so,
Ryan Purvis: Oh, don't you? Yeah. Well, you can do it on your app. You can do it on the phone.
Heather Bicknell: Yeah. I haven't gotten big in audio books, but anyway, how have you been?
Ryan Purvis: Yeah, fine. Fine. Yeah. Audio books have a kind of flip flop between them. Sometimes like, uh, like really get into them. Sometimes it's just like, I can't be bothered. So, you can do an audible subscription for about seven pounds a month and you get one credit.
But if you don't do the subscription and you were to go buy an audio book, you're looking at 21 to 28 pounds a book. So it's really worth it. If you want, and if you are like not buying books, you can just build the credits up and then you'd like, they do like a two for one deal and all that kind of stuff, and that's really worth it.
Cause then you can go back, like two books, but even less. And then I, with Kindle unlimited, If the book is on both, then you don't get to see the book. I mean, the Kindle unlimited is free anyway, but then it will synchronize where you are from a listening point of view to where you are on the reading.
Like, so if you read a couple, like, sometimes I read a book, I'm reading the book and I'm like, geez, I really like, I've got to go do this thing now, but I really don't want to stop reading the book. [00:04:00] Then the audio will pick up from where I've stopped reading and then I can still carry on the story.
Heather Bicknell: That's a cool feature.
Ryan Purvis: Yeah, so now that good. We just busy with work. FinXone is really doing well from a market point of view. It's going to get the product and delivery into a good state now because that's the usual thing when you build a startup. You sort of keep follow you. You spin the plates.
And, when you're splitting the plates, like you're sorting out the sales, the marketing, then you, all of a sudden you realize the product's a little bit behind, so you've got to sort the product out, and then you sort delivery out, and then it all comes at once. So we're, growing nicely. We just reached 30 people.
And we're adding clients every month now, which is great. So it's really on the right trajectory. So that's keeping me busy and then Valuu, we've made some really nice movements with some of the build out of the product. So some of the stuff that I, I was actually looking at some notes I've been asking for since November have finally come into the product.
So it's, it's starting to take shape as well. It was very unstable for a while, not unstable, just the guys who built the first version of it just really didn't understand. [00:05:00] What we're trying to achieve. And they also didn't understand the platform. So we're building it inside of bubble and you know, even I was going through it going, why would you do it this way?
This makes no sense. And they're really just like, if I gave my six year old, my three year old bubble and asked him to build it, that's what I got. And you know, you're paying people money who are professionals and they're not professionals. And then I had another guy come in and work on it, who
had the opportunity to learn how to use it on. It was just the same problem. Just also doesn't get it. So the problem with these low code no code platforms still most of them. I probably consider WordPress to be a low code platform is you still need a developer who knows how to use the tool and knows how to think like a developer.
And is something we're fixing the FinXone. FinXone is, you know, we're trying to make it as end user friendly as possible. And that's where deterministic AI and that will come into it. So I'm enjoying the bubble experience because I'm learning a hell of a lot of things about how crap it can be if you don't do it properly.
But it's a powerful tool. And if you know how to use it, I mean, I do some very basic things in my mind, very basic, and you can build something [00:06:00] pretty quickly. But it's just takes that, you still need to have that logical brain that goes. We need to have a data model. We need to understand how the pages fit together.
We need to, do this integration and integration. And that's obviously important part of building any product really is having that delivery, execution mentality, I guess.
Heather Bicknell: It was interesting. I saw an announcement yesterday where the, Biden administration, they'd put out a kind of like an ask to technology companies to stop using, I think it was C plus plus. Maybe C and C as programming languages because of more, yeah, like there's something about the way they use memory and I think it's like a cybersecurity thing.
So sort of like a, yeah, from that angle, I thought it was really interesting, but it just made me think about, I don't know, other like newer languages or things like no code tools or I kind of skill. Like, I don't know, the, the shift, I suppose, in the tool stack that [00:07:00] we're likely to see from, you know, the next decade.
Ryan Purvis: Yeah, I mean, C++ and C and all those, I mean, they're old languages, and I mean, they're still in use. I mean, it's scary to think they're still in use.
Heather Bicknell: Like 20 percent of the, of software still uses them, I think, is what the article.
Ryan Purvis: Yeah, and, and I mean, one of the main thing with it is that the, um, Because of the memory access that it has, you know, a lot of the sort of, I think the third generation languages, C sharp, Java, et cetera, memory management is handled by the framework.
So you can't really do anything. You can try, but it's still managed by the framework. And You have your garbage collectors to go through and clean up objects that have been closed off that sort of thing. The thing with C ++ Or, or the access that has is that you can literally bulk your whole machine if you get it wrong.
You know, you could write over memory that you shouldn't be writing over and, you know, it has access to all the nuts and the bolts. So, it's, you know, pointers and references are very abstract concepts for a lot of people. So you really find it's a small niche of people that can actually write good stuff at [00:08:00] C++.
And write it well. But they are powerful languages, but in the same token, a lot of the third generation, even fourth generation languages are also very powerful. And then their access to the core things is normally through a much better mechanism. Which again is more controlled. So I, you know, that 20 percent even to me sounds high for what should be out there.
You know, you probably want to see that getting down to like a 5%. Because all the other language should be able to do that. But I even think, you know, we talked about it yesterday. In fact, that with the way the general AI is working and going, your need to write code is going to get less and less and less over time.
There will still be people writing code. Same as there's still people with driving manual cars with no electronics in them. It's just a very small percentage because what's gonna happen is you start generating code and you're gonna need someone understands how to put that code together or to check that it's right, but they're not going to write all the code themselves.
And at some point, the AI is gonna get so good that they're just gonna write the whole thing for you, and that's it. I mean, you're just gonna [00:09:00] basically be using it. Basically, the simple stuff like make me a form that captures these fields that search into a database, you know, ChatGPT can do that now.
You just need to ask the right question. You know, I want to do this database with these fields. I want to be able to edit it. I want to be able to add it to your product engineering. That becomes really important. So that's why I think there will be some level of, Logical brain still involved, but you don't have to be a developer anymore.
Heather Bicknell: You need the skills to kind of audit it. But a lot of the manual stuff is taken away.
Ryan Purvis: The other thing with these mechanisms right now. If you gave the same prompt five times, you get five different results. So, it's very rare that, the LLM, or at least Chantiby, for what I've done it with, It's going to improve on its previous code in an easy way.
So, for example, if I asked it to write me a script to do something. And then I'd say, okay, but your script has made a mistake here, there's an error, fix this error. It'll regenerate the script sometimes, completely different. So now it's fixed the error, [00:10:00] but now it's created two new errors because it didn't actually fix the problem that you were trying to fix, it just fixed what it thought it had to fix.
And then sometimes what it'll do is you'll say, okay, I've got the script, now I want to add this extra field to be done as well. Well, this extra method and it'll only regenerate that method, but then it'll leave out all the other stuff you actually needed to instantiate their method and use it. So it doesn't completely, it doesn't do a hundred percent perfect code.
It's like 50%. So that's where you still need that skill person. And I see it kind of the same way as you're building a PowerPoint deck or a Word document. You're going to write down the key bullet points of what you want to do, and the AI generates an output for you, and you're going to spend your time tweaking it to make it work.
Heather Bicknell: I haven't personally been hands on with Copilot at all, but I'm looking forward to the PowerPoint piece in particular. The ability to create slides from words, I think would be a game changer if it, Yeah, again, if it works kind of well enough, it's always the caveat, the quality needs to be to a certain standard.
Ryan Purvis: Yeah, and it's, [00:11:00] I mean, you see it now with Sora, which is the video generation thing. You're giving the AI a paragraph and it's generating a video for you, which is really about, you know, 30 frames per second of images to create that. It's conceivable that your Copilot in Microsoft, for example, would be taking your paragraph and generating you a couple slides.
Now what it's using to generate the slides from, is going to be based on what it's been trained on. So it may not be Heather's way of making slides. It might be the generic way that most people do it. And it'll be personalized over time and I don't think it's far away where it'll start picking up your style or my style or whatever it is and you can say in my style generate my slides for me.
Ryan Purvis: And that I think should be really, really awesome. Especially if you can get it to generate on the right template. Like the right color scheme and that is crucial.
Heather Bicknell: You know, absolutely. Otherwise it's not really a time saver.
Ryan Purvis: Yeah, because that is the biggest waste of time.
It's like I was writing a proposal yesterday. My word document and I was like, all the colors are wrong. All the bullet points are wrong. All the headers are wrong. And word is a [00:12:00] terrible tool to try and figure that stuff out. It's just not usable. And I mean, that's why there's products that help you do it.
But, you know, that's what I would really help me. It's just like, here's my document in free form. How I've written it. Turn this into my proposal. Yeah. With T's and C's.
Heather Bicknell: Yeah, absolutely. Have you done any new AI experiments or had any different, I guess tweaks to what you've been doing since we've, Last chatted?
Ryan Purvis: So we've actually gone away a bit from the AI stuff. So, we were doing all our marketing using ChatGPT and, Tapio and various other tools. And we actually come backwards. We've come back to manual work. So what we're doing now is I'm recording, WhatsApp voice notes.
In response to topics and then Anabelle is taking those and turning them into posts with Will and with Christian. And we posted those, good old fashioned way manually. or maybe they're scheduling a little bit, but, there's no AI involved in that too much anymore. Unless we're doing that to get the topics.
I don't know. But what we were finding is that people getting irritated with the generated posts because they always have emoticons in it. They always look a bit weird. I thought Tapio was actually not [00:13:00] bad. But then we stopped using Tapio and started using ChatGPT and it just wasn't as good.
So yeah, that's cut down a bit. I still use it as a general tool. Like if I'm writing a proposal, I still put my stuff in and say, here's what I want to write. Please rewrite this in professional tone as a salesperson or whatever it is, or as an architect. And it does that for me nicely. And, yeah, it's nothing really new.
I think I think it's now become part of my workflow, so I don't even notice I'm doing it. Yeah, how about yourself?
Heather Bicknell: Interesting. I'd be curious if the if you're seeing the social posts perform better with or without AI. But, no, you know, look, I haven't used it too much for work applications. I think where I struggle with it is a lot of what I do at least professionally relies on the most up to date data.
And it's really sort of kind of real time what's happening in the market. And I find that obviously with like free ChatGPT that is, trained on old data. I think, but even the, GPT four, isn't like quite. I don't think it's accessing the modern. I could be wrong. [00:14:00] I don't think it's like the latest, you know, stuff that's coming out on the web.
So there's sort of like that whole element of of research where it could be really potentially helpful, but, can't quite tap into it. But, I have been using it to summarize different documents and kind of, some different tools to pull in different documents and kind of try to summarize things together.
You just find, like. It's still, you know, it's hard to get past some hallucination behavior or like is so confident about stating the wrong thing. So I'm still, I'm experimenting, but still kind of struggling to find like my killer use cases.
Ryan Purvis: Yeah, yeah, that's exactly the challenges. It's actually more of an age technology at this point in time.
As much as some people are saying it's a front and center one. I still think it's edge. But it's useful. It's useful. I saw an interesting video, this morning, I think and I'm not sure who it was, but it was, I think it's a minister, must have been a minister in Singapore, might even be their prime minister.
I don't know. It was on LinkedIn. And, basically they have put in place a grant for over 40 year olds to get further [00:15:00] educated. Because they realize that AI is going to change things. And they're basically paying their people, which has always been their model to get their people educated, to skill up, to handle AI. So you can go back on the government's tab and learn how to use AI technologies.
Which is quite a fundamentally different thing. Most countries don't do that. Like no one does that.
Heather Bicknell: Yeah. Well, that's very, very forward thinking. That is interesting. I mean, it does feel like, again, I know we've talked about this many a time, but in terms of the technological leaps that have happened to personal computing and the internet and all the rest, it's like the next phase. So it is, going to more biased to younger generations eventually in the workforce who, kind of have grown up with it and know how to harness it.
So focusing on some education and having it be coming from the government makes a lot of sense. I think, even just being able to sniff it out and misinformation and all the rest.
Ryan Purvis: Yeah, I mean, even the, well, I mean, Singapore's always had that model that that's what made them, what they are is that they've educated their, population.
I mean, I, think they're a small population by a couple million. And that was [00:16:00] their, inflection point was focusing on their education across the board. So I'd watch them as a very strong player. Going forward in AI markets. Because that's the biggest problem with this is if you're not using this stuff, you're going to fall so far behind.
It's not like you're falling behind like a week or two, you're falling behind by months, years, but every day you're not using it. Because even if you go, what Saron does with generating video, I mean, how many roles are you going to have? How many digital people were creating videos as a job, graphic design, et cetera, and AI is going to do that for you and do it for you for cheap, like really cheap.
So, you know, you've got to make that part of your service offering because the creativity is going to come from the human, the throughput, the muscle is going to come from the compute. And I mean, I'm expecting to see more and more data centers going up. But just more and more AI servers being put in them.
So, you know, the power demand is going to go up. And it's kind of that thing where they're going to build capacity and you're just going to see it all of a sudden it's everywhere.
Heather Bicknell: Yeah, actually, Gartner had some really interesting, stuff about this for their 2024 tech predictions, but [00:17:00] actually hadn't thought about it
to this degree. But generative AI is so power intensive that, sort of green it and sustainability and any, even just like the amount, like we're going to reach a limit of the amount of, power that we can draw from. Like, it is so energy intensive that sort of part and parcel how large organizations need to think about, their ability to leverage these things because there's only so much resource out there.
So I guess I hadn't thought about, actually drawing so much in the power grid that you sort of max it out, but that's where Gardner's these things kind of come in.
Ryan Purvis: Yeah, I mean, yeah, I mean, we got this huge sun out there that generates a lot of power. So we should be seeing a lot more, solar and wind and all that generation coming in.
But I wonder, I mean, this might be very science fiction, but, you know, are we not seeing at some point what Elon Musk is doing with his Starlink network? The first sort of opportunity to start putting servers in space on the moon. To do computational stuff. I [00:18:00] mean, cause then it's constantly fed solar radiation to generate power and, you know, beaming the power down.
I mean, I remember reading a paper many years ago for, collecting energy with satellites outside of the atmosphere and then beaming it down with lasers to collectors. I mean, it sounds crazy, but actually, maybe that's what will happen. I don't know.
Heather Bicknell: I guess we'll see. The funny thing was Starlink. So wearing my Yosemite sweatshirt today, but I was there, in the fall of last year.
It's a national park, in the U. S., but they there's really good stargazing at night and. The ranger at the park was talking about how some nights we'll see starlink, when they're doing the stargazing and that it's like, it's a pretty creepy thing to see that, you know, network of satellites in the sky.
Because most of the time we don't think about it. But when you, you know, it's like an alien kind of a thing, just, just kind of throwing, satellites up there. It's just, it's to think about, just what we have allowed to happen in there, I guess, is just thing that it's like changing the landscape of the night sky, to [00:19:00] the naked eye.
That's just, I hadn't realized.
Ryan Purvis: Yeah, I thought, I thought they actually had them painted black now. Something at the bottom so that you couldn't see them. I thought I remember reading something like that. I mean, I could have misread it. But, I know what you mean, because it is very futuristic. I mean, if you think about, like, I'm watching the Foundation series on Apple TV, off and on.
And, some of that stuff, I mean, that's really old. Like, I think it's really old written science fiction, but it's been really modernized. The CGI and that, and having the sort of Dyson sphere coverage, which is what Sonic is in some respect, it's like all the satellites covering the planet. Yeah. Not obviously, but it's, it's kind of a virtualized Dyson sphere, I guess.
It's quite a freaky concept. I mean, to conceive about what's happened to do that and to keep those satellites in orbit and to keep them in, you know, there must be a little, adjustments going on all the time and, you know, data being sent between them. I mean, I'd love to know how that's all working.
I mean, it must be absolutely fascinating.
Heather Bicknell: I'm sure it would blow some minds. Well, I know we're about out of time here, but it was good to catch up and I can't [00:20:00] wait to hear what you think of Dune.
Ryan Purvis: Yeah. Let me know when you've watched it. Yeah. Tell me.
Heather Bicknell: Yeah, we'll do.
Ryan Purvis: Super. Thanks. Talk to you later.
Heather Bicknell: Have a good weekend.
Ryan Purvis: You too.
Bye.
Thank you for listening to today's episode. Heather Beckner is our producer editor. Thank you, Heather. For your hard work on this episode, please subscribe to the series and rate us on iTunes at the Google Play Store.
Follow us on Twitter at the Dww Podcast. The show notes and transcripts will be available on the website, www.digital workspace works. Please also visit our website, www.digital workspace works. And lastly, if you found this episode useful, please share with your friends or colleagues.
===
Ryan Purvis: [00:00:00] Hello, and welcome to the Digital Workspace Works podcast. I'm Ryan Purvis, your host, supported by our producer, Heather Bicknell. In this series, you'll hear stories and opinions from experts in the field, stories from the front lines, the problems they face and how they solve them, the areas they're focused on from technology, people, and processes to the approaches they took that will help you to get to the scripts for the Digital Workspace inner workings.
How you doing?
Heather Bicknell: I'm good. It's you know, it's Friday. I'm going to see Dune 2 tonight. So good.
Ryan Purvis: Yeah. Did you watch Dune 1 again before you watched Dune 2 or not?
Heather Bicknell: Yeah. Kind of just installments throughout the week. But, no, it's, Dune 1. It's still good. It holds up. I'm excited to see all of the events that unfolds, what happens in the second half of that book.
Ryan Purvis: Cause it's not, it's it's not really a sequel. It's really the second part of the first movie.
Heather Bicknell: Yes, yeah, and they're really the first movie is [00:01:00] sort of just the initial setup of them, you know on Arrakis, so Yeah, lots of lots more like magical sort of things to happen in the second half.
Ryan Purvis: Hmm. Yeah No, I think it's gonna be really good. We might go watch it tomorrow night if we can get tickets.
My wife's never seen Dune, never read the books So it'll be interesting for her to be exposed to it.
Heather Bicknell: How far have you, I've read through Dune Messiah, so I read the first book and the second book. How far did you get in the series? There's so many and I don't know if it's worth it.
Ryan Purvis: Yeah, I'll be honest. I think I stopped after a while.
I think I read two or three of the books, and then I stopped. And I think I stopped just because I was struggling with the language a bit. And I started reading. There's another like other authors that I like to read. Their stuff came out and I was like, I'll just read those books because I know those are good books to read, et cetera.
And that's probably what happened is I just refocused and then I haven't gone back to them. But I have been thinking about getting, the audio books. I mean, a lot of audio books recently. And I find [00:02:00] that's actually a good way with some of these sort of like, Wheels of Time and, The Lord of the Rings and all those sorts of books, which are very long reads.
But you can audio book them and do them in one and a half times speed. And then it's not, it's not as bad, you know, so, so that's what I mean. That's what I've been thinking of doing.
Heather Bicknell: Yeah. Well, if it's a good narrator, let me know. Maybe I'll do that too.
Ryan Purvis: Yeah. It's, it's very good.
Like, Alex Hamazi, I've been listening to a lot of his stuff. He's got three books out of very much sales, marketing stuff. So I like to read his books and listen to the audio book separately. And that's really nice because you can listen to like, I listened to the audio book first and you kind of get a sense of what's coming and what's interesting.
And then you go read the book, but you then you just basically go straight ahead to where you want to look at. And I think there's actually a way with whisper sync that you can flag stuff. And say, this is interesting. So when you read the book, it actually synchronizes it to the book. I haven't figured out how to do that yet, but I think it can do that.
Heather Bicknell: Is WhisperSync, is that the name of an app or is that a feature? I haven't heard that term before.
Ryan Purvis: That's part of, the Kindle, Audible.
Heather Bicknell: Okay. [00:03:00] I don't have Kindle, so,
Ryan Purvis: Oh, don't you? Yeah. Well, you can do it on your app. You can do it on the phone.
Heather Bicknell: Yeah. I haven't gotten big in audio books, but anyway, how have you been?
Ryan Purvis: Yeah, fine. Fine. Yeah. Audio books have a kind of flip flop between them. Sometimes like, uh, like really get into them. Sometimes it's just like, I can't be bothered. So, you can do an audible subscription for about seven pounds a month and you get one credit.
But if you don't do the subscription and you were to go buy an audio book, you're looking at 21 to 28 pounds a book. So it's really worth it. If you want, and if you are like not buying books, you can just build the credits up and then you'd like, they do like a two for one deal and all that kind of stuff, and that's really worth it.
Cause then you can go back, like two books, but even less. And then I, with Kindle unlimited, If the book is on both, then you don't get to see the book. I mean, the Kindle unlimited is free anyway, but then it will synchronize where you are from a listening point of view to where you are on the reading.
Like, so if you read a couple, like, sometimes I read a book, I'm reading the book and I'm like, geez, I really like, I've got to go do this thing now, but I really don't want to stop reading the book. [00:04:00] Then the audio will pick up from where I've stopped reading and then I can still carry on the story.
Heather Bicknell: That's a cool feature.
Ryan Purvis: Yeah, so now that good. We just busy with work. FinXone is really doing well from a market point of view. It's going to get the product and delivery into a good state now because that's the usual thing when you build a startup. You sort of keep follow you. You spin the plates.
And, when you're splitting the plates, like you're sorting out the sales, the marketing, then you, all of a sudden you realize the product's a little bit behind, so you've got to sort the product out, and then you sort delivery out, and then it all comes at once. So we're, growing nicely. We just reached 30 people.
And we're adding clients every month now, which is great. So it's really on the right trajectory. So that's keeping me busy and then Valuu, we've made some really nice movements with some of the build out of the product. So some of the stuff that I, I was actually looking at some notes I've been asking for since November have finally come into the product.
So it's, it's starting to take shape as well. It was very unstable for a while, not unstable, just the guys who built the first version of it just really didn't understand. [00:05:00] What we're trying to achieve. And they also didn't understand the platform. So we're building it inside of bubble and you know, even I was going through it going, why would you do it this way?
This makes no sense. And they're really just like, if I gave my six year old, my three year old bubble and asked him to build it, that's what I got. And you know, you're paying people money who are professionals and they're not professionals. And then I had another guy come in and work on it, who
had the opportunity to learn how to use it on. It was just the same problem. Just also doesn't get it. So the problem with these low code no code platforms still most of them. I probably consider WordPress to be a low code platform is you still need a developer who knows how to use the tool and knows how to think like a developer.
And is something we're fixing the FinXone. FinXone is, you know, we're trying to make it as end user friendly as possible. And that's where deterministic AI and that will come into it. So I'm enjoying the bubble experience because I'm learning a hell of a lot of things about how crap it can be if you don't do it properly.
But it's a powerful tool. And if you know how to use it, I mean, I do some very basic things in my mind, very basic, and you can build something [00:06:00] pretty quickly. But it's just takes that, you still need to have that logical brain that goes. We need to have a data model. We need to understand how the pages fit together.
We need to, do this integration and integration. And that's obviously important part of building any product really is having that delivery, execution mentality, I guess.
Heather Bicknell: It was interesting. I saw an announcement yesterday where the, Biden administration, they'd put out a kind of like an ask to technology companies to stop using, I think it was C plus plus. Maybe C and C as programming languages because of more, yeah, like there's something about the way they use memory and I think it's like a cybersecurity thing.
So sort of like a, yeah, from that angle, I thought it was really interesting, but it just made me think about, I don't know, other like newer languages or things like no code tools or I kind of skill. Like, I don't know, the, the shift, I suppose, in the tool stack that [00:07:00] we're likely to see from, you know, the next decade.
Ryan Purvis: Yeah, I mean, C++ and C and all those, I mean, they're old languages, and I mean, they're still in use. I mean, it's scary to think they're still in use.
Heather Bicknell: Like 20 percent of the, of software still uses them, I think, is what the article.
Ryan Purvis: Yeah, and, and I mean, one of the main thing with it is that the, um, Because of the memory access that it has, you know, a lot of the sort of, I think the third generation languages, C sharp, Java, et cetera, memory management is handled by the framework.
So you can't really do anything. You can try, but it's still managed by the framework. And You have your garbage collectors to go through and clean up objects that have been closed off that sort of thing. The thing with C ++ Or, or the access that has is that you can literally bulk your whole machine if you get it wrong.
You know, you could write over memory that you shouldn't be writing over and, you know, it has access to all the nuts and the bolts. So, it's, you know, pointers and references are very abstract concepts for a lot of people. So you really find it's a small niche of people that can actually write good stuff at [00:08:00] C++.
And write it well. But they are powerful languages, but in the same token, a lot of the third generation, even fourth generation languages are also very powerful. And then their access to the core things is normally through a much better mechanism. Which again is more controlled. So I, you know, that 20 percent even to me sounds high for what should be out there.
You know, you probably want to see that getting down to like a 5%. Because all the other language should be able to do that. But I even think, you know, we talked about it yesterday. In fact, that with the way the general AI is working and going, your need to write code is going to get less and less and less over time.
There will still be people writing code. Same as there's still people with driving manual cars with no electronics in them. It's just a very small percentage because what's gonna happen is you start generating code and you're gonna need someone understands how to put that code together or to check that it's right, but they're not going to write all the code themselves.
And at some point, the AI is gonna get so good that they're just gonna write the whole thing for you, and that's it. I mean, you're just gonna [00:09:00] basically be using it. Basically, the simple stuff like make me a form that captures these fields that search into a database, you know, ChatGPT can do that now.
You just need to ask the right question. You know, I want to do this database with these fields. I want to be able to edit it. I want to be able to add it to your product engineering. That becomes really important. So that's why I think there will be some level of, Logical brain still involved, but you don't have to be a developer anymore.
Heather Bicknell: You need the skills to kind of audit it. But a lot of the manual stuff is taken away.
Ryan Purvis: The other thing with these mechanisms right now. If you gave the same prompt five times, you get five different results. So, it's very rare that, the LLM, or at least Chantiby, for what I've done it with, It's going to improve on its previous code in an easy way.
So, for example, if I asked it to write me a script to do something. And then I'd say, okay, but your script has made a mistake here, there's an error, fix this error. It'll regenerate the script sometimes, completely different. So now it's fixed the error, [00:10:00] but now it's created two new errors because it didn't actually fix the problem that you were trying to fix, it just fixed what it thought it had to fix.
And then sometimes what it'll do is you'll say, okay, I've got the script, now I want to add this extra field to be done as well. Well, this extra method and it'll only regenerate that method, but then it'll leave out all the other stuff you actually needed to instantiate their method and use it. So it doesn't completely, it doesn't do a hundred percent perfect code.
It's like 50%. So that's where you still need that skill person. And I see it kind of the same way as you're building a PowerPoint deck or a Word document. You're going to write down the key bullet points of what you want to do, and the AI generates an output for you, and you're going to spend your time tweaking it to make it work.
Heather Bicknell: I haven't personally been hands on with Copilot at all, but I'm looking forward to the PowerPoint piece in particular. The ability to create slides from words, I think would be a game changer if it, Yeah, again, if it works kind of well enough, it's always the caveat, the quality needs to be to a certain standard.
Ryan Purvis: Yeah, and it's, [00:11:00] I mean, you see it now with Sora, which is the video generation thing. You're giving the AI a paragraph and it's generating a video for you, which is really about, you know, 30 frames per second of images to create that. It's conceivable that your Copilot in Microsoft, for example, would be taking your paragraph and generating you a couple slides.
Now what it's using to generate the slides from, is going to be based on what it's been trained on. So it may not be Heather's way of making slides. It might be the generic way that most people do it. And it'll be personalized over time and I don't think it's far away where it'll start picking up your style or my style or whatever it is and you can say in my style generate my slides for me.
Ryan Purvis: And that I think should be really, really awesome. Especially if you can get it to generate on the right template. Like the right color scheme and that is crucial.
Heather Bicknell: You know, absolutely. Otherwise it's not really a time saver.
Ryan Purvis: Yeah, because that is the biggest waste of time.
It's like I was writing a proposal yesterday. My word document and I was like, all the colors are wrong. All the bullet points are wrong. All the headers are wrong. And word is a [00:12:00] terrible tool to try and figure that stuff out. It's just not usable. And I mean, that's why there's products that help you do it.
But, you know, that's what I would really help me. It's just like, here's my document in free form. How I've written it. Turn this into my proposal. Yeah. With T's and C's.
Heather Bicknell: Yeah, absolutely. Have you done any new AI experiments or had any different, I guess tweaks to what you've been doing since we've, Last chatted?
Ryan Purvis: So we've actually gone away a bit from the AI stuff. So, we were doing all our marketing using ChatGPT and, Tapio and various other tools. And we actually come backwards. We've come back to manual work. So what we're doing now is I'm recording, WhatsApp voice notes.
In response to topics and then Anabelle is taking those and turning them into posts with Will and with Christian. And we posted those, good old fashioned way manually. or maybe they're scheduling a little bit, but, there's no AI involved in that too much anymore. Unless we're doing that to get the topics.
I don't know. But what we were finding is that people getting irritated with the generated posts because they always have emoticons in it. They always look a bit weird. I thought Tapio was actually not [00:13:00] bad. But then we stopped using Tapio and started using ChatGPT and it just wasn't as good.
So yeah, that's cut down a bit. I still use it as a general tool. Like if I'm writing a proposal, I still put my stuff in and say, here's what I want to write. Please rewrite this in professional tone as a salesperson or whatever it is, or as an architect. And it does that for me nicely. And, yeah, it's nothing really new.
I think I think it's now become part of my workflow, so I don't even notice I'm doing it. Yeah, how about yourself?
Heather Bicknell: Interesting. I'd be curious if the if you're seeing the social posts perform better with or without AI. But, no, you know, look, I haven't used it too much for work applications. I think where I struggle with it is a lot of what I do at least professionally relies on the most up to date data.
And it's really sort of kind of real time what's happening in the market. And I find that obviously with like free ChatGPT that is, trained on old data. I think, but even the, GPT four, isn't like quite. I don't think it's accessing the modern. I could be wrong. [00:14:00] I don't think it's like the latest, you know, stuff that's coming out on the web.
So there's sort of like that whole element of of research where it could be really potentially helpful, but, can't quite tap into it. But, I have been using it to summarize different documents and kind of, some different tools to pull in different documents and kind of try to summarize things together.
You just find, like. It's still, you know, it's hard to get past some hallucination behavior or like is so confident about stating the wrong thing. So I'm still, I'm experimenting, but still kind of struggling to find like my killer use cases.
Ryan Purvis: Yeah, yeah, that's exactly the challenges. It's actually more of an age technology at this point in time.
As much as some people are saying it's a front and center one. I still think it's edge. But it's useful. It's useful. I saw an interesting video, this morning, I think and I'm not sure who it was, but it was, I think it's a minister, must have been a minister in Singapore, might even be their prime minister.
I don't know. It was on LinkedIn. And, basically they have put in place a grant for over 40 year olds to get further [00:15:00] educated. Because they realize that AI is going to change things. And they're basically paying their people, which has always been their model to get their people educated, to skill up, to handle AI. So you can go back on the government's tab and learn how to use AI technologies.
Which is quite a fundamentally different thing. Most countries don't do that. Like no one does that.
Heather Bicknell: Yeah. Well, that's very, very forward thinking. That is interesting. I mean, it does feel like, again, I know we've talked about this many a time, but in terms of the technological leaps that have happened to personal computing and the internet and all the rest, it's like the next phase. So it is, going to more biased to younger generations eventually in the workforce who, kind of have grown up with it and know how to harness it.
So focusing on some education and having it be coming from the government makes a lot of sense. I think, even just being able to sniff it out and misinformation and all the rest.
Ryan Purvis: Yeah, I mean, even the, well, I mean, Singapore's always had that model that that's what made them, what they are is that they've educated their, population.
I mean, I, think they're a small population by a couple million. And that was [00:16:00] their, inflection point was focusing on their education across the board. So I'd watch them as a very strong player. Going forward in AI markets. Because that's the biggest problem with this is if you're not using this stuff, you're going to fall so far behind.
It's not like you're falling behind like a week or two, you're falling behind by months, years, but every day you're not using it. Because even if you go, what Saron does with generating video, I mean, how many roles are you going to have? How many digital people were creating videos as a job, graphic design, et cetera, and AI is going to do that for you and do it for you for cheap, like really cheap.
So, you know, you've got to make that part of your service offering because the creativity is going to come from the human, the throughput, the muscle is going to come from the compute. And I mean, I'm expecting to see more and more data centers going up. But just more and more AI servers being put in them.
So, you know, the power demand is going to go up. And it's kind of that thing where they're going to build capacity and you're just going to see it all of a sudden it's everywhere.
Heather Bicknell: Yeah, actually, Gartner had some really interesting, stuff about this for their 2024 tech predictions, but [00:17:00] actually hadn't thought about it
to this degree. But generative AI is so power intensive that, sort of green it and sustainability and any, even just like the amount, like we're going to reach a limit of the amount of, power that we can draw from. Like, it is so energy intensive that sort of part and parcel how large organizations need to think about, their ability to leverage these things because there's only so much resource out there.
So I guess I hadn't thought about, actually drawing so much in the power grid that you sort of max it out, but that's where Gardner's these things kind of come in.
Ryan Purvis: Yeah, I mean, yeah, I mean, we got this huge sun out there that generates a lot of power. So we should be seeing a lot more, solar and wind and all that generation coming in.
But I wonder, I mean, this might be very science fiction, but, you know, are we not seeing at some point what Elon Musk is doing with his Starlink network? The first sort of opportunity to start putting servers in space on the moon. To do computational stuff. I [00:18:00] mean, cause then it's constantly fed solar radiation to generate power and, you know, beaming the power down.
I mean, I remember reading a paper many years ago for, collecting energy with satellites outside of the atmosphere and then beaming it down with lasers to collectors. I mean, it sounds crazy, but actually, maybe that's what will happen. I don't know.
Heather Bicknell: I guess we'll see. The funny thing was Starlink. So wearing my Yosemite sweatshirt today, but I was there, in the fall of last year.
It's a national park, in the U. S., but they there's really good stargazing at night and. The ranger at the park was talking about how some nights we'll see starlink, when they're doing the stargazing and that it's like, it's a pretty creepy thing to see that, you know, network of satellites in the sky.
Because most of the time we don't think about it. But when you, you know, it's like an alien kind of a thing, just, just kind of throwing, satellites up there. It's just, it's to think about, just what we have allowed to happen in there, I guess, is just thing that it's like changing the landscape of the night sky, to [00:19:00] the naked eye.
That's just, I hadn't realized.
Ryan Purvis: Yeah, I thought, I thought they actually had them painted black now. Something at the bottom so that you couldn't see them. I thought I remember reading something like that. I mean, I could have misread it. But, I know what you mean, because it is very futuristic. I mean, if you think about, like, I'm watching the Foundation series on Apple TV, off and on.
And, some of that stuff, I mean, that's really old. Like, I think it's really old written science fiction, but it's been really modernized. The CGI and that, and having the sort of Dyson sphere coverage, which is what Sonic is in some respect, it's like all the satellites covering the planet. Yeah. Not obviously, but it's, it's kind of a virtualized Dyson sphere, I guess.
It's quite a freaky concept. I mean, to conceive about what's happened to do that and to keep those satellites in orbit and to keep them in, you know, there must be a little, adjustments going on all the time and, you know, data being sent between them. I mean, I'd love to know how that's all working.
I mean, it must be absolutely fascinating.
Heather Bicknell: I'm sure it would blow some minds. Well, I know we're about out of time here, but it was good to catch up and I can't [00:20:00] wait to hear what you think of Dune.
Ryan Purvis: Yeah. Let me know when you've watched it. Yeah. Tell me.
Heather Bicknell: Yeah, we'll do.
Ryan Purvis: Super. Thanks. Talk to you later.
Heather Bicknell: Have a good weekend.
Ryan Purvis: You too.
Bye.
Thank you for listening to today's episode. Heather Beckner is our producer editor. Thank you, Heather. For your hard work on this episode, please subscribe to the series and rate us on iTunes at the Google Play Store.
Follow us on Twitter at the Dww Podcast. The show notes and transcripts will be available on the website, www.digital workspace works. Please also visit our website, www.digital workspace works. And lastly, if you found this episode useful, please share with your friends or colleagues.