Your AI Agents Need Checks, Not Blind Trust

[00:04] Dave Pengelley: All right, so uh Richard's often interrogating us on what we're wearing, but what is on your shirt this morning, Richard? [00:11] Richard Webbe: A [00:11] Dave Pengelley: Akka [00:11] Richard Webbe: C D [00:11] Dave Pengelley: Daka. [00:11] Richard Webbe: C I went to the A C D C concert with my son, who's [00:15] Matt: This [00:16] Richard Webbe: only [00:16] Dave Pengelley: They [00:16] Matt: morning [00:16] Richard Webbe: eleven, who loved [00:17] Dave Pengelley: still [00:17] Richard Webbe: it. [00:17] Dave Pengelley: touring. [00:19] Richard Webbe: Well, it might have been last year, but hey. I bought the t shirts. [00:24] Dave Pengelley: Nice, nice. [00:25] Matt: very [00:25] Dave Pengelley: I saw [00:25] Matt: cool. [00:26] Dave Pengelley: them uh live. I saw them twice in Perth. Twenty plus Over 20 years ago. I think the ball breaker tour. [00:32] Richard Webbe: Yep. [00:33] Dave Pengelley: Um, was it just the one? Yeah, I saw them once, I think, because I saw a kiss a few times. I've seen Kiss a handful of times, but Akadaka was one of the first big rock concerts I went to do with my dad as a teenager, and I think there was the ball breaker tour. [00:47] Richard Webbe: Well, [00:47] Dave Pengelley: Wrecking [00:47] Richard Webbe: I went to the [00:48] Dave Pengelley: ball. [00:48] Richard Webbe: I went to the uh Back in Black first concert at the Maya Music Bowl, which was uh historic, [00:56] Dave Pengelley: Yeah. [00:56] Richard Webbe: and uh that song Rock and Roll Ain't Noise Pollution. Was uh turned up to a 12 because it was all the ladies in Turak who complained that the last time they played there, which was 10 or 15 years earlier, before Bon Scott died, they said we don't need this noise pollution. And of course, so they wrote the song Rock and Roll Ain't Noise Pollution, and boy did they pump it out. And if you've ever been to the Sydney Meyer music bowl, there's all these trees around. And I was explaining to my kids the other day when we were walking through there that we didn't get tickets, so we all climbed up into the trees. If you've ever listened to Holly August night, Neil Diamond always acknowledged the uh the tree people. So we're all sitting up in these trees watching the biggest rock concert of the decade at the time. It was just fabulous. A few people did fall out of trees and there were branches broken, but hey, it was fun. [01:45] Dave Pengelley: The eighties, different era, simpler times. [01:50] Richard Webbe: Much, much simpler times, mate. [01:53] Dave Pengelley: With [01:53] Richard Webbe: Oh, [01:53] Dave Pengelley: uh some excellent music carrying out of the seventies and into the eighties. [01:57] Richard Webbe: I think music represents the sentiment. And the emotion of the community at the time. [02:03] Dave Pengelley: Yeah, that doesn't say much for modern music, does it? [02:06] Richard Webbe: No, it's crap. [02:07] Dave Pengelley: Yes. Yes. It's one of these things that I'm I'm working on my games and trying to work out what I'm gonna do for score and music and background and and what I AI gen can do versus and when I say AI gen, I want to actually get it to use MIDI tools to like do proper synth and actually use notes and not just do some random generator thing. And um I saw I can never remember which composer it is. It was like Bark or someone had this game where you combine chord sequences in any order and you can basically make like a quatrillion iterations of songs that make sense because of the maths of music. Um I've got to look that up and see whether I can maybe bring that into [02:46] Matt: Yeah, that'd [02:46] Dave Pengelley: um [02:46] Matt: be cool. [02:47] Dave Pengelley: a synthesizer using the powers of [02:50] Matt: Have you ever [02:50] Dave Pengelley: artificial [02:50] Matt: seen [02:51] Dave Pengelley: intelligence. [02:51] Matt: have you ever seen Strudel? [02:53] Dave Pengelley: Strudel. [02:54] Matt: I think it's called [02:54] Dave Pengelley: No. [02:55] Matt: Strudel. Um it's like a it's a programming language that You you write like you write [03:01] Dave Pengelley: Hmm. [03:01] Matt: it out like code and it actually creates music. It's [03:05] Dave Pengelley: Hmm. [03:05] Matt: very, very cool. I I reckon [03:07] Richard Webbe: That [03:07] Matt: agents [03:07] Richard Webbe: is cool. [03:07] Matt: would could learn that and um mess around with that. [03:11] Dave Pengelley: Yeah, because even for sound effects, I was looking at like um do I need to use Audacity or can I make my own custom tool that does a lot of the just specific effects and and um audio changes? And I was gonna make my own one and then found out someone beat me to it and there's already some really good ones out um six months ago. So it's good. It's amazing time, lots of tools. Lots of technology and lots of podcasts. [03:35] Dave Pengelley: Actually, [03:46] Dave Pengelley: are there lots of podcasts? I think when I I'm surprised by the YouTube algorithm, I shouldn't be, but I am. I mean, I watch all of our content on a separate account, like end to end. Like every podcast, every yeah. Interview I do, even though I'm the one doing it, even though I'm the one hosting it, even though I'm the one who recorded the field note, I'll listen to it end-to-end just to A, make sure that I'm not putting out complete rubbish. And B, if anything, jumps out where I'm like, oh, that could be useful for later. And C, because it's just good for the stats to actually have someone watch something from the start to end. Um, but um, I don't get other podcasts like ours. I get a few of the hype merchants, your Alex Finns and stuff popping up. Because they talk about grockbots and things. But I'm not seeing other just non hype podcast content. [04:37] Matt: Yeah. [04:37] Dave Pengelley: Which is weird. Because as soon as I watch a few videos on game dev, I get heaps of game dev content and from game dev people. If I watch a few things on the news, I get heaps of political news and stuff popping up. But I watch this and get no other similar content. [04:51] Matt: Hmm. Maybe that's because there's like we don't have enough out there to to draw the algorithm towards us to other things. [05:00] Dave Pengelley: Maybe there's just not enough similar content, maybe we are unique. [05:03] Richard Webbe: We [05:03] Matt: I [05:03] Richard Webbe: are [05:03] Matt: think [05:03] Richard Webbe: unique. [05:04] Matt: we're pretty [05:04] Richard Webbe: I've [05:04] Matt: name. [05:04] Richard Webbe: actually been invited as a guest onto a couple of other podcasts on the strength of our podcast. And I did one the other day. And I have to say, we are much more fun. [05:16] Matt: Fun. What do you mean? [05:18] Richard Webbe: Because we're like that. We're just crazy. We're fun. We talk about stuff. And I reckon we talk about stuff that people would really care about rather than the sanctimonious, sterilized business. eco chamber of AI which is becoming boring. [05:33] Dave Pengelley: I mean you you were on a f uh you said you were on another podcast and they're telling them that they're sanctimonious and sterile. Uh that that aside. [05:40] Richard Webbe: Oh, are we live? [05:43] Dave Pengelley: Were they well, I mean, were you on another panel show, or was it a one-to-one interview, [05:46] Richard Webbe: No, [05:46] Dave Pengelley: fireside [05:47] Richard Webbe: no, [05:47] Dave Pengelley: chat [05:47] Richard Webbe: no. [05:47] Dave Pengelley: thing like [05:47] Richard Webbe: No. [05:48] Dave Pengelley: Darren? [05:48] Richard Webbe: These other podcasts were quite good. As you know, my friend Darren who's in sales um [05:52] Dave Pengelley: Mm-hmm. [05:52] Richard Webbe: did it and this um there are a couple of others that I've done. I've got another invite to do another one. So clearly our content is resonant. resonating with people and I didn't mean to sound um but they're very focused on pure business outcomes and it was a very controlled thing whereas I love our podcast because we're so relaxed and irreverent and we just say shit that we want to say. [06:14] Dave Pengelley: I was I was on one recently. Um it finally got published. I was on it like three months ago, but they had a server crash and lost half their recordings and like [06:22] Richard Webbe: Oh [06:22] Dave Pengelley: they thought [06:22] Richard Webbe: dear [06:22] Dave Pengelley: it was gone forever, but they went, Oh, we've recovered it and we just published it. I was like, Oh, thank you. Um, and that was a very different style because he had he had a long list. questions and he was just like, What about this? And tell me about this. And what about this? And I felt almost like interrogated, but it like and just took a different turn than I expected. I thought it was SEO O lab. We were gonna talk about using AI for websites. And he ended up talking about sales leadership and CRMs and all my Salesforce experience. Um which was which was great. I mean it was like good good reminder for me. Like hey I know all this sales stuff. Like I I was immersed in this for over a decade and I don't get to talk about it very often but I I do actually um I picked up a few things in my time at the [07:02] Richard Webbe: That's [07:02] Dave Pengelley: world's [07:02] Richard Webbe: what a [07:02] Dave Pengelley: leading [07:03] Richard Webbe: good interviewer [07:03] Dave Pengelley: CRM. [07:03] Richard Webbe: does, they hijack the attendee so they give honest and clear, unprepared answers. [07:08] Dave Pengelley: Well, yeah, and I mean he wanted to know more. I think it's always that selfish thing as a interviewer sometimes. Like, what can I get out of this? I mean, I find when I do the operator stories, they're quite different because I don't have an agenda. In fact, arguably some of the people we're interviewing could be, you know, competitors for AI consulting and deals, especially if they're local based. But I'm like, there's enough pie here, rising tides. lift all ships and I'm just genuinely curious to hear what they're doing, how they're doing it, what tools they're doing. I I spoke to one guy this morning uh who's gonna be uh on will publish it next week and he's built his own custom harness like a whole like company co-worker harness with a flat price per user all usage included and he's just built skills and devs and stuff and like he's built a full you know [08:00] Dave Pengelley: Thing. So yeah, that seems really good and sounds like really good. But I got to, you know, challenge him and go, what does this mean for security? And how do you do that? How do you do that? And not to catch him out because I just assumed he had good answers and he did. And so that's one of those things. I've got I'm not shilling any specific person, but I'm just curious around why did they go down that path and what did they do that for? And where did this come from? And uh it's that's that's interesting. I mean, the guy that we um the operator store we launched this week, Clive uh Mayhu, who Who is the CEO, co-founder of golf.ai and fellow author such as yourself, Richard. He's just launching his book this week, Vertical AI. He tells me very proudly in the interview that he saw the possibilities of using AI in the golf context and went and spent a quarter of a million dollars on the domain name golf.ai. [08:52] Matt: Amazing. [08:54] Dave Pengelley: Like he's you you might have run past him before, Richard. He was the uh managing director for NetStur. Australasia back in the nineties. [09:01] Richard Webbe: Oh, what's [09:02] Dave Pengelley: You [09:02] Richard Webbe: his name? [09:03] Dave Pengelley: Clive Mayhew, you guys are similar age, [09:05] Richard Webbe: Oh, [09:05] Dave Pengelley: so [09:05] Richard Webbe: that [09:05] Dave Pengelley: you [09:06] Richard Webbe: name, [09:06] Dave Pengelley: probably [09:06] Richard Webbe: that name does ring a bell. [09:07] Dave Pengelley: you you [09:08] Richard Webbe: I [09:08] Dave Pengelley: probably cross [09:08] Richard Webbe: tell you [09:08] Dave Pengelley: paths [09:09] Richard Webbe: some. [09:09] Dave Pengelley: or run in similar circles. [09:10] Richard Webbe: Yeah, yeah, yeah. Well, he would have been working for Mr. Andreessen in those days. [09:14] Dave Pengelley: Yeah. [09:15] Richard Webbe: And um there was a few multi-billionaires and then multi-millionaires lost in those days. Like I remember a few of them, my friends who were in startups, and [09:24] Dave Pengelley: Mm-hmm. [09:24] Richard Webbe: they look at the paper and they say, Oh, Richard, I'm worth a hundred million dollars now. I go, well done. And then of course the dot-com. crash came and they hadn't vested their shares so they couldn't sell them and they'd come back to me and go, hey Richard, if you've got a job, I'm no longer worth a hundred million dollars [09:38] Dave Pengelley: Well, [09:38] Richard Webbe: anymore. [09:38] Dave Pengelley: it it's uh it seems seems Clive through his entrepreneurial pursuits has done fairly well for himself if he [09:43] Richard Webbe: I [09:43] Dave Pengelley: can drop [09:43] Richard Webbe: think [09:43] Dave Pengelley: two [09:43] Richard Webbe: you [09:43] Dave Pengelley: hundred and [09:43] Richard Webbe: might [09:44] Dave Pengelley: fifty [09:44] Richard Webbe: have. [09:44] Dave Pengelley: K on a domain. [09:45] Richard Webbe: Yes, I think [09:45] Dave Pengelley: Um [09:45] Richard Webbe: you might have. Actually, you know, it's interesting, you know, we're here to talk about our AI and I'm not derailing like I usually do. But [09:52] Dave Pengelley: Thank [09:52] Richard Webbe: on [09:52] Dave Pengelley: you. [09:52] Richard Webbe: that point, um we had the Melbourne Cup for those calling from watching from overseas and interstate yesterday. which always attracts a lot of information [10:02] Dave Pengelley: I [10:02] Richard Webbe: and [10:02] Dave Pengelley: forgot [10:02] Richard Webbe: a little [10:02] Dave Pengelley: it was on. [10:03] Matt: I [10:03] Dave Pengelley: I [10:03] Matt: didn't [10:03] Dave Pengelley: didn't [10:04] Matt: even [10:04] Richard Webbe: eh [10:04] Matt: know it [10:04] Dave Pengelley: even [10:04] Matt: was a [10:04] Dave Pengelley: know. [10:04] Matt: bad [10:04] Dave Pengelley: Like it's just bit not working in an office at the moment around other people. Didn't even know it happened. [10:12] Richard Webbe: okay so it did happen [10:14] Dave Pengelley: And [10:14] Richard Webbe: um [10:14] Dave Pengelley: I bet a horse won. I mean, a person riding a horse won. I bet the the person who trained the horse [10:19] Matt: change. [10:20] Dave Pengelley: that the person was riding on one. [10:22] Richard Webbe: yes all of them got accolades exactly in that order as you say but what was interesting i remember thinking to myself now what's mr and miss A six-pack or the average pedestrian going to do, they're going to get their LLM and say, who's going to win the Melbourne Cup? And then place their bet like that. And I had some friends with me and I wanted to prove a point to them. And I said, and they said, Richard, are you going to use your AI magic to find out who's going to win? I said, Yes, I am, but most people will and they'll get it wrong. And they didn't understand. So I proved the point. And I entered in and I picked the winner using my LLM. [10:59] Matt: So you're [11:00] Richard Webbe: And I sorry. [11:01] Matt: So you won millions. [11:03] Richard Webbe: No, no, I didn't bet that much, so I won $85, but [11:09] Richard Webbe: it's only five dollars each way. Uh, and it was point of law, and it was a great win. Um, and it was amazing, but it it wasn't you know the one of the top favorites, [11:17] Dave Pengelley: What what [11:17] Richard Webbe: but it became [11:18] Dave Pengelley: was your magic to to to scamming the horse, uh the horse books? [11:21] Richard Webbe: that's the question I wanted you to ask me. Yes. So what I did is I had to have knowledge about horse racing. Now I used to be a member at Flemington. I'm not Big horse racing person, but I used to hang out with people who were big horse racing people. So I started to enter into my LLM all the variables that I could think of, like tons of them, like even little tiny things, right? And I won't go, but I entered them all in, and as I entered more and more variables in, the order of winning changed. And then I added even more variables. What if it rained more than an hour or less than an hour before the race? And it was only a few millimeters of rain. So now AI is applying its understanding of the weather to the track and how thick it is and what the weighting is. And it came out and told me that point of law would be your best choice. And I picked it and won. Now I was on the phone just as I was coming to do this podcast today to a very authoritative IT and AI person. And they did the same thing in front of their family and consequently lost dramatically, which proves my point. It's not about the answer, it's about the knowledge of the variables and the questions. [12:36] Dave Pengelley: Horse racing is one of those things where there are a lot of variables and there is statistical probabilities engaged in there if you give it the right things. I remember 30 plus years ago, a guy came to my dad. My dad, for many, many years, I don't know what the status of the project is now, but this guy would go to the track and he would get a copy of the results and the data on a three and a half inch floppy. And he would go and load that in. And my dad built this whole like Like access database thing, because this guy had come up with spreadsheets and formulas and algorithms that he [13:04] Matt: I think [13:04] Dave Pengelley: thought [13:04] Matt: it's only [13:04] Dave Pengelley: would predict race [13:05] Matt: there's [13:05] Dave Pengelley: results. And so for deck [13:06] Matt: no [13:07] Dave Pengelley: for decades, this guy was [13:08] Matt: problem. [13:08] Dave Pengelley: tweaking them, manipulating them. Obviously, then the data moved online. And so we had to update the program to pull the online data. But this guy, I think, you know, I don't think he ever made billions off it, but he was making, you know, enough to live off [13:21] Richard Webbe: Yeah, [13:21] Dave Pengelley: through the horses [13:22] Richard Webbe: good. [13:22] Dave Pengelley: because he he'd worked out the statistical probabilities and kept tweaking it and massaging it. And it was about having enough data points and which Which data points got weighted against which thing and which scenarios and which horse on which track surface on which temperature and all the things, right? [13:38] Richard Webbe: Yeah, I and I entered in some outrageous detail that I think helped. Everything from how many races had they won earlier in the day to how tired the jockey might be to the colour of their silk to how often they'd ridden that parti ridden that particular horse to I I just got as much variables I could and put it in. And it said very convincingly that point of law will win. [14:02] Matt: I got a question. [14:04] Richard Webbe: Yep. [14:05] Matt: So what happens when the ability to pull that off is not a big deal? [14:14] Richard Webbe: That is the billion dollar question. And that's what I was saying in one of my articles a while ago. And we've spoken about that that the probability now reduces. Now, um, who's our mate that started, I always forget his name, OpenAI. Guys know his name. [14:31] Dave Pengelley: Almond. [14:31] Matt: Your man. [14:32] Richard Webbe: Sorry? [14:33] Dave Pengelley: What button? Same Alman. [14:35] Richard Webbe: Yeah, yeah, Altman. And I believe it was him. I was reading, um, I believe it was him, and he was talking about how, like I think I wrote my article a f a few months ago, that facts will no longer be a mystery. Information is no longer a mystery. So the whole dynamic of business is going to change. So the people that are going to get hurt are lawyers and accountants and stuff like that, because really their knowledge of the facts is. What they trade on, but their thought leadership is not what they trade on, but that's what they should be trading on. And Sam Altman believes, uh, like with Elon, in the new world economy, that trust is going to take over from knowledge. And with a world of AI that is trying to create as much fake news as possible, this is going to create a very new and interesting paradigm. [15:26] Dave Pengelley: If people use it. [15:29] Dave Pengelley: Because [15:29] Richard Webbe: Yep. [15:29] Dave Pengelley: I mean I I this is not a political podcast, but I just saw a clip around the midterms and some of the stuff that's been put out around various candidates. And there's still so much clickbait nonsense of fake quotes that the and then the this this particular presenter was saying, here's a fake quote of something that this this particular candidate never said, but it's attached with a piece of video clip. Most people won't watch the video clip. They'll just see a video clip and go, oh, there's the evidence it happened, and there's the misquote that never happened. But they'll they'll go two and two and go, yep, I read there's a quote. And I saw there was a video, I didn't have to watch the video. That quote's damning. I'm making my choice and voting based on you know complete falsehoods. [16:08] Richard Webbe: Yep. [16:08] Dave Pengelley: Um, so until you you know that that means then people like Elon need to have the AI looking at every video, looking at every claim and every post, comparing the two. That's possible, but someone's gonna have to pay for those tokens. Um [16:23] Richard Webbe: Oh, yeah, look, I I think that's at the tactical practical level, but when we think about it at at a bit bigger of a stretch. Strategic problem when radio came along. I think of the War of the World. Right? So when our friend, um God, I've forgotten his name, the greatest actor and uh uh voiceover man in history uh who used to advertise Xerox. Um you'll remember the beard. [16:49] Dave Pengelley: Joe North. [16:50] Richard Webbe: No, no, no. Um, I'll remember his name in a second because this is embarrassing because I'm a big fan. Um, and he produced one of the most definitive movies about a personality. Um, he read The War of the Worlds on radio, and about 30% of the audience for in that part of the US thought that the US was actually being attacked by aliens. They had to put special messages out, uh, stop people losing their mind, uh, and because people weren't quite used to the imagery of the radio and a very well-produced radio show that made it all seem real. So, what's gonna have to happen, and Alvin Toffler used to write about this, [17:27] Matt: Was it [17:27] Richard Webbe: uh [17:27] Matt: also in Wells? [17:28] Richard Webbe: that's [17:28] Matt: Sorry. [17:29] Richard Webbe: it awesome well. Thank you very much. [17:30] Dave Pengelley: Which [17:30] Richard Webbe: I should [17:30] Dave Pengelley: is [17:31] Richard Webbe: have [17:31] Dave Pengelley: funny [17:31] Richard Webbe: known [17:31] Dave Pengelley: because [17:31] Richard Webbe: that. [17:31] Dave Pengelley: because it was adapted by the the novel by HG Wells with different [17:35] Richard Webbe: That's [17:35] Dave Pengelley: wells, [17:35] Richard Webbe: exactly [17:35] Dave Pengelley: different spelling. [17:36] Richard Webbe: very true. [17:36] Matt: His [17:37] Richard Webbe: And he read [17:37] Matt: his voice [17:37] Richard Webbe: it. [17:38] Matt: is creepy as I [17:39] Richard Webbe: Oh, [17:39] Matt: like [17:39] Richard Webbe: it's [17:40] Matt: it. [17:40] Richard Webbe: brilliant. It's brilliant. If you ever want to entertain yourself one day, put on More of the Worlds by Austin Wells. You'll get it free somewhere and listen to it as you go to sleep. It is utterly unbelievably brilliant. [17:53] Dave Pengelley: Listen to the Jeff Wayne version. [17:55] Richard Webbe: Ah, okay. [18:00] Matt: Wrote out a lot of uh sorry, read out a lot of HP Lovecraft. Like the um the Cthulhu, uh, you know, the that rule anthropic kind of horror stuff. Um the the the one there's such a famous line where he's the guy's like descending into madness and he he he's talking about his experiences, you know, hearing things through the walls and and that kind of stuff. I I don't know, I'm just I'm picking up fragments of memory. I'm gonna have to [18:31] Richard Webbe: I'll [18:31] Matt: dig dig [18:31] Richard Webbe: do [18:32] Matt: into [18:32] Richard Webbe: that. [18:32] Matt: that after this. [18:32] Richard Webbe: Keep throwing them up on the screen. It's good. It's good. So, yes, so um, so what I'm saying is about the War of the Worlds thing had the everyone fooled. Uh, and it was meant to be just entertainment. So, what we're going to have to do, and Alvin Toffler and others wrote it, we need to educate ourselves as a humanity, not through compliance and restrictive practices, but just knowledge to be a discerning human and knowing when to trust and not to trust. My grandfather always said to me, Richard, believe. Leave a quarter of what you read, a quarter of what you hear, and half what you see. And that was probably relevant to his time when he was growing up. There was a famous story about when they first brought the video evidence into sporting. And there was a particular AFL situation, and the one video that captured this guy, it looked like a player had just completely punched and pollaxed a guy in the face. And on the face of that evidence, he should have been banned for life and damned to hell. But the umpire who was there, Saw that someone had run literally run into this guy's elbow, so that his hand came forward and hit the other guy, right? Um, yeah, and it was just the way the camera corded, the way it happened, it was bizarre. Uh, and if he hadn't had the umpire on his side, he would have been banned for life. So we need to educate ourselves as humans, like when we see a message and we go, there's something weird about this message. Be careful. [19:55] Dave Pengelley: That makes me think, and I I don't remember the details of it, but during COVID, there was that scene where it looked like someone punched a horse. in Melbourne. Apparently they didn't they didn't actually punch a horse, but there was this this news clip where it looked like someone fully just tried [20:07] Richard Webbe: That [20:07] Dave Pengelley: to [20:07] Richard Webbe: was [20:07] Dave Pengelley: haymaker [20:08] Richard Webbe: during the COVID [20:08] Dave Pengelley: a horse. [20:09] Richard Webbe: protests, and there was a uh I do remember a suspicious video that looked like someone punched a horse. [20:14] Dave Pengelley: Yeah, like [20:15] Richard Webbe: And [20:15] Dave Pengelley: it [20:15] Richard Webbe: I [20:15] Dave Pengelley: fully [20:16] Richard Webbe: I [20:16] Dave Pengelley: looked like this guy just gave this horse a big haymaker [20:18] Richard Webbe: yeah, [20:18] Dave Pengelley: and a big [20:18] Richard Webbe: I [20:18] Dave Pengelley: right [20:18] Richard Webbe: can't [20:19] Dave Pengelley: hander. [20:19] Richard Webbe: remember. I think he was swinging at the horse, and the police officers pulled the horse away at the same time, and the angle looks like Mungo out of Blazing Saddos. [20:27] Dave Pengelley: Yeah, [20:28] Richard Webbe: Does anyone remember [20:28] Dave Pengelley: it [20:28] Richard Webbe: Mungo [20:28] Dave Pengelley: looks [20:29] Richard Webbe: and [20:29] Dave Pengelley: exactly [20:29] Richard Webbe: Blazing [20:29] Dave Pengelley: like [20:29] Richard Webbe: Saddam? [20:29] Dave Pengelley: that. Yeah. Um it's interesting you talk about news and so on and and headlines and stuff. The uh have you heard of the Gelman uh amnesia effect? [20:39] Richard Webbe: Do you know what? I have, but reinvigorate us on what it means. I have actually heard about that recently. Please [20:45] Matt: I'm [20:45] Richard Webbe: say. [20:45] Matt: I'm feeling amnesia. I must have the effect. [20:49] Dave Pengelley: So so so the idea is like say you're you open the newspaper or you're watching watching the news and there's something about the latest AI models, and it's completely wrong. So you you just Called total BS on it. You're like, [21:01] Richard Webbe: Yeah. [21:02] Dave Pengelley: that's not how that works. Why are they using those words? That's not how it fits. Even like um, like I watched 60 Minutes, the their thing about the job ageism, and they were completely over-indexing, blaming AI for every job applicant ever getting rejected. I'm like, there's so much more than just AI. And again, it's that whole let's demonize and make AI evil. Anyway, whatever it is, you guys are experts in AI or business and sales, and you see a news article and it's just complete bollocks. Then that that segment. Finishes, they go to the next segment, which is, you know, about biology of an elephant disease in Africa that you know nothing about. And you just go, Wow, that's really sad about the elephant disease in Africa. That must be all be 100% true. And you can forget that three seconds ago, they were like spouting complete bollocks about something you did know about. So why can you trust the elephant thing? Because that might be just as much bollocks. [21:54] Richard Webbe: Same [21:54] Dave Pengelley: And it's this [21:55] Richard Webbe: sources [21:55] Dave Pengelley: thing [21:55] Richard Webbe: here. [21:56] Dave Pengelley: that when you're watching news, you'll criticize it. An area you know about, but as soon as it's unfamiliar, you just trust it again, and you forget that they were just spouting BS three seconds ago. This [22:07] Matt: That's [22:07] Dave Pengelley: so when [22:07] Matt: fascinating. [22:07] Dave Pengelley: it yeah, and so with any influencer, with any news thing that you see, like it's that consistency. How often do you trust them? And are they actually proven to be trustworthy, or are they two-faced? And just because they're, you know, broken clocks right twice a day, you got to be careful um what [22:25] Richard Webbe: Very [22:25] Dave Pengelley: you [22:25] Richard Webbe: good. [22:25] Dave Pengelley: trust and [22:26] Richard Webbe: Very [22:26] Dave Pengelley: engage [22:26] Richard Webbe: good colour. [22:26] Dave Pengelley: that critical thinking all the time. [22:28] Matt: How do you how do you Establish it here's a big question because like it's very easy to say that kind of stuff and to pull out those kinds of observations, but how do you actually achieve trust? [22:41] Richard Webbe: Ah, trust is um it's the thing that values brands and people the most. Trust. And it's based on the fact that you will do what you say you do or what you said was true. So it needs evidence [22:54] Matt: I think that's [22:55] Richard Webbe: and [22:55] Matt: integrity [22:55] Richard Webbe: it happens. [22:55] Matt: though. I'm not sure that that's [22:57] Richard Webbe: Now [22:57] Matt: trust. [22:57] Richard Webbe: the person delivering it's integrity, trust. It's the person who's receiving it. [23:02] Matt: Right. Like [23:02] Dave Pengelley: So [23:03] Matt: accepting [23:03] Dave Pengelley: I um [23:03] Matt: integrity. [23:04] Dave Pengelley: I did corporate training many years ago uh by the Kobe organization, and it was the speed of trust was the uh workshop, and they answered the exact question, uh Matt, around the foundations of trust. And there you go, integrity at the bottom. So there's the below the surface things that people can't really see, which is like the roots. Um that's your integrity. Uh, and you know, are you whole and congruent? I suppose, and what's your intent, your agenda? And then up the top is the leaves and the branches that everyone can see. And that's your sort of competence. So it's the this stacking thing where it starts with sort of the underlying character of a person and then their competence then builds on that trust and stuff. And obviously this particular website wants to sell me things. But um [23:49] Matt: That's cool. Can you send me that like a screenshot of that? [23:52] Richard Webbe: That's [23:52] Matt: Um, 'cause [23:52] Richard Webbe: great. [23:53] Matt: I would like to do some writing on that. [23:56] Dave Pengelley: Yes, I will I'll I'll I'll send you the the hyperlink. Through [24:00] Richard Webbe: So [24:00] Dave Pengelley: this whole [24:00] Richard Webbe: I [24:00] Dave Pengelley: slot, [24:00] Richard Webbe: just [24:01] Dave Pengelley: so you can [24:01] Richard Webbe: a good [24:01] Dave Pengelley: read [24:02] Richard Webbe: example, [24:02] Dave Pengelley: the thing. [24:02] Richard Webbe: Matt, is I just told you that I used AI to find the winner of the Melbourne Cup. Now, it could have been 99.99% luck, like Dave says. The clock is right twice a day. And so I could have just hit that time of the day when it was right. But I did add some variables and some data in there, excuse me, that might have made some sense, right? Stop laughing at me, Dave. I always get it. Um [24:30] Matt: What kind of phone do you have, Richard? [24:32] Richard Webbe: I have an iPhone. [24:33] Matt: Does it have the switch on the side of it? [24:37] Richard Webbe: That's a very good question. [24:41] Richard Webbe: He's roasting me now. My trust is now at the door. Everything's gone. [24:46] Matt: No, [24:46] Dave Pengelley: I [24:46] Matt: it's [24:46] Richard Webbe: It [24:46] Matt: a genuine [24:47] Richard Webbe: does [24:47] Matt: question. [24:47] Richard Webbe: have the [24:47] Matt: I'm [24:47] Richard Webbe: little [24:48] Matt: a [24:48] Richard Webbe: button. [24:48] Matt: I'm [24:48] Richard Webbe: However, [24:48] Matt: a problem [24:49] Richard Webbe: for [24:49] Matt: solver. [24:49] Richard Webbe: some reason on my phone, [24:51] Dave Pengelley: that's [24:51] Richard Webbe: it doesn't [24:51] Dave Pengelley: a button. [24:52] Richard Webbe: work. [24:52] Dave Pengelley: That's [24:52] Richard Webbe: It [24:52] Dave Pengelley: it's the [24:53] Richard Webbe: doesn't [24:53] Dave Pengelley: new the this [24:53] Richard Webbe: work. [24:53] Dave Pengelley: new button version. I've got I've got the switch, the physical switch, because I've got an older iPhone thirteen dinosaur and [25:00] Matt: Well. [25:00] Dave Pengelley: I uh I got the Pro Max and I hate it because it's like a freaking brick. It just breaks [25:04] Matt: Oh, [25:04] Dave Pengelley: my wrist [25:04] Matt: mine's [25:04] Dave Pengelley: off. Like [25:05] Matt: the [25:05] Dave Pengelley: it's [25:05] Matt: opposite [25:05] Dave Pengelley: just [25:06] Matt: size. [25:06] Dave Pengelley: oh [25:07] Matt: Mine's [25:07] Dave Pengelley: mine's [25:07] Richard Webbe: So, [25:08] Dave Pengelley: just [25:08] Matt: mine's the meaning. [25:09] Dave Pengelley: mine's [25:09] Matt: Check [25:09] Dave Pengelley: so [25:09] Matt: this [25:09] Dave Pengelley: big. [25:10] Matt: boy [25:10] Dave Pengelley: Look at [25:10] Matt: out. [25:10] Dave Pengelley: look at the size of it. Like I've got a big head and like it's massive. Anyway. [25:13] Matt: Yeah. [25:14] Dave Pengelley: Um the [25:16] Richard Webbe: know, you it's a lot of people. [25:16] Dave Pengelley: I was playing a game or something. I wanted to hear some audio and had to take it off mute the other day, which is never off mute. And then someone called me and I was like, what the heck is that? Like, I'm I haven't heard my phone ring in a decade. That's bizarre. [25:28] Matt: Yeah. [25:29] Dave Pengelley: Um Yes, anyway. [25:31] Matt: Very [25:31] Dave Pengelley: Uh [25:31] Matt: good. [25:33] Dave Pengelley: now now now that we're completely de round, let's let's do a thing. [25:44] Richard Webbe: Melbourne Cup. [25:48] Dave Pengelley: I thought [25:48] Matt: That [25:48] Dave Pengelley: I'd [25:48] Matt: um that that intro sound sounds a bit sus. It's kinda like [25:53] Richard Webbe: Yeah. [25:54] Dave Pengelley: To be fair, that's the first time I watched it completely because Ruby made it for me as I was connecting into this meeting. So I'd never actually watched the new uh Winds of the Week bumper. [26:02] Matt: Still [26:03] Dave Pengelley: But I [26:03] Matt: pretty [26:03] Dave Pengelley: thought [26:03] Matt: cool. [26:04] Richard Webbe: All right. [26:04] Dave Pengelley: Um Yes, I thought, you know, I'll just we're time for a new on brand bumper. I don't know, what what what does the audience think? How does that compare to the legacy classic? [26:14] Matt: That's [26:15] Dave Pengelley: I [26:15] Matt: a [26:15] Dave Pengelley: think [26:15] Matt: good question. [26:24] Matt: Because I think the legacy one is more cool, but it's also possibly more slop. [26:30] Dave Pengelley: Oh, it's heaps, it's heaps more slop. I mean, that's why I left the double the week in the new road, the new one as [26:35] Matt: Yeah. [26:35] Dave Pengelley: a bit of a wink and a nod to our legacy. I think the new one needs to be a bit longer. I think it's just too short. [26:41] Richard Webbe: I'm gonna go [26:41] Dave Pengelley: Despite [26:41] Richard Webbe: look at my [26:42] Dave Pengelley: the [26:42] Richard Webbe: nose. [26:42] Dave Pengelley: whoop noises, it's too short. Needs to be longer. Um [26:47] Matt: Very cool. [26:49] Dave Pengelley: yes. Wins of the week. Well, um, Richard is uh gonna leave us because he's gonna go make millions on the horses, betting on the ponies. Spring carnival time. Never been a bit of time, Richard. [27:04] Dave Pengelley: What about you, Matt? And he wins all of the week this week. [27:07] Matt: Yeah, I've got lots. And it it's it's nothing as um monumental, like it's nothing new. It's just progression. It's the same thing that we've been that I've been doing for, you know, every week now. It's just updating my stack, updating my flow, updating the tools that I use, and refining consistently. So um and on that topic, you know, like there's so much value there for other people to be able to, you know, if they're fresh, they can jump in and and start learning from somebody who's kind of gone through and you know gotten a thousand cuts already. Um yeah, so I my win I guess is a mentality shift. It's the the the return to the ability to actually go to content and actually start trying to to preach some of this stuff again. Out of this box. Um, so yeah, I'm gonna I'm gonna be re putting in a lot more attention into my school again moving forward, um, as well as like some actual sort of YouTube content, um trying to post more in X, and then along with that, it's gonna come with some LinkedIn stuff too, because why not? Um, so yeah, and how I've done this, like this mentality shift, is literally driven by how I'm currently getting work done. Like I've not slowed down the amount of work. That I'm doing, it's definitely increasing. And that trust factor is part of this. It's like, how can I trust my systems to work fairly unattended? You know, how do I build up that trust? And then what does that actually mean from a systems engineering perspective? Um, and yeah, it's allowed my brain to kind of get a little bit more of a release and uh provide like a little bit more space. Like I've I've defragged the drive. Now and and there's a little bit more uh disk space in there. So yeah, that's that's probably my win. It is [29:06] Dave Pengelley: Nice. [29:06] Matt: just a bit of a workflow and mentality [29:08] Dave Pengelley: Because [29:08] Matt: shift. [29:08] Dave Pengelley: you've got the GrocBots managing the T3 coding harnesses and stuff, right? So you've got this sort [29:14] Matt: Yeah. [29:14] Dave Pengelley: of multi-layered delegation going on, which [29:17] Matt: Yeah. [29:17] Dave Pengelley: has given you easier ways to manage [29:20] Matt: That [29:20] Dave Pengelley: all. [29:20] Matt: they're they're my two tools. I I use [29:22] Dave Pengelley: Yeah. [29:22] Matt: T3 Code as my main sort of agentic interface, and then I've got Grockbot as my main sort of team interface. So like I was looking at hiring a while ago and I had this funny realization where what I'm currently achieving through a roster of Grokbot agents for 200 US dollars a month is like a tenth, you know, it's like 10% of the cost of what I would have been paying to bring on a junior or someone less junior to do less work. And it kind of blew my mind. mind for a moment and I'm like, okay, well, what does this actually mean if I'm going to continue to scale my individual company, my individual business? I'm actually going to build more human team members. What does that look like? Like do I just copy and paste my existing setup with them? And it's just a matter of each of us managing our own little sub-departments. Yeah, it's a fascinating thing. But um I've not achieved that reality yet. I don't really know what that looks like yet. But yeah, I'm very happy with where everything sits right now. [30:31] Richard Webbe: Excellent. Excellent. They talk about trust and trusting your systems, like the architecture you're creating, Matt, for you and some of your customers. They're saying that it it acts a lot like a biology system and an immune system. [30:46] Matt: Yeah. [30:46] Richard Webbe: And if you think like a biological system or immunity system, then it guides you on where to do your trust checks to make sure your systems are going okay. And they're saying that. human, I won't say human um uh lack of trust, or what's the word I'm looking for? Human suspicion is going to be a highly valued commodity, but not like that. It's like uh a problem-solving suspicion. I'm looking at something, it's working, I'm accepting what's coming in, but where could it go wrong? Kind of like in business, we used to run what we call, and I think Dave and I have spoken of this before, loss reviews, when you're working on a big project or a big deal. Before it goes wrong, you sit down and run a strategy on what went wrong before it happens. And then it goes, and I saw, has anyone seen the film World War Z? [31:41] Matt: I actually have nothing [31:42] Richard Webbe: It is worthwhile seeing, but there's a scene in there in the movie because Israel's borders were so strong, [31:49] Matt: to be [31:49] Richard Webbe: they're [31:49] Matt: there. [31:49] Richard Webbe: the only ones that weren't attacked by the zombie. And and Brad Pitt gets saved by his friend, who's a major spy in the Israeli Mossad, and he goes, Why am I? sitting here and why is most of Israel safe and the rest of the world's being attacked. He said, because we came up with the nine a 10-man philosophy. He said, what's that? He said, well, nine people will accept all the information that's happening and they'll prepare a plan to act on that. But one must always be voted in to be the contrarian and say, I do not believe that. And they would prepare a plan on the basis that it's not going to work. And he said, he goes to him, You're the ninth 10th man. He goes, yep. And I said the others have wood we're and that's why we're safe. [32:32] Matt: That's [32:33] Richard Webbe: It's an interesting, [32:33] Matt: cool. [32:34] Richard Webbe: cool concept, [32:35] Matt: I [32:35] Richard Webbe: isn't it? [32:35] Matt: resonate a lot with that. [32:39] Matt: I always tend to attack the status quo. You know, somebody runs that way, I run the other way. Um it it's this, it's just weird sort of natural tendency. I don't know if I'm that 10th man thing or or what it is. Sometimes it's I'll take a positive approach where everyone's being negative, or everyone's being hyper positive and like maybe blindsided, and I'll take the devil's advocate. Um one interesting thing, I don't I think I've told you guys about this before. There's a concept uh called TRIS, T-R-I-Z. And it's a Russian acronym for a Russian phrase that I'm not going to attempt to pronounce. And it translates to the theory of inventive problem solving. And when you're talking about the idea of security and trusting your systems being analogous to an immune system and a biological system. system that's basically TRIZ in effect where it's looking at an existing situation from the lens of an entirely different discipline. [33:41] Richard Webbe: Yeah. Like [33:42] Matt: Um [33:42] Richard Webbe: like white blood cells running around your body looking for trouble. [33:46] Matt: yeah yeah it's the same as like Red Hat agents looking for holes in your programs. [33:52] Richard Webbe: Fabulous. Oh, I like that analogy. [33:55] Matt: Yeah [33:55] Richard Webbe: True [33:55] Matt: that's [33:56] Richard Webbe: for that. [33:56] Matt: TRIZ [33:56] Richard Webbe: I'm going to [33:57] Matt: is [33:57] Richard Webbe: use [33:57] Matt: fantastic. [33:57] Richard Webbe: that. [33:58] Matt: Like for anyone who's actually like a Like a thinker, you know, in our audience, go down this rabbit hole. Just give yourself the permission because it it's a really, really cool, powerful thing. [34:08] Dave Pengelley: Why why would you not want to say this out loud, Matt? [34:18] Richard Webbe: You went you went [34:20] Matt: Also, [34:20] Richard Webbe: from Russian to you went from Russian to Welsh then, [34:22] Matt: I [34:22] Richard Webbe: and [34:23] Matt: need [34:23] Richard Webbe: I can [34:23] Matt: to [34:23] Richard Webbe: never [34:23] Matt: give [34:23] Richard Webbe: understand [34:23] Matt: credit. [34:24] Richard Webbe: Welsh. [34:24] Matt: I need to give credit where it's due. Rahul's actually in our chat right now from LinkedIn. He is the one that has has put me on to Triz um from from Blake's community on school. So yeah. [34:35] Dave Pengelley: Nice. [34:36] Matt: Runnels [34:36] Dave Pengelley: Well done, [34:37] Matt: and legend. [34:38] Dave Pengelley: well done, Rahul. And uh Phil is back in the comments talking about agents and manage agents and commenting here: the more you scale, the more distant you become from the process, you need more high-level checks and balances. Trust is [34:50] Richard Webbe: Yes. [34:50] Dave Pengelley: always a huge part of delegation, whether a human or AI. Yes, I've been having this chat with some people online and in the comments and stuff as well. And we touched on On this, I think last week, but the trend of the dark factory, and there's more people moving away from that. And I'm I'm in that category. I love the idea of the dark factory of just throw a giant speck at something and then boom out comes the other end, a little package with a bowl on it and a perfect software. And that might work for really well for some people. But for me, I don't think I'm good enough at knowing what I want at the start. I don't think I can I think I know what I want, and I'll write up a specification and get the agents to help me write up a specification. But then as soon as I see The first version, I'm like, ah, that actually is not what I want at all. That should be there. And I want this, and that doesn't fit like I thought it would. And for me, I'm more of an iterative builder and I need to see something, I think, in order for my brain to activate and connect the dots and go, ah. So I'm a bit of a visual learner in that aspect, which has always driven my wife insane. Because we you're talking about, let's let's remodel this room and let's work out what furniture we want and discuss, yeah, we could put a couch there, and if we had that table there and This and then we walk through Ikea. And as soon as I see some of the little showroom showcases, I'm like, oh, what if we did this and this and changed that and did that? And it's completely different. She's like, Why, where was this a week ago? I'm like, I didn't I couldn't see it. It didn't come to me until I was here in the spot. And then I saw the things. And like, like that, that sort of guy seeing all the maths and the algebra in front of him. [36:20] Matt: Yeah. [36:20] Dave Pengelley: Um, the the Gallifanaka's meme, I'm like, it just pops into split into shape. Um, and so [36:26] Matt: So [36:26] Dave Pengelley: for [36:26] Matt: we're [36:26] Dave Pengelley: me, [36:26] Matt: going to [36:27] Dave Pengelley: the whole Dark [36:27] Matt: go [36:27] Dave Pengelley: Factory [36:28] Matt: through [36:28] Dave Pengelley: doesn't work. [36:29] Matt: one. [36:29] Dave Pengelley: I need the Do that eater building. [36:30] Matt: I would look at it from like again, a systems engineering perspective, and and put yourself in that system. So you have now, let's say there's two elements, you know, the the iteration cycle that spits out a thousand ideas and escalates them back based on your previous decisions. And then you go through and you kind of like tinder, swipe left and swipe right on what you actually like and don't like. And then that system starts to learn your sort of parameters and your variables over time. And Yeah, [37:00] Dave Pengelley: The [37:00] Matt: uh, [37:01] Dave Pengelley: depends on the use case and the particular product. And I think, yeah, there's a use case for that. I I am and have spoken before about the drama I have when the AI gives me more work than I gave it. And I'm like, what the heck? How many decisions do I need to make now? Um, but I mean, at the moment, because I'm doing game development, that's hugely iterative. And because I'm not making AI slop stuff, I'm trying to actually craft some experiences that are meaningful and good. And like I said, I think we said last week, the AI can't understand. human taste, frame rates and what that means to the human eye when they see things pop up and movement and motion and feeling, they just don't have that. Um and so it really does need you to play it and go, ah, that didn't feel right. Ah, that didn't make sense. Oh, I wish I'd knew known what was happening when that thing happened. Like I know mechanically what the system's doing, but there was no visual feedback um from a user experience point of view. [37:52] Matt: I [37:52] Dave Pengelley: And [37:52] Matt: would [37:52] Dave Pengelley: so [37:52] Matt: love to know how Lauren went through and actually did that with um with the [37:59] Dave Pengelley: the The T Bow Dog. [38:01] Matt: Yeah, I just sent you the link again. Um, because yeah, she like she's probably like my biggest influence on the whole sort of hands-off factory, build trust in the system, self-verification, you know, build formal verification and allow the system to actually check and test itself based on the goals and the parameters in question. Well, [38:23] Dave Pengelley: Yeah. [38:23] Matt: is that your hand? [38:24] Dave Pengelley: No, no, that's that's that's the game. Um I did story mode, and then it actually Actually out trolls you and says, Why do you just kills you after a little while going story mode? There's not even challenge there. Go go play again. [38:38] Matt: It's very cool. [38:40] Dave Pengelley: Yeah. [38:40] Matt: I've only played it twice so far, but like this whole thing was literally just her cranking through with Groffba. [38:47] Dave Pengelley: I played it once on mobile, and it was a bit annoying on mobile because you didn't automatically pick up the things like the chests and stuff when you clicked on [38:54] Matt: The [38:54] Dave Pengelley: them. [38:54] Matt: music's [38:55] Dave Pengelley: Um, [38:55] Matt: pretty loud. [38:56] Dave Pengelley: I can't change that unfortunately. [39:00] Dave Pengelley: There we are, here we [39:01] Matt: Ah, [39:01] Dave Pengelley: are. Ah, [39:02] Matt: peace. [39:02] Dave Pengelley: phew. [39:05] Richard Webbe: I like [39:05] Dave Pengelley: Like [39:06] Richard Webbe: the uh use of AI for gamification of education. I think it's brilliant. [39:12] Matt: Speaking [39:12] Dave Pengelley: this is for [39:13] Matt: of [39:13] Dave Pengelley: trollefication. [39:13] Matt: it, I've got a really interesting thing about education we can talk about in a sec. I might not I might forget about it. But um yeah, like this kind of thing, from what I understand, like there's a lot of taste and you know human decisions that have gone into this that that Lawrence you know imposed, [39:33] Dave Pengelley: Yeah. [39:33] Matt: but I believe a lot of it is very, very self-driven. There's a there's another concept that she talks about a lot now called TLA or TLA Plus, um, which is another sort of like self-verification system similar to Bend. I can't remember if I told you guys about Bend previously. [39:50] Dave Pengelley: The the other thing with someone like that Tebow Dog is the vampire survivors like genre is pretty well established, and like I've one shot really Basic versions, not of that fidelity with that animations, but simple like prototype shooters with a character that runs around with different weapons that have different profiles that auto-kill everyone around them, with hordes of beasts coming. And I've one-shot that. And so to go from a one-shot, basic working prototype to then let's add some different sprites, let's cram all this like trolling Tebow story on it. Um, it's actually one of the easier genres of games to actually make for something like that. [40:30] Matt: Cool. I haven't done any game dev stuff yet. I uh too too deep in client work. [40:34] Dave Pengelley: Yeah. [40:35] Matt: I need a client that wants a game. [40:37] Dave Pengelley: Um [40:38] Richard Webbe: It's all again, man. It's all again. [40:40] Matt: Rahul, let's come together and make a Triz game. I think that'd be great. [40:46] Dave Pengelley: one of my wins was, and I I shared this in the field note, and so for those that watch the field notes, uh, you're gonna have seen this before. Um, but it's I close the browser tab. Here we go. Um, this library that Technium from um Hermes put up with all these different web libraries and little demos of all the different looks and feels and things. Um I think this is super cool because if you have an idea of a field you want, whether it's a pixely feel or a Braille feel or some kind of animation or cursor-driven thing on your website, he's got like 400 different libraries he's he's put together here with demos, with sample code, with like the NPM where to get it. Um it's Crazy. Like it's just one of those ones where you can just get ideas and go, I really like that look for this, and I like that effect. Oh, I really want this light rendering on my site. What would I call that? How do I get that? And this sort of just shows you the library straight away that gives you the glowing effects [41:44] Matt: What's [41:44] Dave Pengelley: or [41:45] Matt: the [41:45] Dave Pengelley: the [41:45] Matt: um what's the website for that again though? [41:47] Dave Pengelley: webdev.technium.io. [41:49] Matt: I had that um saved a little bit [41:52] Dave Pengelley: Yeah, [41:52] Matt: a [41:52] Dave Pengelley: I [41:52] Matt: while ago. [41:53] Dave Pengelley: shared it in one of our WhatsApp groups, Matt. You would have seen it there. But um, it's just a really handy tool, even if you don't use these specific libraries, but you Just want the the visual language to describe to your LLM, or you want to take a screenshot and say, I really want something like this, but you don't need to reinvent the wheel. Um, even one of the one of the um models in here was called Paper CSS, which does these sort of nice little [42:19] Matt: Oh [42:19] Dave Pengelley: hand-drawn cards and [42:20] Matt: it [42:20] Dave Pengelley: things. [42:20] Matt: is [42:21] Dave Pengelley: I used [42:21] Matt: great. [42:21] Dave Pengelley: that to make a client presentation the other day. I said, Here's here's the CSS library, go use this, and it just drew it all and made it look really nice and not. Like formal but organic, which would be beautiful. This is [42:35] Matt: There [42:35] Dave Pengelley: one I [42:35] Matt: you [42:35] Dave Pengelley: used on [42:36] Matt: go. [42:36] Dave Pengelley: the field note the stir the ink as a demo going. Like, if you wanted that kind of effect on your website, where would you even start? How would you know what to ask for? Whereas it's just how's a demo, it's how you do it. So, um, yeah, more resources like that I think are really handy to just inspire people and give you words for your creativity. Um, yeah. [43:00] Matt: I don't know what it is. I I think it's there's a lot of these things, this is kind of like uh this kind of be like a very hipster statement. A lot of these things are super cool until everyone uses them. [43:10] Richard Webbe: Yeah. [43:10] Matt: Um, [43:10] Dave Pengelley: I know. I mean, it's like every website, every website's covered in pills, right? And I'm kind of sick of the pills, and some of my websites still probably have a little bit of that, but I try and go away from that generic AI on most of my sites. I showed you the Tin McGames one last week with little kennel and stuff, right? Um Yeah. [43:28] Matt: I'm I'm not sure what saying. [43:30] Richard Webbe: line between confirmate conformity, fashion, and a competitive advantage. And they are hard things to get the balance right. It depends where you are. If [43:43] Matt: Yeah. [43:43] Richard Webbe: you're talking fashion, you want to be unique enough to stand out, but not so unique that you look like an idiot. And that's what happens, right? And I'm sure all of us have watched someone walk down a runway on a fashion thing on the telly and someone's dressed like a goat and you go, Could that be fashion? Or someone will get something from that and say, Oh, we're into goat materials or into fluffy things, or you know, but it is that fine line because as soon as you follow the tier two fashion, like as soon as you become a hijack marketer and you copy someone, you'll get that convenience of association, but you won't get the uniqueness of success. And never is it more true. What are the most fashionable things? You know, art, um, clothing, music. music and music's a great one when they do mashups right they pinch uh here we go dad's gonna create something uh that we will all relate [44:37] Dave Pengelley: Derelite. [44:37] Richard Webbe: to um [44:38] Dave Pengelley: You talk about fashion. Zoolander walking in at it's literal garbage. [44:45] Richard Webbe: yes oh man uh have you seen alley g's new movie he's got that giant blue puffer jacket on it just sorry it was so big it was cracked [44:53] Dave Pengelley: Anyway. [44:53] Richard Webbe: me up so coming back to that you know you mash up right you hear music And because you know a certain beat, and then they cross that with another voice that you know, it all of a sudden becomes very needed and familiar because they've stolen and mashed in a bunch of things that it knows you've already heard multiple times. So familiarity is a big part of that. So, how do you be different and familiar at the same time? And this is where human human trust and human creativity will come into it. [45:24] Matt: Oh man, I'm getting so many weird parallels in my head right now of of what you're saying. Like I'm picturing like you know DJ sets where they do exactly that and they mix in all these different pieces and it plays on nostalgia, it plays on like different emotions and then pulls you through. Like I wonder, like that's probably one of the most common things that I can think of where that that arena is actually sort of flexed. But I wonder if it's possible to do that in different mediums and in different arenas. And this comes back to like the whole cognitive security thing and like trusting and Being skeptical and all that, like [46:01] Richard Webbe: NLP, [46:01] Matt: could [46:02] Richard Webbe: NLP is the human version of that. So you've heard of neurolinguistic [46:05] Matt: be [46:05] Richard Webbe: programming. So I ran sales training and go-to-market and executive training for many years and I avoided [46:12] Matt: a little [46:12] Richard Webbe: the NLP story. One, [46:13] Matt: bit more than that. [46:13] Richard Webbe: because everyone was doing it and I thought it was boring. But number two, NLP is about acting and looking familiar with the person you're talking to in terms of phrases and hand movements, but not copying them exactly, just giving them a familiar bridge to connect you with. Film. And if you find a very neurodiverse person and you're talking to them, sometimes every time you say something, they'll say, Oh, yeah, I've done that too. That's their natural extension to say, we are the same and we should be talking together. They're not trying to trick you, they're really trying to connect with you, but it can backfire. And so the more AI automates all of this stuff, and the more, you know, as Dave's showing us very cleverly, how you can reproduce this for games and Business tools like you do, at the same token, there'll be a line there that if you step over it, it'll be negative. [47:06] Matt: Yeah. That's it. I I my headline today when I did my prep for this show was is AI making us independent or more dependent on different things. [47:17] Richard Webbe: And and yes, yes, and yes. Go on, Dave, sorry. [47:20] Dave Pengelley: Yeah, no, I I I think it it is a bit a bit of both, and I think that's that's the trap, and it's like um we mentioned last week, is AI a new industrial. Revolution, or is it a new renaissance? And who is using it in which way? Some people are going to be using it to automate, optimize, downsize, upsize, scale, and sort of industrial revolutionalize it. Other people are going to be using it for renaissance purposes to innovate and create and try new things and do things that were beyond the before. And so they become dependent on it because but in a way that is dependent to do the new things they could never do before. [47:56] Richard Webbe: Yes. [47:56] Matt: Yep. [47:56] Dave Pengelley: Right? [47:57] Matt: It's [47:57] Dave Pengelley: Um [47:57] Matt: kinda like losing an um [47:58] Dave Pengelley: losing your car. Like if you're used to being on a drive, like down to the shops and bring a whole load of groceries back, and then all of a sudden you lose that. And now great, now I've got to walk back. Then I can't carry as much. I'm gonna have to do extra trips. I've got to do just in time trip and go there every day because I can't carry like a full bootload home uh anymore. Or maybe I don't get um the the the case of soft drink or whatever because I don't want to carry that the five kilometers home and um we just we don't have those things. So yeah, um all these things, you know, it comes back to the user. And the intention, and you talk about, you know, trust. What is what is your integrity? What is your trust? Why are you building these things? Why are you producing these things and putting them out there? Is it because you want to scam quick bucks off people? Or is it because you actually are genuinely trying to improve the world through new experiences, new optimizations, new productivity, new spare time or games or entertainment or movies or whatever, right? Like um, that side I showed, uh, we've got a great comment here from Isaac. He'd use uh the website to develop the skills. But it's cool. So it wouldn't just ape it directly, but would go, okay, how do they do that? And wants to learn from that, which I think is another great application. You don't have to use those exact things. But again, that's that inspiration and creativity is why I share them because people don't know this stuff exists. And to your point, Richard, they just go, hey, I give me a great website. Like what? [49:22] Dave Pengelley: To what point? And this is the the chat I had with um um uh [49:29] Dave Pengelley: name's came. Uh, the guy from Pitch Kitchen um had on the show a few weeks ago. Uh, and he was that getting the narrative and the marketing messaging was before you get AI to build a website, before you get it to build SEO pages. What's the truth of your business? What are you actually promoting? What do you do? What makes [49:46] Richard Webbe: Absolutely. [49:46] Dave Pengelley: you unique? Get [49:47] Richard Webbe: Absolutely. [49:47] Dave Pengelley: that messaging right first before you handle the AI. So like your horses, how many input variables are you giving this thing? How much context are you giving it in order to get a unique differentiated result that it solves your particular problem? [50:00] Richard Webbe: What was great about the horse racing is a lot of that data was available on the website infinitum. And all I had to do is draw the LLM and give it a weighting on how much to carry in each part. So when you think about designing a website, you don't think of it from your point of view, think of it from your customer's point of view. So I'd say to the LLM, give me a great website that someone between the ages of two and four would find very interesting and relates to icons and characters that are familiar. Now that's a whole different question to give me a great website. So again, I think putting the trust thing to one side from a marketing point of view, you've just got to describe the problem you're trying to solve properly with a lot of prompts and a lot of things. On the other side, I believe trust there's going to be an opportunity to sell trust as a service. Now I'm just laughing here because I bought some books for my kids online. I'm an impulse buyer, I totally admit that. Was great. He bought me all these books and history, usually from the Reader's Digest, which would come in every few months. That little hey, look at this great book. And my dad would immediately write away and ended up with this great. I'm looking around because I've got this really great book that was called The Last Two Million Years. Basically described the entire human race and the planet Earth. And there's a lot more of those coming out. And I had a great one for the kids. And then it said, Hey, buy this history timeline book. It's 30 pages long, it rolls out to 30 meters. It's great. You know, 119 or 50, whatever. I thought, oh, the kids would love that. So I didn't think. And I just bought it. I got no acknowledgement of the purchase. I did it because it was connected to a site that I just bought from and had bought from before. I've got no references. I've got nothing. And so I immediately went right. So I grabbed um the uh data they had from where I bought it and I dropped it into my LLM and said, tell me, it This is a scam, isn't it? Comes back and said, Richard, yes, many red flags on this one. I can see on the web numerous complaints about what you've done, people not receiving the product, blah, blah, blah. I thought, oh bugger. So then, you know, my trust had just been betrayed. Uh, my experience has gone up a little bit. I think for a guy who knows cyber and purchasing an AI, it shouldn't have happened. But here, I put in the complaint to my bank, I won't say who, and they just sent me a letter just now while we're talking, could you please send Use the fact that it hasn't been delivered. How do you get evidence for that? Could you please send us the email where they said they're not going to deliver? Could you? They're wanting me to give evidence that doesn't exist, [52:43] Dave Pengelley: Yeah. [52:43] Richard Webbe: which is why I put a stop on it. It is just illogically stupid, and I'm about to write that email [52:49] Dave Pengelley: I [52:49] Richard Webbe: back to [52:50] Dave Pengelley: I [52:50] Richard Webbe: the [52:50] Dave Pengelley: nearly [52:50] Richard Webbe: bank. [52:50] Dave Pengelley: got caught on that last year, leading into Christmas. I saw an ad for magnet chess and the ad looked amazing and I was like, oh, that's so good. And bought it on the spot. And then, you know, the weeks went by and I'm like Where did that order go? I haven't seen that. And it came like two days before Christmas. Legit product, magnet chest, really fun, really cool. But it was one of those Instagram purchases where I saw the video, I went, Oh, that looks heaps fun. Buy straight away. But they did really well off their Instagram promotion and they got a lot of orders and they just had a backlog and did come just in time. But I was like, Where's my order history? Where's my confirmation? Where is this? But I got scammed. I nearly lost a bunch of money buying a car off car sales because [53:25] Richard Webbe: Wow. [53:25] Dave Pengelley: it was dodgy people with dodgy services. And I managed to get it back, but the We're like, yeah, you know, what are we gonna do? It's five grand, like we've got bigger fish to fry. Like, do you really want to press this further? And I was like, no, fine. [53:39] Dave Pengelley: Um, but you know, people like us that should know better, we still get caught up in the moment sometimes. Um, and justify and make mistakes. We're gonna wrap it up very shortly, guys, on the pro topic of AI slop and why are people putting it out there. I did see this one yesterday from itch.io, which is a sort of indie um game hosting platform. platform and [54:00] Matt: Terabytes. [54:00] Dave Pengelley: they just deleted a couple terabytes of AI slot. So they are getting people just spamming crap games up onto their service and they are working to try and get rid of it. So the [54:12] Richard Webbe: But [54:12] Dave Pengelley: ones [54:13] Richard Webbe: in [54:13] Dave Pengelley: that [54:13] Richard Webbe: all fairness, one man's slop is another man's [54:16] Dave Pengelley: no [54:16] Richard Webbe: art. [54:16] Dave Pengelley: but when you get the same thing and you you just change the the like I saw this this whole someone shared a screenshot this page of the same game and they just changed the title sort of like aliens in the ocean in space in a west And in a this, in a that. So they're making minimal changes and pumping out 50 copies of the same exact rubbish. That's just spamming the market. That's the old school SEO keyword stuffing thing. And they're like, we don't want that. If you're just mass producing rubbish, that's not what we want. We want people actually making games and putting them on here. So good on them. Um, we didn't even get to news. Uh, what was the big news? Um, Terror 6.1 dropped from OpenAI. OpenAI. I doubled their limits, but now they're gonna make their limits use more usage across all their products. Um, [55:07] Matt: Clint [55:07] Dave Pengelley: which I'm [55:07] Matt: Ultra. [55:08] Dave Pengelley: not um I mean, maybe maybe with double usage is fine, but I like because I was deliberately sort of scamming the system a little bit by keeping chat separate from codex and doing a whole bunch of planning and chat and then getting codex to do the delivery. Um, and so they're going, no, no, chat's gonna use your codex quota now. It's one quota usage. I was like, ah, come on. But I've said, but we'll give you double the quota now, all the stuff they It on dev day around halving it, they've gone, no, no, we're now doubling it. That's hard to keep up, right? Um, I [55:36] Matt: Yeah. [55:36] Dave Pengelley: put a thing out on LinkedIn that we've been predicting all this tokenomics stuff for months. [55:40] Matt: I really like that video. It was very good. [55:43] Dave Pengelley: Yeah, I knew we've been talking about even in those clips. I was I've there's a few ones where I'm like, we've been talking about this for weeks, we've been talking about this for months. Um, [55:50] Matt: Yeah. [55:50] Dave Pengelley: but I just went to my um my homies agent that's got all the transcripts and said, pull the quotes where we've talked about, you know, tokenics and and subscription. Charges and stuff, and I went and found like the eight videos and with the time codes, and I was like, Ah, I don't have all of these as MP4s offline on my PC. Sometimes they're just something. So I was like, can you use YT DLP and make a Python script that will just go and pull these segments? Next thing you know, I've got a folder on my computer with eight segments with these little one-minute clips. And so then I was like, Great, drop them into script, just kill the little bits of quote that I didn't want that are brought in, like the the wrapper. I was like, bang, done. [56:30] Matt: Perfect. So good. [56:31] Dave Pengelley: Um [56:32] Matt: How's that for a time saving though? Like that's the kind of stuff that you know you don't really think about, but [56:36] Dave Pengelley: I [56:36] Matt: like [56:36] Dave Pengelley: never [56:36] Matt: it's [56:36] Dave Pengelley: would have made that if I'd had to go through and manually find those clips and copy and paste them out manually and do that. I never would have made that video. Even with that time saving, I still barely made the video because I'm like, oh, still a hassle. I've got to then put them into script and [56:49] Matt: a good idea. [56:49] Dave Pengelley: I do need to tidy it up and just reformat it into vertical because I don't trust the AI with some of that stuff to get the right section of quote that I'm happy with. But um win of the week, yeah. That that could be [57:00] Richard Webbe: I [57:00] Dave Pengelley: that [57:00] Richard Webbe: see [57:00] Dave Pengelley: could be a [57:00] Richard Webbe: you. [57:01] Dave Pengelley: win of the week that I actually uh made that YouTube thing and I put a few animated shorts up on our YouTube as well. [57:07] Matt: Nice. Super [57:09] Richard Webbe: Yeah. [57:09] Matt: cool. Now we're just [57:10] Dave Pengelley: Um [57:10] Matt: gonna get to like Alex or Mersey putting out one an hour. [57:14] Dave Pengelley: yeah. Um, and then of course there is all the consternation around Grokbot and um quota limits. And did they go up? Did they go down? Was it a visual glitch? Elon has now come and said, Oh, yeah, they they cost too much to run, we're not making enough money, which people are saying. Saying, didn't you say money's not going to exist, dude? Like, what the heck? Um and then [57:35] Richard Webbe: Bit [57:35] Dave Pengelley: there's [57:35] Richard Webbe: early, [57:35] Dave Pengelley: this. [57:36] Richard Webbe: bit early. [57:36] Dave Pengelley: And then there's this guy that's gone particularly gone a little bit viral uh with excuse the language. Go F yourself, Elon. You screwed us over by cutting the weekly limits on Grockpot. You took advantage of the limit display glitch to F us in the A. And even people who weren't hit by the glitch, including me, still watch the ultralim melt away like snow in the sun, U S O B. And it goes on. [57:58] Matt: Wow. [58:00] Richard Webbe: There's probably a committee of three hundred people that made those decisions, but Elon gets that punch right in the nose. [58:07] Dave Pengelley: Well, [58:07] Matt: Yeah. [58:07] Dave Pengelley: because because everyone, and there was no official announcement that I saw, man. Did you see there was no official [58:12] Matt: No. [58:12] Dave Pengelley: post around what happened? But like I watched all weekend and I had about 30% usage. And I was like, cool, I'm just just gonna hold out. I don't need to use Grokbot over the weekend. Um and then Monday morning was at 70%. Like I'd like I'd lost half my usage, and I'm actually going to my my help desk bot going, which bot used the usage? Where did this go? Did I have a background automated process that ran something? Like, what the heck? It's like, no, no, uh, most of the work was on the animation on the like last Thursday. And I'm like, that doesn't explain why it's jumped up in the last 24 hours. And then [58:42] Matt: Yeah. [58:42] Dave Pengelley: I was like, check X, and it said found this post where Elon's like complaining about the cost of things. Um, [58:49] Richard Webbe: Yeah. [58:49] Dave Pengelley: and then [58:49] Matt: It's [58:50] Dave Pengelley: 24 [58:50] Matt: five [58:50] Dave Pengelley: hours [58:50] Matt: things. [58:50] Dave Pengelley: later, mine had gone back down to 30%. And I was like, What is even happening here? [58:56] Matt: Yeah. Mine never uh mine never moved. never went backwards. There was no rollback for me. But it absolutely went to two times burn rate for a good you know day or so. I don't know where it's at now, but like I literally have like per percentage of your weekly limit, [59:17] Dave Pengelley: Yeah. [59:17] Matt: you get X amount of tokens. Like I've got all those tokenomics worked out. And yeah, like it was super super clear that it went it doubled. Like it went two times for for a good day or so. [59:30] Dave Pengelley: My usage literally went from thirty something to seventy something and then back to thirty something in twenty four hours. It was weird. [59:35] Matt: Yeah. [59:39] Dave Pengelley: Um that that didn't really double, but it looked like it doubled. [59:43] Richard Webbe: I [59:43] Dave Pengelley: Um [59:43] Richard Webbe: can't wait for all these people that were complaining about data centers that when what they want done in their billing or their environment or their virtual world goes up tenfold because no one let them build more data centers to keep doing it for them. [01:00:00] Matt: That's another weird thing. It's gonna continue to get weirder, I think. Uh I don't really know where I don't think [01:00:04] Dave Pengelley: So [01:00:04] Matt: many people understand where that's actually going to go. [01:00:07] Dave Pengelley: and [01:00:07] Matt: Um, [01:00:07] Dave Pengelley: I mean you you you mentioned that we're dependent more dependent on on all this sort of technology. I I suppose I'd close out the show by going, don't become like dependent on it for your for everything, but also take advantage of it now. It's never gonna be I can't say it's never gonna be cheaper than it is now, but the odds are the tokenomics will shift and maybe as [01:00:27] Matt: I'm if not not sure you're able be do going that. to to [01:00:27] Dave Pengelley: they get the models cheaper. You can have some level of AI in the future that's probably better than what we have today, that's cost effective. But also now's the time to get on some of these tools and use them while they're arguably cost effective to really get creative and try things. [01:00:46] Matt: Hundred [01:00:46] Richard Webbe: Yeah, [01:00:46] Matt: percent. [01:00:46] Richard Webbe: yeah, [01:00:46] Matt: Personal [01:00:47] Richard Webbe: right. [01:00:47] Matt: and business, like, yeah, I I reckon that there it there is definitely value there right [01:00:52] Dave Pengelley: Yeah. [01:00:52] Matt: now. [01:00:52] Dave Pengelley: There's there's there's a window of opportunity which may be closing or it may, you know, drop away for some period of time and then come. Back and then we we don't know what's going to happen with the tokenomics on this stuff. The Frontier labs are changing their plan costs and quotas by the hour. Whether that's Grok, whether that's OpenAI, I don't know what Anthropic's doing at the moment. Um, at the moment, things like OpenAI, they're like, yeah, we're uh doubling your quota, but more things are gonna count against it. But also dots are still unlimited, and we're gonna change which model the dots use, which means your dots probably gonna get dumber, but we can give it to more people. And it's like, man, they're just This will yeah, it's that's crazy. So get on it. [01:01:35] Matt: Yep. Good point. Good way to end [01:01:39] Dave Pengelley: Good [01:01:40] Matt: it. [01:01:40] Dave Pengelley: way to end it. Get on it. Or as as the late great Sir Patrick Stewart would say, make it so. [01:01:46] Matt: Hmm. Am I the only one that believes he's died twice? [01:01:50] Dave Pengelley: Yes. [01:01:50] Matt: Like I I f I my [01:01:52] Dave Pengelley: X-Men [01:01:52] Matt: partner told me [01:01:53] Dave Pengelley: wasn't [01:01:53] Matt: that. [01:01:54] Dave Pengelley: real. [01:01:54] Matt: No my [01:01:54] Dave Pengelley: That [01:01:55] Matt: partner [01:01:55] Dave Pengelley: was a movie. [01:01:56] Matt: Yeah, she she's she told me oh Patrick Stewart's passway. I'm like, didn't you? Like pass away a few years ago. [01:02:01] Dave Pengelley: No. [01:02:02] Matt: Like I I must be misconstruing it with somebody [01:02:04] Dave Pengelley: Must [01:02:04] Matt: else. [01:02:04] Dave Pengelley: be someone else. No, so Patrick, uh, this year, just now. [01:02:08] Matt: Yeah, right. [01:02:09] Dave Pengelley: Um, [01:02:09] Matt: Interesting. [01:02:09] Dave Pengelley: and funnily enough, I saw William Shatner on an interview and he's ninety five, he's like ten years older, and he's still going like a trucker. It's just crazy. [01:02:18] Matt: Nuts. [01:02:19] Dave Pengelley: Um not that they're the same person. They just both happen to be captains of the enterprise. Um [01:02:26] Matt: Cool. [01:02:27] Dave Pengelley: yes. [01:02:27] Matt: Thanks for [01:02:28] Dave Pengelley: Uh [01:02:28] Matt: this. [01:02:29] Dave Pengelley: thank you. Good chats. Uh thank you for those that joined us online. Make sure you have liked and subscribed. You tell your friends. You drop a comment in the chat. Thanks for those that joined us live in the comments. Love seeing some engagement during the show. Thank you for that. Check out the field notes. Check out the operator stories. I'm interviewing really interesting people with different perspectives on all this technology. It's just, you know, different points of view. And like this show, it's not a hype show. We're not shilling for them. We're just, I'm just hearing what they've got to say, where they're at, how they're helping clients in different ways. And that's amazing how much of this stuff just comes back to those bare basics of business. And then using AI as an accelerator once you know what the business should be doing. So not rocket science. Just start playing. [01:03:13] Matt: You'll find all them on the YouTube channel, right? [01:03:15] Dave Pengelley: They're all on the YouTube channel. aioperatorspod.com. They're on the website too. Check them out. Thank you, Australia. Thank you, the world. Thank you, gentlemen. We'll see you next time.