All episodes
What We Cut From Our AI Workflows This Year
livestream1:37:4618 Nov 2026
In the final weekly panel of 2026, the AI Operators compare which habits they would keep, what they would cut from their agent workflows, and where faster prototyping changes the work. They also discuss how to evaluate the sources and instructions behind an AI answer.
Show notes
In the final weekly panel of 2026, the AI Operators compare which habits they would keep, what they would cut from their agent workflows, and where faster prototyping changes the work. They also discuss how to evaluate the sources and instructions behind an AI answer.
Dave Pengelley, Richard Webbe and Matt Slager are joined by Joanna Jones, Kate Fabian, Manjula Kerai and Shawn Hain.
What the panel unpacked
- Cut what no longer helps. Shawn describes shrinking his framework, while Dave talks about revisiting the prompts behind the show's own production workflow as models change.
- Make the next decision easier. Matt asks his agent to take questions one at a time. The panel compares using a model to guide the conversation with delegating a task to a model suited to the work.
- Prototype before committing. Faster experiments let builders explore more ideas, including game concepts, without investing in the finished product first.
- Look behind the answer. The panel discusses how instructions, tools and accessible data can shape an apparently neutral AI response.
Transcript
[00:00] Dave Pengelley: raid the uh the Christmas cupboard for a hat, but then if I wore a hat, as we talk about every week, Richard, people wouldn't see my incredible hair.
[00:10] Dave Pengelley: No hat.
[00:11] Richard Webbe: I get more comments about your personality and hair than anything else on the planet.
[00:16] Dave Pengelley: Excellent. As it should be. It's
[00:18] Richard Webbe: And
[00:18] Dave Pengelley: funny.
[00:18] Richard Webbe: they're
[00:18] Dave Pengelley: I um
[00:18] Richard Webbe: all good, they're all positive.
[00:20] Dave Pengelley: as as you do in Australia, I went and had a uh skin check yesterday,
[00:24] Richard Webbe: Yes.
[00:24] Dave Pengelley: uh as as you should. Um And I was like, oh, he's probably gonna want to check the scalp. And so I was like, no product yesterday, like no, no, nothing. And so all day I just went like this big head is just sort of in fluffy mode with no product holding it together, just using my sunglasses, like a headband. And then like this morning when I woke up, it was just like
[00:46] Dave Pengelley: So anyway.
[00:47] Richard Webbe: Ah, very,
[00:48] Dave Pengelley: Uh
[00:48] Richard Webbe: very cool.
[00:49] Dave Pengelley: the trials and travails.
[00:51] Richard Webbe: How do you do your
[00:51] Dave Pengelley: But
[00:51] Richard Webbe: skin checks? Do you use the new AI machine they
[00:54] Dave Pengelley: no,
[00:54] Richard Webbe: have?
[00:54] Dave Pengelley: I went to a doctor who uh Who felt me up all over
[00:58] Richard Webbe: Wow,
[00:59] Dave Pengelley: and got
[00:59] Richard Webbe: that's
[00:59] Dave Pengelley: his little thing out? He's like, uh got his little microscope with a little light on it. He's like, he runs over and looks for little moles and freckles and things and goes, oh, what's that? Uh no, that's all right. And keeps moving.
[01:09] Richard Webbe: so
[01:10] Shawn Hain: Yeah.
[01:10] Richard Webbe: they
[01:10] Dave Pengelley: Very
[01:10] Richard Webbe: have this
[01:10] Dave Pengelley: invasive.
[01:10] Richard Webbe: new machine and it's about a hundred and fifty AI little cameras in a sort of uh a giant dome and they ask you to strip off naked. Um I don't get completely naked, I just wear my jocks and a couple of footy socks. And then I um on my phone. Thank you very much, Matt. Anyways.
[01:31] Richard Webbe: So then I stand there, you know, naked, and that takes this giant scanning photo. And then they take you into another room and you lie there. And AI automatically goes to each mark on your body. It's oh darry, that's just a people, darry, that's some girl scratched your back last night, and it just goes to all the bits and pieces.
[01:49] Dave Pengelley: I've seen that on TikTok at the very popular channel.
[01:54] Matt Slager: So what what's the is that that's just
[01:57] Richard Webbe: Yeah, it is. And what they do is it it uses AI. They're very good. They got this giant chart on the wall that tells you of all the different carcinomas and moles and things that you can get. Uh, and it identifies all of them, said that's okay. But when there's a couple that's not sure about, that's when the doctor comes along. Uh, mine's a female doctor, and gets out her little magnifying glass and just sticks it on bits and pieces and said, you know, that's okay. That's the lovely Joanna.
[02:25] Dave Pengelley: Excellent
[02:25] Richard Webbe: And she's
[02:25] Dave Pengelley: clinician.
[02:25] Richard Webbe: dressed.
[02:26] Dave Pengelley: We're
[02:26] Shawn Hain: Yeah.
[02:26] Richard Webbe: Lovely.
[02:26] Dave Pengelley: talking about skin cancer checks, Joanna. You joined just at the right time.
[02:30] Joanna Jones: What's happening?
[02:30] Dave Pengelley: We're talking about skin cancer checks and AI and people touching each other with magnifying glasses to analyze things.
[02:39] Joanna Jones: Good morning, sounds right up the nurse alley, hey.
[02:42] Richard Webbe: Joanna,
[02:42] Dave Pengelley: Yep.
[02:43] Richard Webbe: you're very shiny today. Is that for Christmas? You look lovely.
[02:46] Joanna Jones: Did a Christmas doo-doo?
[02:48] Dave Pengelley: Ooh.
[02:49] Richard Webbe: I bought my little friend.
[02:51] Joanna Jones: Yeah.
[02:52] Dave Pengelley: I've got my beard.
[02:53] Richard Webbe: He's got his beard.
[02:55] Joanna Jones: Oh brilliant, absolutely
[02:56] Dave Pengelley: Yeah.
[02:57] Joanna Jones: brilliant.
[02:57] Dave Pengelley: Yep. So we can
[02:59] Joanna Jones: Matt,
[02:59] Dave Pengelley: do this.
[03:00] Joanna Jones: Sean,
[03:00] Shawn Hain: Yeah.
[03:00] Joanna Jones: where's your
[03:02] Shawn Hain: No,
[03:02] Joanna Jones: sparkle?
[03:02] Shawn Hain: my I I think I know where my Christmas stuff is. It's in a box at the back corner of my garage after we moved, and I don't think it's coming out this year.
[03:11] Richard Webbe: Yeah.
[03:11] Dave Pengelley: So I think I've had this on a hook on the wall next to my desk for four years now, since the last time when I was still at Salesforce. And I did a Christmas a Christmas video uh to my team. Um and I
[03:26] Richard Webbe: Son came up to me and said, Dad, can we set up the Christmas tree? I said, sure. He goes, where is it? I go, I don't know. And I had a look, and it's sitting in my lounge room from when I pulled it down last year in the bag. So I didn't even bother moving it from the spot. So we just took it out and stuck it together.
[03:43] Matt Slager: I think that's
[03:43] Dave Pengelley: We've
[03:44] Matt Slager: a good
[03:44] Dave Pengelley: got
[03:44] Matt Slager: one.
[03:44] Dave Pengelley: two in the back corner of the yard that we need to um burn because we get live ones. And so then when it's finished, we just throw it in the corner and ostensibly to have some kind of like fire pit during winter once it's dried out and we're too Years behind on burning our Christmas trees. So we're about to have a third one in the corner.
[03:59] Richard Webbe: Do
[03:59] Matt Slager: Well,
[03:59] Richard Webbe: you use that's a good question? Do you use do you use real Christmas tea, treats?
[04:04] Dave Pengelley: Yeah.
[04:05] Richard Webbe: Wow.
[04:07] Shawn Hain: Never done that.
[04:08] Richard Webbe: I've
[04:08] Dave Pengelley: Never
[04:08] Richard Webbe: never
[04:09] Dave Pengelley: done it.
[04:09] Richard Webbe: done
[04:09] Dave Pengelley: I've
[04:09] Richard Webbe: that.
[04:09] Dave Pengelley: got
[04:09] Shawn Hain: Yeah.
[04:10] Dave Pengelley: Santa beard stuck in my mouth now. It's
[04:12] Joanna Jones: Uh
[04:12] Dave Pengelley: weird.
[04:12] Matt Slager: I'm not sure.
[04:12] Joanna Jones: the
[04:13] Shawn Hain: Yeah,
[04:13] Dave Pengelley: Um
[04:13] Shawn Hain: not mine always in plastic, so
[04:15] Dave Pengelley: that's that's that's
[04:16] Richard Webbe: Oh.
[04:16] Dave Pengelley: what she said, right?
[04:17] Richard Webbe: There
[04:18] Dave Pengelley: Um
[04:18] Richard Webbe: there are
[04:18] Joanna Jones: colour.
[04:18] Richard Webbe: too many jokes to do with that exactly.
[04:20] Shawn Hain: one. that's the
[04:24] Dave Pengelley: uh yeah, no, we we started the live tree, I don't know. When nearly a decade ago, we've been doing live trees, and it's good. You just got to remember to water them, and you can't get them too early, or they die before Christmas. Um, they're
[04:37] Richard Webbe: They
[04:37] Dave Pengelley: a
[04:37] Richard Webbe: smell
[04:37] Dave Pengelley: bit
[04:38] Richard Webbe: nice.
[04:38] Dave Pengelley: they smell nice. The dropping pine needles is a pain because they don't vacuum particularly well, it's all sweeping them up. I wish we had more
[04:46] Richard Webbe: Very
[04:47] Dave Pengelley: variety
[04:47] Richard Webbe: true.
[04:47] Dave Pengelley: of Christmas trees, like in America where they've got all the different kinds of firs and pines and things. Here we're pretty much stuck with one
[04:53] Richard Webbe: Yeah.
[04:53] Dave Pengelley: kind of pine.
[04:55] Joanna Jones: Yeah,
[04:55] Richard Webbe: Yeah.
[04:55] Joanna Jones: I remember when I was growing up that we used to go into the bush and get a Christmas cherry tree.
[05:00] Dave Pengelley: Mm-hmm.
[05:00] Joanna Jones: I love the smell of that. But yeah, Dad used to, we all used to go in and sneak around. I mean everybody was doing it then. That was the you know, a little while ago. But
[05:12] Dave Pengelley: Yeah,
[05:12] Joanna Jones: yeah,
[05:12] Dave Pengelley: we're
[05:12] Joanna Jones: I loved having a live tree.
[05:15] Dave Pengelley: something nice about it.
[05:17] Joanna Jones: Oh,
[05:17] Dave Pengelley: Um
[05:17] Joanna Jones: I was just, yeah, it smells beautiful.
[05:20] Richard Webbe: My dad bought us a synthetic Christmas tree when we were very, very young, and it lasted I would say with my mum probably at least 40 years, probably longer. And I said, Dad, why do we have a synthetic tree? And he has this this many years ago. I said, it's environmentally friendly, Richard. We don't have to cut down trees. And we
[05:41] Dave Pengelley: Have
[05:41] Richard Webbe: use
[05:41] Joanna Jones: And
[05:41] Richard Webbe: it
[05:41] Dave Pengelley: you
[05:41] Richard Webbe: every
[05:42] Joanna Jones: then
[05:42] Richard Webbe: year.
[05:42] Dave Pengelley: have
[05:42] Richard Webbe: And we use
[05:43] Dave Pengelley: you
[05:43] Richard Webbe: it
[05:43] Dave Pengelley: seen
[05:43] Richard Webbe: every
[05:43] Dave Pengelley: the um
[05:43] Richard Webbe: year.
[05:43] Dave Pengelley: the Ryan Reynolds Will Ferrell movie, the the Scrooge Adaptation Spirited on Apple
[05:48] Richard Webbe: That
[05:48] Dave Pengelley: TV?
[05:48] Richard Webbe: it's wonderful.
[05:50] Dave Pengelley: So that I mean that
[05:50] Joanna Jones: we
[05:50] Dave Pengelley: whole plot of that is that he's pitching the the natural trees association against the plastic tree people
[05:55] Richard Webbe: That's
[05:55] Dave Pengelley: and trying to make it
[05:55] Richard Webbe: right.
[05:56] Dave Pengelley: a cloud. Last warfare around.
[05:59] Richard Webbe: That is a beautiful um uh s a salesman's etiquette. I I watched that, I thought it was brilliant.
[06:07] Dave Pengelley: Yes, yes. Well, it is a uh we uh for those that are just joining and catching up, this is gonna be the last show of the year. That's why I brought in some of the ensemble of extra guests and friends we've had through the year. Uh, many people were not as uh available, unfortunately, as is the case on on this sort of show. But uh yeah, welcome everyone. This is AI Operators and I should probably run the thing.
[06:40] Dave Pengelley: I believe this is show forty
[06:42] Matt Slager: I
[06:42] Dave Pengelley: four.
[06:42] Matt Slager: don't know.
[06:44] Richard Webbe: Wow,
[06:44] Dave Pengelley: Hard.
[06:44] Richard Webbe: well done, Dave. You've done such a great job rallying us, producing this, doing the bubble overs or those credits, all it's
[06:52] Dave Pengelley: I
[06:53] Richard Webbe: brilliant.
[06:53] Dave Pengelley: I didn't realize we started In January, Matt, you mean Mano. When
[07:00] Matt Slager: Is
[07:00] Dave Pengelley: I was
[07:00] Matt Slager: it
[07:00] Dave Pengelley: looking
[07:00] Matt Slager: January?
[07:01] Dave Pengelley: back, yeah, I looked back at episode one, the start of new chat episode, our first one. That was January 25. That was before Australia Day.
[07:09] Joanna Jones: Oh, wow.
[07:10] Shawn Hain: Cheese,
[07:11] Dave Pengelley: So we've been doing this
[07:12] Shawn Hain: the
[07:12] Dave Pengelley: all year.
[07:13] Shawn Hain: the year is just flying by.
[07:16] Dave Pengelley: Right?
[07:16] Joanna Jones: Oh,
[07:17] Dave Pengelley: Like
[07:17] Joanna Jones: magic, you've brought magic, Dave.
[07:19] Dave Pengelley: uh, I wish I wish the channel had seen a little bit more magic. I mean we've we've cracked a hundred, we're up to like a hundred and four followers. We're getting there. Um, but it's been good. It's been good. Uh, I've really enjoyed it. And I want to say thank you up front to all the people that have been part of making it successful. Uh, Matt and Richard, especially for your regular contributions, but Joanna and Sean for your coming in and out. And Sean coming up to the IRL event was excellent. Like, no, this has been a team effort, and I thank everyone.
[07:45] Matt Slager: Yeah,
[07:45] Richard Webbe: Yeah,
[07:45] Matt Slager: for
[07:45] Richard Webbe: well
[07:46] Matt Slager: sure.
[07:46] Richard Webbe: done, Dave. Um Joe, I just want to ask a sensitive question and you don't have to answer it, but I know you're a nurse, so you can handle these things as well as an AI specialist. But the fact that you've got a cocktail dress on. Did you go home last night?
[08:04] Joanna Jones: Just checking to make sure the children aren't around.
[08:10] Richard Webbe: It looks
[08:11] Joanna Jones: No,
[08:11] Richard Webbe: lovely.
[08:11] Joanna Jones: it's
[08:11] Richard Webbe: It's
[08:11] Joanna Jones: just
[08:12] Richard Webbe: it's very Christmassy, I agree.
[08:14] Joanna Jones: Okay, well, if we're going to be ruthless like this, I will show you I have jeans and my fluffy
[08:22] Joanna Jones: My fluffy slippers on as well. I my son just said to me, you're doing that thing they were doing in COVID, shirt for the top, jocks on the top, bottom. I
[08:31] Dave Pengelley: Yeah,
[08:31] Joanna Jones: was like,
[08:31] Dave Pengelley: yep.
[08:32] Joanna Jones: I'm not in
[08:32] Matt Slager: Guilty.
[08:32] Joanna Jones: I'm not in jokes.
[08:33] Matt Slager: I don't ever die.
[08:35] Dave Pengelley: Yeah. I've done many interviews and and and customer things and stuff where I'm wearing like board shorts or sports shorts on the bottom and wearing an ice shirt and stuff up top. Um I'm actually enjoying the sneakers today. Not that anyone really cares, but um
[08:49] Richard Webbe: I caught it with an old I caught it with an old friend from Ericsson today and during the COVID because it was Ericsson and they're over. And they knew we had all the big lockdowns and all the roads were shut down. I was busy on my pushbike riding, doing my kilometers up while I was doing my conference calls. And I was on with, gee, I think I was on with one of the very, very, very senior leaders at Erickson. And he said, Richard, I hear a noise. Where are you? I said, Well, I'm actually out on my bike riding. He goes, No, you're not. I go, Yes, I am. And of course I had my phone mounted on the handlebars. So I turned the video on and he sees the stars going by in my head. Come up and go, Hello.
[09:31] Joanna Jones: Oh, it's pretty amazing where you can record and take calls from now. And instead of the phone box.
[09:38] Richard Webbe: Oh, yes.
[09:39] Matt Slager: Yeah.
[09:40] Dave Pengelley: Yeah,
[09:40] Richard Webbe: Yes.
[09:40] Dave Pengelley: I remember the episode where I had to go to do I was helping a charity organization had to run a workshop and I had to leave the this call halfway through and I connected in from my phone from the train. Like the train signal was a bit spotty, but it was amazing that I could even still participate in the uh the call.
[09:55] Shawn Hain: Yeah.
[09:56] Richard Webbe: That was that's brilliant. I did have the naked son come in the room when I did have the CO of Telstra and Ericsson online and one of the execs said, Richard, and I said, Oh, it's pants off Tuesdays. We always had pants off Tuesdays.
[10:11] Joanna Jones: And
[10:11] Shawn Hain: Yeah,
[10:12] Joanna Jones: I think Matt,
[10:12] Shawn Hain: just
[10:12] Joanna Jones: you
[10:13] Shawn Hain: around
[10:13] Joanna Jones: um brought
[10:13] Shawn Hain: it.
[10:13] Joanna Jones: you came to um one of the podcasts from South Australia.
[10:17] Matt Slager: Yeah, from the beach in Adelaide.
[10:19] Joanna Jones: Oh, I still remember that water. I look beautiful.
[10:22] Matt Slager: Yeah, that was pretty cool. You went allowed in though, they had all these signs saying that there was like bad out
[10:32] Joanna Jones: South Australia.
[10:34] Matt Slager: Yeah, like out of everyone, Dave, you've made every single one of these forty four.
[10:39] Dave Pengelley: Well, I mean it was it was it's kind of my been my baby, right? I'm the one who pulled it together, so I figure I gotta turn up every week. Um Yeah, it's it's it's it's yeah, it's a it's a pleasure and honor. And and you know, doing the expanded content as well, it's again, it's I think we've we've seen growth and um it's been good. The operator stories are amazing. I didn't expect them to take off the way they have. Like, um, the view counts are uh one thing, but just the the caliber of guests and people that are reaching out that want to have chats is just phenomenal. Like, um, yeah, the guy I spoke to who I'm pushing out on Monday, ELAD, he
[11:15] Dave Pengelley: has fundamentally changed the way internet traffic moves around and stuff with some of the technologies he's
[11:20] Richard Webbe: Wow.
[11:20] Dave Pengelley: built and sold over the years. Like it's just crazy,
[11:23] Joanna Jones: Wow.
[11:23] Dave Pengelley: the the guys that I'm getting on this show. Like it's Um
[11:26] Joanna Jones: Yeah.
[11:26] Dave Pengelley: really interesting. So
[11:27] Matt Slager: That's
[11:27] Richard Webbe: Yeah,
[11:27] Dave Pengelley: yeah.
[11:27] Richard Webbe: I
[11:28] Matt Slager: impressive.
[11:28] Richard Webbe: noticed there was a lot of publicity this week about um pretty much AI and AI bots are now the largest traffic on the internet.
[11:39] Dave Pengelley: Yeah.
[11:39] Richard Webbe: Not just
[11:40] Joanna Jones: Yeah.
[11:40] Richard Webbe: our agents are doing it for
[11:41] Dave Pengelley: And
[11:41] Matt Slager: It's
[11:41] Richard Webbe: us.
[11:42] Matt Slager: an
[11:42] Dave Pengelley: then
[11:42] Matt Slager: ultimatum,
[11:43] Dave Pengelley: who
[11:43] Matt Slager: too.
[11:43] Dave Pengelley: I spoke to his latest technology synapse is basically a way of creating secure, private, zero trust pathways, agent to agent communications. Um crazy stuff.
[11:56] Joanna Jones: No, I look forward to hearing that one. That hasn't gone out
[11:58] Dave Pengelley: No,
[11:58] Joanna Jones: yet.
[11:59] Dave Pengelley: that goes out on Monday. So this with this week I just published Nathan Ahern, an Aussie bloke who has run various um he ran a huge MSP and built a bunch of technologies, and he's been part of sort of advising multi-million dollar acquisitions and sales and things. And now he's run around the world talking to family officers and investors and basically just sort of transferring innovation knowledge around doing talks and let's just living the dream. So yeah, amazing, amazing different variety of people we get. On the show.
[12:27] Joanna Jones: Have
[12:27] Matt Slager: I wonder
[12:27] Joanna Jones: you
[12:27] Matt Slager: if
[12:27] Joanna Jones: got
[12:27] Matt Slager: uh,
[12:27] Joanna Jones: a list of all your guests there, Dave?
[12:30] Dave Pengelley: Uh yes, you can find them at AIOperatorspod dot com and click on operator stories in the menu bar.
[12:38] Richard Webbe: Good
[12:38] Joanna Jones: Beautiful.
[12:38] Richard Webbe: segue. Well done, Joe.
[12:42] Dave Pengelley: Uh let me
[12:43] Richard Webbe: Hey um hey um just before I shut up, um Santa gave me some updates on how AOA is going to change Christmas this year. Would you like to hear them?
[12:54] Matt Slager: yes.
[12:55] Joanna Jones: I'm not sure.
[12:57] Shawn Hain: Okay.
[12:58] Richard Webbe: So the first one, of course, is shopping's going to be a lot easier. We send our bots out to profile our people and they'll go find the best bargain for the best thing.
[13:08] Dave Pengelley: I'm sorry about
[13:09] Richard Webbe: But
[13:09] Dave Pengelley: that actually.
[13:09] Richard Webbe: yeah, the next thing, of course, is everyone's going to be doing personalized AI slop, which no one will appreciate anymore because we've gone past that so quickly. So anyone under the age of five is allowed to produce AI slop, and that's cute. Anyone older than that, that's not cute. And you're just lazy and we hate you. So go away and don't even give me a present. But every time you do something customized and you write it out by hand or do it physically, you're going to get extra kudos because you actually did it yourself. Planning menus for people is going to be much, much better because you just put in vegetarian here, blah, blah, blah, and it'll come up with a nice lunchtime menu. And when you've got those guests that hate each other, and I was playing with this, you can rate and profile them. On where they sit, and it'll plan the seating at the table for you so you get less
[14:01] Shawn Hain: Oh
[14:01] Richard Webbe: arguments.
[14:02] Shawn Hain: so they're as close together as possible or as far away as
[14:06] Richard Webbe: Well,
[14:06] Shawn Hain: possible.
[14:07] Matt Slager: Or
[14:07] Richard Webbe: this
[14:07] Matt Slager: did
[14:07] Richard Webbe: is
[14:07] Matt Slager: it
[14:07] Richard Webbe: a
[14:07] Matt Slager: put you
[14:07] Richard Webbe: good
[14:08] Matt Slager: on your
[14:08] Richard Webbe: point.
[14:08] Matt Slager: own table, Richard?
[14:09] Richard Webbe: Sean, you are on the money. It depends on the outcome that you want. And I hear the dog's too close to someone. But the best one is I think we know the fight comes up between the family debates over what's happening in the world because we all haven't sat to each other for a while. Just leave Claude or Gros. Or whatever on in the background with voice, listening, and when someone speaks crap, it'll go, that's wrong, and the arguments will be like this long. They won't be like 20 minutes over turkey, it'll be like, Oh, might kill the personality of Christmas lunch. That was Santa's view of how AI Christmas is going to be this year.
[14:45] Matt Slager: I like it. So, with the seating arrangements, did it did it put you on your own table, Richard?
[14:50] Richard Webbe: Now that's very uh that's very funny of you, Matt. Um didn't you know that Dave and I do the jokes here? How are you doing?
[14:57] Matt Slager: I'm just stepping
[14:58] Dave Pengelley: He's he's
[14:58] Matt Slager: in.
[14:58] Dave Pengelley: me mixing it up, mate. Mixing it up. Um no, I mean that's that's a good point. And and I think you know, on on the topic of AI and Christmas and so on, there's something uh I would like to share with you all. What?
[15:12] Richard Webbe: Here
[15:12] Dave Pengelley: No,
[15:12] Richard Webbe: we
[15:12] Dave Pengelley: no,
[15:12] Richard Webbe: go.
[15:13] Dave Pengelley: look. Um coming soon. What's coming soon? Oh no, what is it? What is it, Dave? What is it? Anyone, any guesses? What's coming soon?
[15:22] Matt Slager: Is it a is it a marketing email list?
[15:26] Dave Pengelley: Dun dun d refresh. Why isn't this working? So why
[15:32] Shawn Hain: Live
[15:32] Dave Pengelley: is it got
[15:32] Shawn Hain: demos.
[15:32] Dave Pengelley: passed
[15:33] Richard Webbe: Coming
[15:33] Dave Pengelley: with it?
[15:33] Shawn Hain: Never
[15:33] Richard Webbe: soon,
[15:33] Shawn Hain: fun.
[15:34] Richard Webbe: a failed demo while
[15:35] Dave Pengelley: Hey,
[15:35] Richard Webbe: doing at lunch.
[15:36] Dave Pengelley: here we
[15:36] Richard Webbe: Oh,
[15:36] Dave Pengelley: go.
[15:36] Joanna Jones: I love tech difficulties that makes me feel old warm and fuzzy.
[15:41] Richard Webbe: yeah. Oh, we're getting merch. Are we getting our own merch for AI
[15:44] Dave Pengelley: Getting
[15:44] Richard Webbe: operators?
[15:44] Dave Pengelley: merch. It's
[15:45] Shawn Hain: No,
[15:45] Dave Pengelley: live.
[15:45] Shawn Hain: that's
[15:45] Dave Pengelley: I just
[15:46] Shawn Hain: cool.
[15:46] Dave Pengelley: might just made the side go live. This is not a political podcast. Uh you can wear this to your Christmas dinners.
[15:52] Shawn Hain: Yeah, I think I might need that.
[15:54] Dave Pengelley: Yeah. And we've got some Too, we've got that on a mug. I like this one. My agents on a smoke o. Um,
[16:00] Joanna Jones: Can I get a
[16:01] Dave Pengelley: and
[16:01] Joanna Jones: pink one?
[16:03] Dave Pengelley: uh uh
[16:03] Richard Webbe: Oh
[16:03] Dave Pengelley: the the
[16:04] Richard Webbe: gosh,
[16:04] Dave Pengelley: the contrast
[16:04] Richard Webbe: it's always
[16:05] Dave Pengelley: doesn't
[16:05] Richard Webbe: feminism
[16:05] Dave Pengelley: work well on pink
[16:06] Richard Webbe: coming
[16:06] Dave Pengelley: Joe.
[16:06] Richard Webbe: into
[16:06] Dave Pengelley: We'll
[16:06] Richard Webbe: it, but
[16:06] Dave Pengelley: talk to you offline, but uh by default we're just doing the blue and the black in the t-shirts. But yeah, we've got the robot on a smoke oh, we've got um the human in too many loops, and when I say human, I mean robot once again, because that's what we're going with.
[16:23] Richard Webbe: I'm I'm not sure what saying.
[16:23] Dave Pengelley: So yeah, just a few bits of fun merch.
[16:26] Shawn Hain: Vot.
[16:26] Richard Webbe: My book in the merch yet.
[16:27] Dave Pengelley: No, because that's on Amazon. I'm I might be able to list external products here if you really want it on here.
[16:33] Richard Webbe: Maybe
[16:33] Dave Pengelley: We can we can talk about that offline, uh Richard. But yes, we have a merch store. It's just got a few basic products out there. Uh you can switch to AUD. I think I've made the selected all the print on demand products to print out of Australia, so they should arrive here before
[16:47] Richard Webbe: how
[16:48] Dave Pengelley: Christmas.
[16:48] Richard Webbe: long did it take you to produce this stuff?
[16:50] Dave Pengelley: Uh the robots did it, mate. I had the corral and caress and and you know, cajole them, but They're
[16:56] Richard Webbe: Okay,
[16:57] Dave Pengelley: robots.
[16:57] Richard Webbe: some more questions. How did you source the t shirts?
[17:00] Dave Pengelley: Well, so there's there's a third-party print-on-demand service that does it all
[17:04] Richard Webbe: Is
[17:04] Dave Pengelley: for me.
[17:04] Richard Webbe: it
[17:04] Dave Pengelley: I just I just upload the designs and they've created the shopfront for me and they manage the ordering and
[17:11] Richard Webbe: China or local or
[17:13] Dave Pengelley: China. Uh no, no, it's it's they print local. So there's this network of print-on-demand printers and print factories around the world, and now there's all these printing services that sort of white label them and print your stuff on.
[17:26] Matt Slager: Very cool. They're pretty good prices. Like a lot of custom shirts like that are not usually in the twenty dollar range.
[17:31] Dave Pengelley: Yeah, like look, but they're they're we're not making heaps of money off them, but I just want to get them out there.
[17:37] Richard Webbe: Well, that might keep some money off anything.
[17:39] Dave Pengelley: No, no, right. I mean
[17:41] Joanna Jones: But
[17:41] Dave Pengelley: and
[17:41] Joanna Jones: you can gift every um person that comes on the podcast a t shirt and a mug, or something like that.
[17:46] Dave Pengelley: Yeah, yeah, as as we we we grow and and and get stuff, I'd like to do more of that kind of gear. Um but yeah, I think uh the political podcast is a great one for those Christmas conversations. Just remind people. At Christmas. It's good to have chats around what is going on. And I think, you know, I would encourage, like, we need to have reasoned conversation, but name calling and devolving is not what it is. We need we need like a little slap button whenever we do this is not a political conversation where we can just like push, push. Just that's all the channel would be.
[18:16] Joanna Jones: Or
[18:16] Matt Slager: I
[18:16] Dave Pengelley: Um
[18:16] Joanna Jones: you
[18:16] Matt Slager: have
[18:16] Joanna Jones: just
[18:16] Matt Slager: a
[18:16] Joanna Jones: get a
[18:16] Matt Slager: uh I do
[18:18] Dave Pengelley: I
[18:18] Matt Slager: I
[18:18] Dave Pengelley: didn't.
[18:18] Matt Slager: have something like for those chats, like you know, if your family are gonna be like, Oh, you know, you're the you're the weird AI one, you know, you're the one that's causing all the data centers. You
[18:27] Richard Webbe: Yeah.
[18:27] Matt Slager: know, it's such a good opportunity to try and like gently have those chats and and clear the truth.
[18:33] Richard Webbe: You
[18:33] Matt Slager: Because
[18:33] Richard Webbe: think you think the Christmas table is the way to have those chats?
[18:36] Matt Slager: well, I know that you like I will get attacked with that for
[18:39] Richard Webbe: I know
[18:40] Matt Slager: sure.
[18:40] Richard Webbe: I already
[18:40] Matt Slager: Yeah.
[18:40] Joanna Jones: Is
[18:41] Richard Webbe: know.
[18:41] Joanna Jones: it? Yeah.
[18:41] Richard Webbe: Yeah.
[18:42] Matt Slager: And like literally, I was just doing a bit of research before this today because my partner who used to work at Bunnings and her mum still does, they've recently had 350 layoffs, and and my partner. was like going, it's all because of AI, like you know, it's such a bad thing. I'm like, well, let me let me just try and dig a bit deeper. And uh it's
[19:04] Richard Webbe: This
[19:05] Matt Slager: not
[19:05] Richard Webbe: is not this is not a political podcast, but there's many other things causing layoffs in this country today that is nothing to do with
[19:12] Dave Pengelley: Well
[19:12] Richard Webbe: AI.
[19:13] Dave Pengelley: no, I I heard I heard great news. This is not a political podcast, but I heard great news from our prime minister that uh he's setting up a whole new bureaucracy for AI safeguards because we need more people employed by the government in government
[19:24] Matt Slager: a
[19:25] Dave Pengelley: jobs.
[19:25] Matt Slager: I'm bit gonna little more not than do it.
[19:27] Dave Pengelley: To tackle inflation
[19:28] Matt Slager: Angela.
[19:29] Dave Pengelley: and everything else.
[19:29] Richard Webbe: I
[19:29] Dave Pengelley: So
[19:30] Richard Webbe: always
[19:30] Dave Pengelley: this is
[19:30] Richard Webbe: think
[19:30] Dave Pengelley: not
[19:30] Richard Webbe: when
[19:30] Dave Pengelley: a
[19:30] Richard Webbe: I
[19:30] Dave Pengelley: political
[19:30] Richard Webbe: look at the
[19:31] Dave Pengelley: podcast.
[19:31] Richard Webbe: news, hey.
[19:33] Manjula Kerai: Hey, how's it
[19:34] Richard Webbe: Hey
[19:34] Manjula Kerai: going,
[19:35] Richard Webbe: Mandula,
[19:35] Manjula Kerai: everyone?
[19:36] Richard Webbe: how are you?
[19:36] Dave Pengelley: Welcome
[19:36] Manjula Kerai: Very
[19:36] Joanna Jones: Manchula,
[19:37] Manjula Kerai: well, how
[19:37] Dave Pengelley: back.
[19:37] Manjula Kerai: you doing?
[19:38] Joanna Jones: good morning.
[19:39] Manjula Kerai: Good to see you.
[19:41] Joanna Jones: It's
[19:41] Dave Pengelley: Ah.
[19:41] Joanna Jones: great to see you.
[19:44] Richard Webbe: You're
[19:44] Joanna Jones: I think
[19:44] Richard Webbe: on
[19:44] Joanna Jones: it
[19:44] Richard Webbe: the
[19:45] Joanna Jones: is.
[19:45] Richard Webbe: road.
[19:46] Manjula Kerai: I have parked up at IKEA.
[19:49] Matt Slager: Very
[19:49] Dave Pengelley: Ah,
[19:49] Matt Slager: good.
[19:50] Dave Pengelley: I know well.
[19:52] Matt Slager: Is it
[19:52] Shawn Hain: Always
[19:52] Matt Slager: is that a
[19:52] Shawn Hain: a
[19:53] Matt Slager: pitch?
[19:53] Shawn Hain: good choice.
[19:53] Matt Slager: Manjula,
[19:53] Joanna Jones: Yes.
[19:54] Matt Slager: you're gonna you're gonna pitch a solution there?
[19:56] Manjula Kerai: Oh, um nope.
[19:59] Dave Pengelley: Not a not a lot of teachers that I care.
[20:02] Matt Slager: No
[20:03] Richard Webbe: I bought an IKEA shop once, so I'm still trying to put it together.
[20:10] Dave Pengelley: Yes, whole shop. He bought the whole box and dice.
[20:14] Matt Slager: flicks.
[20:16] Matt Slager: Did the did it come with instructions?
[20:19] Richard Webbe: No, it did, but they're useless. However, I have used AI on those instructions and it's pretty cool. It makes up for those gaps of you don't know which way to put what where and when you get confused with the diagram.
[20:32] Matt Slager: You
[20:32] Manjula Kerai: Yeah.
[20:32] Matt Slager: could literally just ask Claude to make your own animation.
[20:37] Richard Webbe: Yes, no, all all of that. I had to turn the thing around and can this be I yep works
[20:41] Joanna Jones: Why don't
[20:42] Richard Webbe: well.
[20:42] Joanna Jones: they come with a nobi?
[20:44] Richard Webbe: Now
[20:45] Matt Slager: True.
[20:45] Richard Webbe: Joanna, that is a brilliant comment.
[20:48] Joanna Jones: Why not?
[20:49] Richard Webbe: Yes, a little
[20:50] Joanna Jones: Right.
[20:50] Richard Webbe: QR code.
[20:51] Joanna Jones: John Ansley, if you're
[20:53] Richard Webbe: Yeah,
[20:53] Joanna Jones: watching,
[20:53] Richard Webbe: super John
[20:54] Joanna Jones: take
[20:54] Richard Webbe: and Sam.
[20:55] Joanna Jones: note. There's
[20:56] Richard Webbe: I
[20:57] Joanna Jones: a
[20:57] Richard Webbe: agree.
[20:57] Joanna Jones: lead,
[20:58] Richard Webbe: There
[20:58] Joanna Jones: something
[20:58] Richard Webbe: is.
[20:58] Joanna Jones: to follow up.
[21:01] Dave Pengelley: Connecting all the dots from uh from the year of threads. Well done, Joe.
[21:04] Richard Webbe: You
[21:04] Matt Slager: We
[21:05] Richard Webbe: know,
[21:05] Matt Slager: we
[21:05] Richard Webbe: Joe,
[21:05] Matt Slager: can have
[21:05] Richard Webbe: I'm
[21:05] Matt Slager: like
[21:05] Richard Webbe: worried
[21:05] Matt Slager: a reference
[21:06] Richard Webbe: that there'll
[21:06] Matt Slager: for
[21:06] Richard Webbe: be
[21:06] Matt Slager: me
[21:06] Richard Webbe: QR
[21:06] Matt Slager: to think
[21:07] Richard Webbe: codes on on on patients when they go to the operating theater and I'll see the uh
[21:11] Matt Slager: about it.
[21:11] Richard Webbe: I'll see the surgeon clip on one of John's QR codes on the on the patient to know what to do
[21:16] Manjula Kerai: Oh
[21:17] Richard Webbe: to do
[21:17] Manjula Kerai: my
[21:17] Richard Webbe: to
[21:17] Manjula Kerai: gosh.
[21:17] Richard Webbe: them.
[21:18] Joanna Jones: No,
[21:18] Manjula Kerai: Instructions.
[21:18] Joanna Jones: it'll probably be more likely to have care instructions. I like to have a shave like this. I like to, this is my morning routine. This is how I like my coffee. This is what I do don't eat.
[21:29] Richard Webbe: Absolutely.
[21:30] Joanna Jones: Give you a just have a quick scan in the morning before you meet them.
[21:33] Richard Webbe: Mm-hmm.
[21:34] Matt Slager: Reminds me of uh shameless uh do not resuscitate
[21:42] Manjula Kerai: Oh my gosh.
[21:43] Matt Slager: Frank.
[21:43] Richard Webbe: Yes.
[21:44] Joanna Jones: Sometimes there's enough of them and sometimes there's not enough discussion around that.
[21:48] Richard Webbe: Yes.
[21:48] Joanna Jones: So
[21:49] Manjula Kerai: Yeah.
[21:50] Richard Webbe: Shameless.
[21:51] Manjula Kerai: It's
[21:51] Richard Webbe: What a great show, Matt.
[21:53] Joanna Jones: Uh,
[21:53] Richard Webbe: I I had a feeling you'd like a show like that. It is pretty cool.
[21:57] Matt Slager: Yeah, it's uh it it hits close to home that one.
[22:00] Dave Pengelley: Never watched it.
[22:01] Richard Webbe: Don't
[22:01] Dave Pengelley: I watched like
[22:01] Richard Webbe: say
[22:02] Dave Pengelley: half
[22:02] Richard Webbe: that.
[22:02] Dave Pengelley: the first episode and it wasn't wasn't my thing. I couldn't do it. I was like
[22:10] Matt Slager: I said it hits close to home, meaning I relate.
[22:14] Shawn Hain: I
[22:14] Dave Pengelley: we've been watching
[22:15] Shawn Hain: no
[22:15] Dave Pengelley: Silicon
[22:15] Shawn Hain: idea.
[22:15] Dave Pengelley: Valley, which has been very interesting. Obviously, a lot of it's all pre-AI stuff. So um
[22:21] Manjula Kerai: Oh
[22:21] Dave Pengelley: but still super fun. Well, they're all coding by hand, mate. They all accept right code. That's I think it's a different show today. Um yeah, the world is moving. Let's talk about some news.
[22:44] Dave Pengelley: So the big news I saw this morning was
[22:51] Dave Pengelley: this one.
[22:52] Matt Slager: Oh, yes. Is
[22:53] Dave Pengelley: Yes.
[22:53] Matt Slager: it really news?
[22:55] Dave Pengelley: No, it is I mean They new models, mate. New models are news, yay!
[22:58] Matt Slager: Six
[22:59] Dave Pengelley: And they've
[22:59] Matt Slager: like
[22:59] Dave Pengelley: got this
[22:59] Matt Slager: Sol.
[22:59] Dave Pengelley: this video with it where they're all using like 90s tech,
[23:02] Matt Slager: Actually
[23:03] Dave Pengelley: but
[23:03] Matt Slager: the video
[23:03] Dave Pengelley: that
[23:03] Matt Slager: was probably the most impressive thing about the whole release. Yeah.
[23:08] Dave Pengelley: like this is a fun video. I don't think they're trying to take on the whole Microsoft thing with the Excel and the Lyft. Um,
[23:13] Matt Slager: Yep.
[23:13] Dave Pengelley: have you seen that ad? The old Microsoft
[23:15] Manjula Kerai: Yeah.
[23:15] Dave Pengelley: one.
[23:16] Matt Slager: Well,
[23:17] Dave Pengelley: Um
[23:17] Matt Slager: that's basically what this astro model is is for. It's that office assistant. It's not a software engineer like previously they've tried to explain. That's definitely what Sol is.
[23:27] Shawn Hain: Yeah,
[23:27] Dave Pengelley: Yeah. Yeah, I thought
[23:29] Shawn Hain: yeah.
[23:29] Dave Pengelley: uh the video video was fun. It's weird seeing like modern web interfaces on these ancient laptops and CRTs.
[23:35] Matt Slager: Yeah.
[23:35] Dave Pengelley: That just that's jarring for some reason.
[23:38] Matt Slager: I liked the um the the like the story of that video as well. Like whoever wrote that, like they did a good job.
[23:45] Dave Pengelley: Yeah.
[23:45] Matt Slager: Especially the one in the in the conference room with this this guy fumbling for the cables.
[23:49] Dave Pengelley: We're
[23:49] Matt Slager: And
[23:50] Dave Pengelley: finding
[23:50] Matt Slager: like they
[23:50] Dave Pengelley: the
[23:50] Matt Slager: don't
[23:50] Dave Pengelley: cable.
[23:51] Matt Slager: they don't have their presentation ready yet and you know it just gets ready in time. It's pretty good.
[23:56] Manjula Kerai: Um,
[23:56] Dave Pengelley: So I I remember going into a conference room once, and this was before they had L C D panels everywhere, and there was still a lot more overhead sort of projectors in rooms, and VGA was still a thing. Uh, and I didn't bring a VGA adapter to one of them. I think just was like only had my HDMI, and it's like, do you have no? Oh, okay. Um, all right, gather around my laptop. We're gonna look at my laptop while I show you this thing because I cannot plug into your presentation system. Uh the the other the other one. They got me a few times was when you get there and they go, Oh no, we have a computer all plugged in. Can you just use our computer that's plugged into our screen? It was like, sure. Our solutions in the cloud. I can ignore the fact I had eight prep browser tabs ready to go bang, bang, bang through my demo flow. I will just log into your browser. That is that is good.
[24:48] Richard Webbe: God help
[24:49] Shawn Hain: Yeah.
[24:49] Richard Webbe: me. I did
[24:50] Shawn Hain: Yeah.
[24:50] Richard Webbe: a demo in my lounge room to the head of engineering of Australia's largest telco. with all these things because our office couldn't handle for some reason their network wasn't so I took them all to my house. My dogs licking their toes are making them copy while the demos on my TV.
[25:11] Shawn Hain: Yeah.
[25:13] Joanna Jones: I think in twenty twenty four when I was um teaching at university, um educating there, I went in, they'd gave us a brand new I think I I may have mentioned this before, but they gave this brand new room with all the brand new tech and I think I might have had 30 minute pre-training and I walked
[25:31] Richard Webbe: Yeah.
[25:31] Joanna Jones: in and I could not get any of it working and it was impressive. It was great.
[25:41] Matt Slager: Okay.
[25:43] Dave Pengelley: Oh. Oh no, Joe.
[25:48] Richard Webbe: Let's all freeze.
[25:53] Shawn Hain: Yeah,
[25:53] Richard Webbe: Joe will
[25:53] Shawn Hain: okay.
[25:53] Richard Webbe: be back. Joe will be back.
[25:56] Matt Slager: loading there's actually a comment about the anthropic ipo
[26:01] Dave Pengelley: Ah
[26:02] Matt Slager: yeah
[26:02] Dave Pengelley: Sergio,
[26:02] Matt Slager: saw that they actually
[26:04] Dave Pengelley: what do
[26:04] Matt Slager: announced
[26:05] Dave Pengelley: you think
[26:05] Matt Slager: some
[26:05] Dave Pengelley: about
[26:05] Matt Slager: prices
[26:05] Dave Pengelley: the anthropic IPO? Looks like it will be well over one T
[26:11] Matt Slager: if it sound feels we're typing that like i was like how do i write run trillion i'm like is it a t
[26:17] Shawn Hain: Yeah.
[26:18] Dave Pengelley: is it's a T T for
[26:21] Shawn Hain: And
[26:21] Dave Pengelley: Trillium.
[26:22] Shawn Hain: yeah, that's huge.
[26:24] Matt Slager: it's nuts
[26:25] Shawn Hain: Yeah.
[26:25] Matt Slager: oh seven
[26:26] Richard Webbe: You're back,
[26:26] Matt Slager: Seventy eight
[26:27] Richard Webbe: Joe.
[26:27] Matt Slager: dollars possibly gonna go bad for eighty dollars. So yeah, e equating to like one point three to one point six
[26:33] Dave Pengelley: It's
[26:33] Matt Slager: trillion.
[26:33] Dave Pengelley: it's all a Ponzi scheme though, right? Isn't it?
[26:37] Matt Slager: Oh double.
[26:38] Dave Pengelley: Oh, she's in twice now.
[26:43] Dave Pengelley: All right.
[26:44] Shawn Hain: There
[26:44] Richard Webbe: There
[26:44] Shawn Hain: we go.
[26:44] Richard Webbe: you go.
[26:46] Dave Pengelley: Okay.
[26:47] Shawn Hain: Yeah,
[26:47] Dave Pengelley: The
[26:48] Shawn Hain: that's
[26:48] Dave Pengelley: the joys. Oh no, she's gone.
[26:50] Richard Webbe: You killed
[26:50] Dave Pengelley: Ah.
[26:51] Richard Webbe: her.
[26:52] Dave Pengelley: Oh, I'll bring back her other one. Is that
[26:54] Richard Webbe: Welcome
[26:55] Dave Pengelley: in?
[26:55] Richard Webbe: back, Joe. Just
[26:56] Shawn Hain: a good thing.
[26:56] Richard Webbe: Keep talking, Joe, don't read Ivy keeps cutting you in and air.
[26:59] Dave Pengelley: What? It's not me. It wasn't me.
[27:04] Dave Pengelley: I can do it to Richard. I should do it more often. Here we go. What is
[27:10] Richard Webbe: I agree
[27:10] Manjula Kerai: Oh.
[27:10] Richard Webbe: with that.
[27:12] Manjula Kerai: Poor Richard.
[27:14] Dave Pengelley: Uh
[27:16] Dave Pengelley: Here we go. He's bad.
[27:22] Dave Pengelley: Oh my.
[27:23] Matt Slager: What
[27:23] Dave Pengelley: Oh
[27:23] Matt Slager: a professional
[27:23] Dave Pengelley: my.
[27:24] Matt Slager: episode.
[27:24] Dave Pengelley: You can you can tell it's a silly season. It is we're all getting a bit silly. It's been a long year. Um, on other news, any other big news? Obviously, the the Jensen stuff's happened in the last week, and video have said he's not coming back.
[27:39] Matt Slager: Yeah, it's crazy. Hey. I was not half expecting it to be like a Sam Alton thing where, you know, he gets booted and then everyone sort of votes him back in. Um, but I guess this one's a little bit more on like a a behavioral, you know, moral sort of reasoning, which makes sense. Um but
[27:55] Shawn Hain: Yeah.
[27:55] Matt Slager: yeah, I wonder if I wonder if Jensen's up for an interview. Uh an operator stories, Dave. You might be able to hit him
[28:00] Dave Pengelley: Here
[28:00] Matt Slager: up.
[28:00] Dave Pengelley: hear the hear the juicy details about Wangate. I mean Wangate. I mean juices Jensen's Wang I uh
[28:07] Richard Webbe: This is not a sex podcast.
[28:09] Shawn Hain: Okay,
[28:09] Matt Slager: It's
[28:09] Richard Webbe: Stop
[28:09] Matt Slager: le
[28:09] Richard Webbe: it.
[28:09] Matt Slager: this this is like less directly related to you know AI operators,
[28:13] Dave Pengelley: if you
[28:13] Matt Slager: but
[28:13] Dave Pengelley: say
[28:13] Matt Slager: the the first Starship
[28:15] Shawn Hain: yeah.
[28:15] Matt Slager: capture, that that was pretty cool. You know, the the successful launch and the successful capture with the with the
[28:20] Dave Pengelley: the
[28:21] Matt Slager: chopsticks.
[28:21] Dave Pengelley: chopsticks? Oh, was that like that that happened this year? But was that this year?
[28:25] Matt Slager: What?
[28:26] Dave Pengelley: What
[28:26] Matt Slager: No,
[28:26] Dave Pengelley: the
[28:26] Matt Slager: that
[28:27] Dave Pengelley: chap
[28:27] Matt Slager: was
[28:27] Dave Pengelley: uh
[28:27] Matt Slager: like
[28:27] Dave Pengelley: chops chopsticks that was last year.
[28:29] Matt Slager: it was a couple of days ago.
[28:30] Shawn Hain: Yeah.
[28:31] Dave Pengelley: Is it different from the rockets that have been descending? Uh, because he's
[28:34] Richard Webbe: It's
[28:34] Dave Pengelley: been catching
[28:34] Richard Webbe: a
[28:34] Dave Pengelley: the
[28:34] Richard Webbe: different
[28:34] Dave Pengelley: booster
[28:35] Richard Webbe: catching
[28:35] Dave Pengelley: rockets for a
[28:35] Richard Webbe: technology,
[28:35] Dave Pengelley: while.
[28:36] Richard Webbe: Dive.
[28:37] Matt Slager: Wouldn't
[28:37] Richard Webbe: It's a
[28:37] Matt Slager: it's
[28:37] Richard Webbe: different
[28:37] Matt Slager: the
[28:37] Richard Webbe: one.
[28:38] Matt Slager: first time that the uh a starship has been successfully recovered,
[28:41] Dave Pengelley: Okay,
[28:41] Matt Slager: which is the
[28:41] Dave Pengelley: I haven't
[28:42] Matt Slager: top
[28:42] Dave Pengelley: I've
[28:42] Matt Slager: of
[28:42] Dave Pengelley: haven't been staying across Elon's giant rockets
[28:45] Matt Slager: But
[28:45] Dave Pengelley: lately.
[28:46] Matt Slager: yeah, that would be directly AI related with all the data telemetry that they capture, you know, being able to process and iterate as fast as they have. You know, there's
[28:55] Shawn Hain: Yeah.
[28:55] Matt Slager: no way they would have been able to do that. without it.
[28:57] Shawn Hain: No,
[28:58] Dave Pengelley: Yeah,
[28:58] Shawn Hain: no way.
[28:58] Dave Pengelley: but
[28:59] Shawn Hain: Just no.
[29:01] Matt Slager: Yeah.
[29:03] Matt Slager: And
[29:03] Shawn Hain: Even just
[29:04] Matt Slager: um
[29:04] Shawn Hain: adjusting the simulation
[29:04] Matt Slager: the the the
[29:05] Shawn Hain: takes
[29:05] Matt Slager: actual
[29:05] Shawn Hain: forever.
[29:06] Matt Slager: model release that is you know really made a difference is I would say the fable. You know the the new fable 5.2 like you know at the start of the year it was probably when I personally was tapering off Claude and uh you know spent half the year just being a GPT person and now I'm very much on the other end of that and it's Seems like anthropic, you know, IPO aside, they're on a generational run, they just keep winning. And uh I I don't I I use Sol a lot and 6.2 Sol is definitely a meaning and full improvement, but Astra is just a joke. I don't understand.
[29:41] Dave Pengelley: I find I find Astra is better than than Sol on some things. Like the definitely. Like when I'm doing like some of the game dev and other bits and pieces I'm doing uh and the 3D stuff, I think Astra is definitely better than Sol. And I will switch to Astra deliberately. Like I I'm keen to see significant. six point two Astra and see what a level up it is because they're saying it's better for um game like UI and different the design interfaces and stuff, which it's been a huge gap in the Jipity model so far.
[30:06] Matt Slager: Yep. I think
[30:07] Richard Webbe: For
[30:07] Matt Slager: the best
[30:07] Richard Webbe: us
[30:08] Matt Slager: way to
[30:08] Richard Webbe: mere
[30:08] Matt Slager: use it is
[30:08] Richard Webbe: for
[30:09] Matt Slager: as
[30:09] Richard Webbe: us
[30:09] Matt Slager: a sub
[30:09] Richard Webbe: mere mortals,
[30:09] Matt Slager: agent.
[30:10] Richard Webbe: for us mere mortals, give us a use case comparison of the difference.
[30:15] Matt Slager: Between just those two.
[30:16] Richard Webbe: Well between any, what what you're experiencing, give us a use case comparison.
[30:20] Matt Slager: Well I think of like hu like humans communicate, like we we need to have better communication. For us to think well. You know, if you're
[30:28] Richard Webbe: Yes.
[30:28] Matt Slager: speaking to someone and they're not making sense, it just feels weird and it's kind of like a disconnect and you kind of feel like walking away. Um,
[30:33] Richard Webbe: Yeah.
[30:34] Matt Slager: that's what it's like working with Astra. So that the best way to use it right now, uh objectively, from at least mine and anecdotal other people's experience, is you speak to Fable or you speak to Opus, and then they use Astra as a subagent for those things that Dave's just said are,
[30:53] Richard Webbe: Yes.
[30:53] Matt Slager: you know, their strong suits. So like game development, 3D stuff. stuff, uh, office related things like they displayed in their launch video.
[31:01] Richard Webbe: So it's a horses for courses kind of supply chain where this is better at capturing your logic, this is better at calculating those things, this is quicker to do that. Is that the sort of thing we're talking about?
[31:11] Matt Slager: Yeah, yeah. The best way to describe it is when uh you would have been using Opus 5 before they did the the the 5.5 release
[31:19] Richard Webbe: Yep.
[31:19] Matt Slager: and it was terrible to communicate with
[31:22] Richard Webbe: Yeah.
[31:22] Matt Slager: that that's the same thing that Astra's still doing. It's very much GPTism It speaks in its own language. The responses that you get back, you have to spend like 10 minutes trying to figure out what's even saying. Um so yeah, that that's basically the experience of using it right now.
[31:38] Richard Webbe: I agree. You know, when I was in maths and I first discovered simultaneous equations, I got so excited because now I could calculate something to a point with multiple inputs changing at the same time. And so those that understand their maths understand how that was really. Exciting because I get these simultaneous equations and use them to actually create an answer, where in the past it'd be just trying to find a spot on a graph. And I find some of the new AR models exciting as that. And uh, how do I describe that? I was trying to solve a problem with multiple inputs. And not only was my I was I was using multiple LM, but not only was it finding the answers for me through effective research, but it was suggesting. back to me the calculations I should be doing that it thought I was missing out on and I was most impressed.
[32:32] Matt Slager: That's cool. Yeah, I think those moments where you feel feel the AGI when uh you have those like those pushback moments, it's collaborative, it's kind of brainstorming, it's have you suggested this? It's not just you know a vague question that you've asked and then the wrong answer because it's just trying to answer your question.
[32:51] Richard Webbe: Exactly. So I found it lately. I don't know if it's my slow brain or it's actually actually doing it but they're much more collaborative and cooperative, quicker and more intuitive.
[33:02] Matt Slager: Hmm.
[33:02] Richard Webbe: And uh I was well, you know, I was playing around some pretty important things to me and it was really good.
[33:08] Matt Slager: I just think it's funny. Like in February, you know, I I had a couple of little pithy statements here to pull out, but like in February we had a a chat window. We were using cord code in the terminal or whatever it may be. And today I did a account before jumping onto this. It's probably gone up since now. Today I have 12,439 AP. Agent threads running through T3 code. And yeah, it's ridiculous. Like the workflow constantly like upgrading over time, going from, like you said, Richard, like some sort of simple question and answer to then pushback, enrichment, you know, brainstorming collaboration. And now you've got the ability to actually fan that out within an agent swarm to then come back to you with even more powerful responses. Like the ability to get stuff done just keeps increasing. It's nuts.
[33:57] Richard Webbe: And it's like when you first uh not only not only so when you first discover for those that are mass people integral calculus and your lecturer says to you, Oh Richard, we're finding a point on a graph, but the challenge is the point on the graph is infinitely small. So there's no definitive answer. So we need to come up with a little formula that describes where that point would be in certain conditions. And you go, Oh wow, I didn't know mass could do that. And AI is doing the same thing now, it's fabulous. You're
[34:26] Matt Slager: I find
[34:26] Richard Webbe: grimacing,
[34:27] Matt Slager: like
[34:27] Richard Webbe: Dave. Did I sound as ignorant as I am, or
[34:30] Matt Slager: No,
[34:30] Dave Pengelley: No,
[34:30] Matt Slager: you sound
[34:30] Dave Pengelley: you're
[34:31] Matt Slager: very
[34:31] Dave Pengelley: you
[34:31] Matt Slager: intelligent.
[34:31] Dave Pengelley: you sound like you know a lot more about maths than I do. I've forgotten pretty much all of my high school math math. Even uni, I did uni level calculus in computer science, and what you're saying just sounds like witchcraft to me. It's bizarre.
[34:43] Matt Slager: I actually genuinely miss calculus. I don't know why. I
[34:46] Dave Pengelley: Oh
[34:46] Matt Slager: just had like
[34:47] Dave Pengelley: man.
[34:47] Matt Slager: a bit of a nostalgia
[34:47] Richard Webbe: I don't
[34:48] Shawn Hain: All
[34:48] Matt Slager: feeling.
[34:48] Shawn Hain: right.
[34:48] Richard Webbe: know.
[34:49] Dave Pengelley: I mean, man, Jula, how how's your how's your uh uh um journey been lately with the AIs and the different models and stuff? What do you Using nowadays?
[34:58] Manjula Kerai: I'm still um using what I've been using. So um what's it called? With Claude, I haven't actually checked out the new models.
[35:08] Dave Pengelley: Hmm.
[35:08] Manjula Kerai: But the thing is, like it can what I'm using it for, it can do all of that right now. So I haven't actually looked at the new, you know,
[35:17] Dave Pengelley: Yeah,
[35:17] Manjula Kerai: the new
[35:17] Dave Pengelley: fair
[35:17] Manjula Kerai: stuff.
[35:18] Dave Pengelley: enough.
[35:18] Manjula Kerai: But
[35:18] Dave Pengelley: Which
[35:18] Manjula Kerai: I mean, yeah, I will I will check it
[35:20] Dave Pengelley: I
[35:20] Manjula Kerai: out.
[35:20] Dave Pengelley: mean,
[35:21] Manjula Kerai: Um
[35:21] Dave Pengelley: so it's one of the things we we we say on this show all the time, though don't don't taste shiny objects for the sake of shiny objects. If the tool you're using Using
[35:26] Manjula Kerai: Yeah.
[35:26] Dave Pengelley: does the job you needed to do, then that's doing the job you needed to do. If you're getting the results you need, if you're being able to aspire and be creative and ask big questions and it's getting the result without you pulling your hair out and going crazy and running round in circles, which you know, Matt Richard, you asked about the difference between some of the models. Some of the models is difference is you can get there, but you're gonna have to wait an extra five hours after you go through 30 iterations versus do
[35:49] Richard Webbe: Yes.
[35:50] Dave Pengelley: the thing. Oh, good. There's there's the thing. It's done. Like first shot. Like that, that's one of the differences.
[35:56] Richard Webbe: them Dave moments and I get them all the time now so I tell my my AI to do this for me and it goes I can do that and I'm thinking yeah Dave said this happens so just do it don't tell me you can do it just
[36:07] Dave Pengelley: Yeah.
[36:08] Richard Webbe: execute your moron
[36:09] Dave Pengelley: I've been I've been
[36:10] Shawn Hain: Yeah.
[36:10] Dave Pengelley: I've been leveraging Sir Patrick Stewart a lot lately and just saying make it so. Make it so.
[36:16] Dave Pengelley: Um been been excellent.
[36:18] Matt Slager: That's
[36:18] Dave Pengelley: Uh
[36:19] Matt Slager: a that's another like question that I had. Like you guys sort of gone through a few um changes throughout the year. Like what's what's and how what's One habit that you started, or maybe you were aware of it, maybe you're not, um, that you're gonna keep doing.
[36:32] Dave Pengelley: Well hang on. If we're if we're talking about the year as a whole, can I try a thing? I don't know if I've got the right things here. Let's try this. That that's that that's not the right thing.
[36:47] Dave Pengelley: Try one, hang on. I've got new things. Hang on.
[37:03] Manjula Kerai: Nice.
[37:03] Richard Webbe: Both
[37:03] Matt Slager: Even
[37:04] Richard Webbe: done.
[37:04] Matt Slager: with the old sound, too, that was perfect.
[37:07] Shawn Hain: Yeah.
[37:07] Manjula Kerai: And the kangaroo,
[37:08] Dave Pengelley: Yeah,
[37:08] Manjula Kerai: love it.
[37:09] Dave Pengelley: that's um I try I tried various versions of getting different different AIs to to build versions of that and they were rubbish. I ended up getting Hermes to stick that out through foul through um through my news portal sub because Grok and that were just not getting the job done. But Foul. I'll nail that first time.
[37:28] Manjula Kerai: Nice one.
[37:29] Richard Webbe: Joe, you're back. We missed you, darling.
[37:32] Dave Pengelley: So now
[37:33] Joanna Jones: Did did someone censor me? Was I talking about the union?
[37:40] Richard Webbe: Joe,
[37:40] Joanna Jones: I literally
[37:41] Richard Webbe: if
[37:41] Dave Pengelley: we
[37:41] Richard Webbe: censoring
[37:41] Joanna Jones: just
[37:41] Dave Pengelley: we
[37:41] Joanna Jones: got shut out.
[37:43] Richard Webbe: if
[37:43] Dave Pengelley: you've
[37:43] Richard Webbe: censoring
[37:43] Dave Pengelley: you've
[37:44] Richard Webbe: was a thing, I wouldn't make it on at all. So no one censored you.
[37:48] Dave Pengelley: you've joined back just in time for wins of the year. We're talking about like the habits of change, things that are that we're doing big, big events, things in the year that with AI that We've we've had, you know. Oh,
[37:58] Joanna Jones: The broadcast.
[37:58] Dave Pengelley: Joe's signal, Perth.
[38:01] Shawn Hain: Yeah. Um I guess for me in terms of habits, I think Matt you asked the question originally, the biggest one is making sure when the AI is getting creative, but it's not getting creative in a solved space.
[38:14] Matt Slager: Yeah.
[38:14] Shawn Hain: Like you don't pay attention, it goes off on a tangent, like I'm gonna reinvent the wheel. I'm
[38:20] Richard Webbe: So,
[38:20] Shawn Hain: like, but there's ten wheels to choose from, just use the wheel. You don't need to make your own. I had one the other week where it
[38:26] Richard Webbe: I a good question. that's think
[38:26] Shawn Hain: tried to recreate celery.
[38:28] Matt Slager: Yeah, wow.
[38:29] Shawn Hain: Which is a huge management thing for those of you that aren't overly technical. It's like, why are you doing this? Just
[38:36] Matt Slager: Yeah.
[38:36] Shawn Hain: I don't see if even in the spec to use celery.
[38:39] Matt Slager: My
[38:39] Shawn Hain: And
[38:39] Matt Slager: favorite
[38:39] Shawn Hain: you know
[38:40] Matt Slager: is when you have a solution like that and then you know it's perfect. It's bespoke, it's exactly what you want. And you go, I'm thinking of using this, and they go, Hmm, it looks like it's a bit of a young project. It's not uh validated yet. Like let's just make up our own instead. Like
[38:56] Shawn Hain: Yeah. You're just like no, let's let's not
[39:01] Matt Slager: Yes.
[39:01] Dave Pengelley: Even the even that video, like I tried Grok, but I tried a few different things, and Grok imagined it to make it. Uh, and then I tried Chat GPT to make it, and it tried to put me in Chat GPT work to make it, like, wouldn't make it in pure chat. So I wanted to burn some tokens. I was like, okay, do it. I ended up with four iterations, they all sucked. And then I go into Hermes and I'm watching the the traces in Hermes, and it starts to do like draw like drawing path libraries and stuff, and I was
[39:26] Shawn Hain: a get lot of things.
[39:26] Dave Pengelley: like don't make it out of SVGs
[39:29] Shawn Hain: Oh, wow.
[39:30] Dave Pengelley: use your video generation tool go use foul and then it comes back and goes oh you're right the correct way to do this is to use a video generation tool are you kidding the the average the average cost of using foul is two dollars i don't know what it is through news portal do you approve one render yes
[39:48] Richard Webbe: I I like you said, Sean, I find and Matt's asking, I find one habit I do now, of course, is I go straight to my AI in the morning. Help me make sure I've got
[39:57] Manjula Kerai: Prince
[39:57] Richard Webbe: my day in sync and
[39:59] Manjula Kerai: Prince
[39:59] Richard Webbe: my little bit of an ADHD mind needs to be reminded of
[40:03] Manjula Kerai: Prince
[40:03] Richard Webbe: where I'm going and what I'm focusing on for the day.
[40:05] Manjula Kerai: Prince
[40:06] Richard Webbe: The other new habit that I have is getting my AI
[40:08] Manjula Kerai: Prince
[40:09] Richard Webbe: to remember to cross-reference itself. So I'll tick the memory button you guys told me to tick, and hopefully all the chats are connecting, but they don't always do that. And so I
[40:19] Manjula Kerai: Prince.
[40:19] Richard Webbe: get a bit frustrated, like Sean and Dave, and it says, Oh, I wonder what that was. And I go, well, we spoke about that 10 minutes ago, Moron. Do it. And they go, oh, you're right. Sorry about that. Let me go and check that chat. So there's a little bit of that. I'm I'm hoping as these models get better, we'll get less and less of that.
[40:37] Dave Pengelley: It's it's weird. I mean, I I talked about in the field note on Friday where I yelled at my AI and told it don't be lazy because we're just doing these minor incremental things. Like Sean, you said don't reinvent the wheel when you're in s solved space, but there's also the problem where it's built a version one and then locks itself into that version one so hard. It won't
[40:57] Shawn Hain: Yeah.
[40:57] Dave Pengelley: expand the borders. And so you know how you haven't solved this. This is not the right thing. That was a MVP draft. Stop doing tiny iterations. I want a full refresh. And yeah, you've just got to sort of
[41:13] Matt Slager: It's
[41:13] Dave Pengelley: call
[41:13] Matt Slager: funny
[41:13] Dave Pengelley: it out.
[41:14] Matt Slager: like that that thing there, Dave, I think is actually really dangerous because a lot of people get lulled into a false sense of security with how good the models are getting.
[41:23] Dave Pengelley: Hmm.
[41:23] Matt Slager: And if they go and do that and you're not on your toes and you don't have Have your sort of skepticism radar and your intuition on point to actually push back. And they say, Do you want to get this thing shipped? And you're like, Yeah, yeah, yeah, yeah. I just get it done. But it's not what you wanted. And
[41:37] Dave Pengelley: No.
[41:37] Matt Slager: it could be something entirely destructive, you know, especially if it's client work or working in a company. So yeah, it's a bit spooky.
[41:46] Dave Pengelley: You mean you mean telling you not to be lazy is dangerous, or just letting it do the tiny incremental iterations that aren't going anywhere? Or both.
[41:54] Matt Slager: Is you being on point and actually into intuitively knowing
[41:57] Dave Pengelley: Yeah,
[41:57] Matt Slager: that something's
[41:57] Dave Pengelley: yeah.
[41:57] Matt Slager: wrong. But yeah, like that that's something that I've been feeling a lot because I have, you know, this crazy orchestration system and I get super cognitively burnt out with decision fatigue. Right. And I wake up in the morning and I've got 20 things from each bot in my Grok bot. You know, like, you want to ship this?
[42:15] Dave Pengelley: You
[42:16] Matt Slager: We've
[42:16] Dave Pengelley: need
[42:16] Matt Slager: got this
[42:16] Dave Pengelley: the human
[42:16] Matt Slager: thing.
[42:16] Dave Pengelley: in too many loops, Mugg.
[42:18] Matt Slager: Yeah, pretty much. And yeah, all of it, I'm like, mate, ask me one thing at a time so I can actually intuitively comprehend what you're asking.
[42:26] Dave Pengelley: Yeah.
[42:26] Matt Slager: So that
[42:27] Dave Pengelley: It's
[42:27] Matt Slager: I can, you know,
[42:27] Dave Pengelley: it.
[42:27] Matt Slager: give you proper feedback.
[42:29] Dave Pengelley: I I don't know whether I'm getting stupider or what's going on, but like that cognitive load thing with AI agent decision making is real. And thank you, Richard, for acknowledging that I probably am getting stupider. Um the I've been doing lots of game development stuff and I'm trying to work out which ones I can get to market and which area. Like I'm pivoting and I've got some other um things on the horizon,
[42:50] Manjula Kerai: Oh,
[42:50] Dave Pengelley: but a lot of my AI time at the moment is is in game dev of all things, uh not client work. Uh but But I have so many ideas and I keep getting game ideas and keep getting game ideas. And I got another one on the weekend and I started working on it. And the problem with games, games are ridiculously complex, not just rules and mechanics and engines, but then if as their dialogue, narrative, story, all the pros that most games need, unless it's just a simple like button mashing arcade type thing. Um, and so I built this one and I thought, oh, this is it, this is it, this is super chill, cozy, like no stress game. And then I realized, oh, there's a lot of dialogue that needs to be written. And it had sort of sampled out 30 pre-written little segments for me. And I was like, I am overwhelmed by the 30 things that I didn't write and I have no ownership of. It comes back to my robots from before my IRL talk. If there's all this content and data that I have no emotional connection to and it just created it, it's really hard for me to objectively sit through, review, and understand and connect all the threads, especially if there's multi-tiered dialogue trees and things. I'm like, you need to clean
[43:59] Richard Webbe: Imagine
[43:59] Dave Pengelley: slate this for me. And
[44:01] Richard Webbe: imagine
[44:01] Dave Pengelley: I need
[44:01] Richard Webbe: yeah, imagine the complexity around Grand Theft Auto Six, five thousand people, thirteen years, one point five billion dollars.
[44:12] Dave Pengelley: Exactly, but there's 5,000 people. And there are people whose job is just story. And there's people that are just how do cars drive. And there's a person whose
[44:20] Richard Webbe: Exactly.
[44:20] Dave Pengelley: job is just the the how do we make people dance. Like that's so disaggregated.
[44:25] Richard Webbe: Hey, Phil.
[44:26] Dave Pengelley: Whereas when you're the one person trying to make all the decisions about all the look, the feel, the tone, the mechanics, the simulations, the dialogue, and the narrative and the character development, it's a lot. So I was like, clear this, clear the books. I need to think about from a like, I don't want AI slop this game. And so I need to just create my own cards, my own dialogue trees from scratch, starting with the character. What is that character about? What are they offering the player? And I just need an editing. tool give me a mobile friendly tool where I can create the dialogue on the go and I'm sitting there going I like the world building big picture strategy stuff that's why I was good as a leader I don't want to be doing all the in the weeds writing
[45:09] Richard Webbe: Hey
[45:09] Dave Pengelley: character
[45:10] Richard Webbe: um
[45:10] Dave Pengelley: dialogue
[45:10] Richard Webbe: while while you were chatting on that, Phil just put a great little tip in the comments you might want to paste up. I thought it was a good comment on double
[45:18] Matt Slager: I got many thoughts
[45:19] Richard Webbe: checking.
[45:19] Matt Slager: on that. Feel always comes in with uh good advice. I think we're on similar
[45:22] Dave Pengelley: and
[45:22] Matt Slager: wavelengths.
[45:23] Dave Pengelley: one one one day's the the the uh the the clouds We'll align the cloud stars and we'll get we'll get Phil on the show. Uh, because he does his own IoT podcast as well. Um, yeah. My tip for the day is build SFPs, install them so agents check the correct activity workflows, have a knowledge manager agent that updates changes, you'll find better runs, improved efficiencies. You keep training them and they follow processes, not rebuild routines every time.
[45:47] Matt Slager: So like there's so many ways to look at this, but the simplest, simplest way to look at this is it's literally just inputs. It's noise that the agent harness can pick up on. And I has having this chat with someone um yesterday actually. I think it's fascinating that we all can have the exact same Same AI subscription, you know, use the exact same model and the exact same reasoning. And because of the environment that each of these harnesses are actually operating under, you know, the fact that they can get exposed to these SOPs that Phil's talking about or the knowledge base, you know, one file change on your system, like even just it could be the same file net labeled differently, you know, it's just adding to the entropy generator, you know, like
[46:31] Dave Pengelley: Yeah.
[46:31] Matt Slager: every single person's version of the exact same thing. Actually will produce different results, um, which is fascinating.
[46:38] Richard Webbe: I think
[46:39] Matt Slager: And
[46:39] Richard Webbe: it's
[46:39] Dave Pengelley: I
[46:39] Richard Webbe: kinda
[46:39] Dave Pengelley: think
[46:39] Richard Webbe: cool. I think
[46:40] Dave Pengelley: and
[46:40] Richard Webbe: that's kinda cool.
[46:41] Dave Pengelley: and the and the legacy aspect of it, and I know we've discussed, you know, the the feedback from Anthropic, the best thing you can probably do is delete your claw.md file uh in the today's era because it's probably got a lot of legacy instructions that the earlier models were dumber and needed that much hand holding. New ones, you're actually just prescribing bad activities for it. And
[47:00] Matt Slager: yeah.
[47:00] Dave Pengelley: I think with some of my things, especially in Hermes, as I've iterated this thing from anti-gravity back in January when we first started this, through into clawed code workflows, now into Hermes workflows, I've got some bad junk in there, even just the way it uses plane, which uh was a migration from when I was using linear. And so I've got all these like overly specific instructions where it just guardrails things sometimes and goes, No, I can't do that. I need your specific. I'm like, no, you don't. Why why do you need that? And it's just interpreted like these layered rules for through various sort of skill and agent definitions to put all this sort of baggage in.
[47:38] Matt Slager: You said the magic word there, like you said, skill. Um, from what what Phil's talking about there, I believe literally can just be baked into your skill systems.
[47:46] Dave Pengelley: Yeah.
[47:46] Matt Slager: But much like the agents.md or core.md, you you you need to go through your skills. You know, the the agents that you had writing your skills at the start of the year probably wrote really bad skills. You know, the best skills are exactly what you're saying, Dave. They can't be super specific. I think there's there's two types of information. There's the general information, which is like the vision, the goal, the value, the reason, you know, those kinds of things they can be in prose. Whereas the real deep, detailed specifics need to be baked mechanically into your agent harness or into your code base itself. You can't like you shouldn't have the agent having to juggle the specifics, it should be just there as variables, as like parameters almost,
[48:27] Dave Pengelley: Hmm.
[48:27] Matt Slager: if that makes sense.
[48:28] Dave Pengelley: Yeah,
[48:28] Matt Slager: So yeah.
[48:29] Dave Pengelley: we got Blake Blake says throw away all the tools.
[48:33] Matt Slager: Yes.
[48:33] Dave Pengelley: Yeah.
[48:34] Shawn Hain: Yeah.
[48:34] Matt Slager: Very much that
[48:35] Shawn Hain: Like
[48:35] Dave Pengelley: Um
[48:35] Richard Webbe: It's
[48:35] Shawn Hain: I
[48:35] Richard Webbe: like it's like retraining your dog. You you you you you gotta give them a whole new way of doing things, new ways of working.
[48:41] Shawn Hain: Yeah, I guess
[48:42] Dave Pengelley: sure.
[48:42] Shawn Hain: uh about a month back I gutted my framework and it's a third of the size
[48:47] Matt Slager: we
[48:47] Shawn Hain: that it
[48:47] Matt Slager: have
[48:47] Shawn Hain: was and it performs better now.
[48:50] Dave Pengelley: Yeah.
[48:50] Matt Slager: do that. to
[48:50] Dave Pengelley: Yeah.
[48:50] Shawn Hain: And I'm about to do it again with a new
[48:52] Dave Pengelley: All
[48:52] Shawn Hain: generation
[48:52] Dave Pengelley: all
[48:53] Shawn Hain: models that have came out since I've done
[48:54] Matt Slager: Yeah.
[48:54] Shawn Hain: it.
[48:55] Dave Pengelley: the things we do for this show. Like all the comic scripts and stuff are all uh a mix of Python skills, Python scripts, but those Python scripts have baked-in prompts and which are arguably like a skill that determines the models to do specific things in a specific order. And even some of those, like the the way the new models are interpreting the scripts and the way they're pulling out the fragments to try and build the storyboard for the comics, it's bad most of the time. And so I need to go and reassess it and rewrite and go, no, not that slide. No, you misunderstood. That wasn't Matt who said that or. It just just it's pulling bad comic ideas. I think I don't actually think I managed to pull a put a comic together for last week because it was just so bad. I think we actually skipped episode forty-three on the uh the old LinkedIn comics because it was just too hard. And I just never got
[49:39] Shawn Hain: Yeah.
[49:39] Dave Pengelley: there fighting it. So yeah, I mean all these things, like uh the thumbnails we do, like these are all iterative things that we've built over a period of time. But as the models get better, uh and to the point that then they overbake things. Like when I asked Hermes
[49:55] Manjula Kerai: Well
[49:56] Dave Pengelley: For the the podcast Hermes to make the animation, it was so big on thinking about SVGs and because we make comics and other things, it just got overwhelmed and thought it had to build it itself versus going to a video generator, which is probably just uh things in its rule set that it's read and interpreted that way versus the core harness, which should have said go use video generation.
[50:17] Matt Slager: Hmm.
[50:20] Matt Slager: So I I don't know if we actually spoke about our wins of the year, but uh what what what's surprised you guys the most, you know, like as far as wins? Because for me, like the stuff that I'm doing now, I never would have thought I would be able to do. Like if you asked me last year. At the same time. I think we're probably around the time that Opus four five came out last year. Oh
[50:38] Shawn Hain: Okay.
[50:38] Matt Slager: uh Kate's here. Hello.
[50:41] Kate Fabian: Better late than never, right?
[50:43] Dave Pengelley: Welcome to the party, pal.
[50:45] Matt Slager: Perfect time to answer a question.
[50:47] Richard Webbe: I should have done this from my car. Everyone's doing it from their car. What's going on?
[50:52] Matt Slager: That's time.
[50:55] Dave Pengelley: The ladies are busy. things to do, they're out and about.
[51:00] Dave Pengelley: Ah, yes,
[51:00] Kate Fabian: Ran
[51:01] Dave Pengelley: and
[51:01] Kate Fabian: over
[51:01] Dave Pengelley: and
[51:01] Kate Fabian: a
[51:01] Dave Pengelley: and
[51:01] Kate Fabian: bit.
[51:02] Dave Pengelley: Am Amber sends her apologies, but but her uh her kids have taken ill, so she's uh on nurse mama. So um,
[51:10] Manjula Kerai: I think what surprises me, um w with your question, Matt, what surprises me is that I'm able to develop things that I've haven't been able to do before. And I remember a time where a few years ago I was thinking I'd love to be able to create plugins and created my own plugin the other day for a customer project so it was like oh I created that you know um so yeah it's surprising me how how much productivity we can do and what we can do with the the technology these days so
[51:41] Matt Slager: Do
[51:41] Manjula Kerai: and
[51:41] Matt Slager: you
[51:42] Manjula Kerai: you know
[51:42] Matt Slager: think
[51:42] Manjula Kerai: use your imagination you can do anything
[51:44] Matt Slager: what yes, yeah, you literally just said the magic words there is it's all down to your ability to think about it and actually comprehend. Um do do you think, Mangela, that that what you've discovered there and you know that exact point, how much of the world knows what you've just described?
[52:01] Manjula Kerai: I don't think they do unless they've you know they're diving deep into AI and and using their imagination. And and my father actually, he he taught me to to do all of this because he would say to me, Mandrula, how do you do this on the computer? And I'd say, No, you can't do that. He goes, but what if you do this? And I'm sure you can get it done somehow, you know? So that kind of that made me tap into my imagination. Imagination more and thinking actually the possibilities are endless. That's what's helping a lot.
[52:31] Richard Webbe: Manjula, that's a great example that you just picked there. My father, I've spoken to him before, was um was dyslexic and an artist, but a super smart guy and an inventor. And when he couldn't do something because he couldn't read about it or understand it, he just created a new way. And necessity is the mother of invention. And the interesting thing about what you guys are saying is I remember when I was first in the computer industry, which was a fair while ago, but we won't Go into that. Um, people would come to me and say, Should I buy this laptop? Because I hear there's a new one coming out next week and it might be better. And when you're talking about plugins and algorithms and method and harnesses and stuff like that, a lot of people are sitting back going, Oh, I'll wait for that to come. But you hit the nail on the head, I need it now. So even though someone may create one tomorrow, you and Dave and Sean and Matt and Kate, you all create them now, and that's the time advantage. That's the time competitive advantage. And I think that's really clever because your experience is ahead of everyone else's, your access. You know, we've had a few people in this program that are creating their whole own harness environments and stuff like that. And I used to think, oh, why are they wasting their time? Because I know someone's gonna, and then I suddenly realize, no, they're first. Their customers will get the advantage first. And those few weeks could be the difference between market winning and market losing.
[53:55] Matt Slager: Plus, it also changes the way you think.
[53:58] Richard Webbe: That's
[53:58] Matt Slager: That
[53:58] Richard Webbe: it,
[53:58] Matt Slager: and I
[53:59] Richard Webbe: that
[53:59] Matt Slager: think that's
[53:59] Richard Webbe: explained.
[53:59] Matt Slager: the long-term win. You know, if you spend the time to build your own harness and understand it and figure out everything there. Then it doesn't matter if your product flops, you have so much more experience and so much more understanding than just the average punter.
[54:11] Dave Pengelley: And I I was I was out this morning and someone sort of was put what are you doing today? I was like oh have a podcast and midday like oh what's your podcast about? I was like, Oh it's about AI. Oh that's not very interesting. Like they're like this is the world, like this is this is No, like people are just rejecting this sort of technology.
[54:30] Richard Webbe: She's back. She's
[54:32] Dave Pengelley: Back, Joe's
[54:32] Richard Webbe: back.
[54:33] Dave Pengelley: back. Laptop always and all. She's see how you go, Joe. Um,
[54:37] Shawn Hain: Yeah.
[54:38] Dave Pengelley: but yeah, I mean, we that means like the the that competitive advantage, that ability to get out first still exists
[54:44] Shawn Hain: Yeah.
[54:44] Dave Pengelley: um for us.
[54:46] Matt Slager: So and I I've got a bit of a topical statement and now that we've got um majorlicade and oh we did
[54:51] Dave Pengelley: Oh,
[54:51] Matt Slager: have Joe.
[54:51] Dave Pengelley: we lost Joe.
[54:52] Matt Slager: Uh
[54:55] Matt Slager: this is
[54:55] Shawn Hain: Almost.
[54:55] Matt Slager: this is a bit of a
[54:58] Matt Slager: tapping out. So yeah. Um the the the thing I think, you know, back to that Christmas. Table discussion that we're all going to get attacked by Kate, if you missed that earlier, basically, you know, we're there we're the weird AI people in the family. So, you know, why are you taking people's jobs? You know, that kind of stuff. Um, it makes me think of like stuff in the past, like history, you know, where you're right.
[55:21] Dave Pengelley: Classic rigid.
[55:22] Matt Slager: Like in the in the US, for example, farms literally used to employ 41% of the US workers in the 1900s. Now it's under 2%.
[55:31] Richard Webbe: Yeah.
[55:31] Matt Slager: You know, the di did those people that were you know kicked out of their jobs, uh how did they go? Like, you know, yes, it's a it's a sad thing, it's it, but it's also a really interesting thing because every time this happens, there's always a new opportunity. There's always a rebalancing. You
[55:46] Richard Webbe: So
[55:46] Matt Slager: know,
[55:46] Richard Webbe: true, so true. Where are the typing pool ladies? And they're all gone and they're doing something else.
[55:50] Matt Slager: ATMs, um another
[55:53] Manjula Kerai: Yeah.
[55:53] Matt Slager: one was elevator operators. Uh you know, if people like to talk about the weaving loom, you know, 98% of the work that instantly was automated.
[56:02] Manjula Kerai: Yeah.
[56:03] Matt Slager: Yeah, there there's all Always gonna be this weird shift. Like, how do you guys feel about that?
[56:07] Manjula Kerai: I I feel like these are just tools that that we're we're creating and and AI is another tool and it's gonna help us with our work and humans are gonna evolve anyway. Like in terms of work, we're evolving, like we we're doing things, we're doing more things that we Than we could have done without AI. So it's it
[56:31] Shawn Hain: Yeah.
[56:31] Manjula Kerai: is going to be a change and and hopefully a good change as long as you know we use it in the right manner. And uh so yeah, I feel like it's it's just normal to change, you know, as times go on.
[56:44] Shawn Hain: Yeah. I my thoughts on it's gonna be painful for certain industries and certain professions, they're in a lot of trouble in the sense that the AI does it close enough to as good at a fraction of the cost. But they can reskill. It's the exact same thing we've gone through dozens of times with c changing technologies like this. Um it's just gonna be painful and the fear mongering now is probably more widespread than it was previously.
[57:13] Dave Pengelley: I think there's a learning curve as well where people work out like, oh, they've overcorrected and went AI can do everything, and then they go, Oh, AI can't actually do those things, and there's problems, and we probably need humans in some of those positions back again. And so we're gonna see this bit of swinging and and back and forth. The um the the Queensland guys that I spoke to a few weeks ago, Turner and Barnett on the podcast, like they had someone say, We want to replace 50 workers with AI. And the guys went, uh no, I'm not, we we're not taking your money because we don't think that's actually viable. Um and like that's that's just not the kind of business we want to enhance and and give people jetpacks, they said, not tear jobs out from underneath them. Sorry.
[57:56] Richard Webbe: I think I think you're the nail on the head then, Dave. A business that comes out and says, my objective is to get rid of 50 employees is a stupid business. Because that's not the objective, right?
[58:07] Dave Pengelley: Yeah.
[58:08] Richard Webbe: We're here to be witness to each other's lives, to be more productive, to make create better lives and better quality and all of that. So the objective is not to get rid of people. That may be a tool you use. I remember I was paid for by Cisco, funny enough, to fly all over the world and explain what a VPN is. And all the IT managers used to freak out because they'd built an empire on managing the information. Now the VPNs were going to be automated and put on these servers and these internal, and they're all panicking, they're going to lose all these staff. And I just created a very simple uh diagram and it said, see the people over here who are now bored indexing and directoring for those files. Well, you don't need them anymore because we've got a way of doing that quite automated, but you can put them on the marketing side or the product development side. Or the make your staff happy side. So I think the and and and Sean was alluding to it before the redeployment of your team is what it's all about, not the killing of them. But every time I hear a company go, I want to get it rid of 300 people, I go, your business is not going to last long because businesses are built on people. And if you try and just maximize profit to get rid of people, you're barking up the wrong flaming tree, I tell you now.
[59:21] Matt Slager: Yeah, there was another study I found when I was doing my research. Um resume.org had a survey where uh managers
[59:30] Richard Webbe: This would be good.
[59:31] Matt Slager: managers fully admitted that they Frame layoffs as because of AI because the investors prefer it.
[59:43] Matt Slager: Yeah,
[59:43] Richard Webbe: It's
[59:44] Matt Slager: it
[59:44] Richard Webbe: just
[59:44] Matt Slager: was
[59:44] Shawn Hain: But
[59:44] Richard Webbe: that
[59:44] Matt Slager: like
[59:44] Richard Webbe: a
[59:44] Shawn Hain: that
[59:44] Richard Webbe: whack.
[59:45] Shawn Hain: the the thing is though investors have kind of gotten to the point now where all they're expecting is layoffs and increased profits. That's all they care about. They're not caring about what this is going to look like in fifty years. When the company has no staff that know
[59:58] Dave Pengelley: Hmm.
[59:58] Shawn Hain: what's going
[59:58] Dave Pengelley: Well,
[59:58] Matt Slager: a lot of people.
[59:58] Shawn Hain: on.
[59:59] Dave Pengelley: yeah, they they think AI is the is the magic, and great to see their forward thinking and they're using AI. That that's driving for productivity inefficiency. And oh look, they've reduced headcounts versus anything else that's actually true about changing headcounts.
[01:00:11] Richard Webbe: And it's because they're scared. They don't know what the changes are, and we're trying to solve those problems by having these discussions in the open. So people understand it's not that bad. It's actually really good. I know of um several senior executives I've spoken to. Right up to sea level. And I keep asking them, so what are your business drivers? And they're a bit lost. And I said, Well, your business drivers is let's look at it. Hang on, sell more stuff, provide better service, return more to the investor. Right? It doesn't say get rid of staff as a way of doing that. And I remember worked on a big project. Well, one of the projects I worked on was the wireless networking of power meters in Australia with a company who's Work for it was groundbreaking technology. You don't see people driving around reading meters anymore, that's all gone. Those people have got other jobs now. But what it was, it was a really efficient customer service version, and the data that you get from that now, you plug that into AI, and now the power and utilities
[01:01:13] Dave Pengelley: I mean,
[01:01:14] Richard Webbe: can provide
[01:01:14] Dave Pengelley: like
[01:01:14] Richard Webbe: so
[01:01:14] Dave Pengelley: like
[01:01:14] Richard Webbe: much better
[01:01:15] Dave Pengelley: I talked
[01:01:15] Richard Webbe: service.
[01:01:15] Dave Pengelley: I've I've talked about a few times, I'm sure, told the story that I would um build systems and databases and things in various of my early jobs in order to make to remove the Grunt work from myself. I was like replacing myself by building tools and systems because I didn't want to do that
[01:01:31] Richard Webbe: But
[01:01:31] Dave Pengelley: manual
[01:01:31] Richard Webbe: you
[01:01:31] Dave Pengelley: data
[01:01:31] Richard Webbe: were doing
[01:01:32] Dave Pengelley: entry
[01:01:32] Richard Webbe: better things
[01:01:32] Dave Pengelley: and
[01:01:32] Richard Webbe: using
[01:01:32] Dave Pengelley: and gross
[01:01:33] Richard Webbe: that creative
[01:01:33] Dave Pengelley: work. Yeah.
[01:01:33] Richard Webbe: prime.
[01:01:34] Dave Pengelley: And I could go do
[01:01:34] Richard Webbe: Yeah.
[01:01:34] Dave Pengelley: I was I could then apply my time to somewhere else in the business.
[01:01:38] Richard Webbe: Yeah.
[01:01:38] Dave Pengelley: Uh and this is that that is happening at scale. Now the issue is the time frames and how quickly it can happen, how quickly people can find new work. I get there's definitely some real concerns around uh that. There's concerns around what you exactly what you said that people are miscommunicating why these restructures are happening and blaming it on AI. When that's not necessarily the case, because then that does add further fear-mongering into the broader populace and puts people off experimenting and trying the new things they should be trying. It's it's a weird, weird time. And this is like you talked about what's changed this year, Matt. 2026, we started the show talking about Maltbot.
[01:02:17] Richard Webbe: Please refresh your memory style.
[01:02:19] Dave Pengelley: So so before OpenClaw, there was ClawedBot, where they would had built this agent harness on top of Clawed. primarily that you know you connected to tools and it could go ahead and do things and and it could make a million dollars overnight while you slept by building a business for you and all the things, all the promises that all the hype merchants sort of sell around such things. But then you know they got into naming issues with Claude over Claude Bot, so they called a malt bot for like three days just as we were going to going to air. And then by the next week I think it was changed to open claw, which is what everyone calls it now. Uh and then you know a few months later Hermes sort of Ate their lunch. But I remember pre-Maltbot when we're looking at like GPT 3.5, GPT 4, long before this show. And I remember saying to someone, the last thing I want is to give one of these LLMs access to external systems. Like it's great. It's in a little walled garden. It's safe. If it goes crazy, it's just a chat. Like that's all it is. I'll copy and paste my code fragments out, whatever. There's no way I want this thing connected to real systems. And now look at us with our Grokbot. And now dots and our Claude Co-works. And it's like, what? I can connect another thing. Please go.
[01:03:34] Shawn Hain: Yep.
[01:03:34] Dave Pengelley: Even my SaaS chart reporter, I've been a bit hesitant with that. I built that in anti gravity over like originally and and deployed it. And I've got stuff in Superbase and Cloudflare and Vercell and it's all this distributed uh setup. But I've been really hesitant to give it direct access to Superbase because I'm like, uh, I know I'll manually copy and paste the SQL things into it and I'll run those because I I just want to make sure. And then over time you go, oh, there's a lot of my time and effort to copy and paste these things. And I'm not really reading the sequel. I might as
[01:04:05] Matt Slager: Yeah,
[01:04:05] Dave Pengelley: well just give it access to Superbase. Because then it can actually also analyze and give me other reporting and data
[01:04:11] Matt Slager: yeah.
[01:04:11] Dave Pengelley: and assess things and do things better in Superbase that I can do by hand. So I'm like, fine, have the keys.
[01:04:17] Richard Webbe: Eventually
[01:04:18] Manjula Kerai: I wonder if
[01:04:18] Richard Webbe: cost
[01:04:19] Manjula Kerai: you
[01:04:19] Richard Webbe: savings
[01:04:19] Manjula Kerai: keep on
[01:04:19] Richard Webbe: and
[01:04:19] Manjula Kerai: top
[01:04:19] Richard Webbe: efficiency
[01:04:19] Manjula Kerai: of it.
[01:04:20] Richard Webbe: and customer service they rock. They do it. And I remember um you know being in the networking game many years ago and we were talking about VPNs before, uh dealing with certain departments of defense around the world without me being specific. And they were saying we'd never ever use a third party network because you know we're the spooks. We've got to protect. They all do it now. Just the higher levels
[01:04:43] Matt Slager: Wow
[01:04:43] Richard Webbe: of security.
[01:04:43] Dave Pengelley: Question.
[01:04:44] Matt Slager: blessed.
[01:04:44] Richard Webbe: Bless you Sean.
[01:04:45] Shawn Hain: Sorry.
[01:04:45] Richard Webbe: Bless you. I thought you were doing the you know one of those Cohen brothers uh bullshit sort of thing.
[01:04:52] Shawn Hain: No.
[01:04:53] Dave Pengelley: Does it have any favor?
[01:04:54] Shawn Hain: That's it.
[01:04:58] Richard Webbe: Yes, it does. I don't know why. They just do it all the time. Anyway.
[01:05:06] Matt Slager: It's
[01:05:07] Dave Pengelley: Um
[01:05:07] Matt Slager: crazy. We missed Blake's comment. He said that beginner mindset might actually be a superpower.
[01:05:13] Shawn Hain: Yeah.
[01:05:13] Matt Slager: Um, which I think is true. And it's similar to what Mangela was saying before. You know, the the idea of not getting bogged down with all of the crazy technicality of your previous in
[01:05:23] Dave Pengelley: yeah.
[01:05:23] Matt Slager: um
[01:05:23] Shawn Hain: Yep.
[01:05:23] Matt Slager: institutional knowledge, and you just say, I just want to make a thing.
[01:05:26] Dave Pengelley: Yeah.
[01:05:26] Shawn Hain: Yeah.
[01:05:26] Matt Slager: Can we do that?
[01:05:27] Dave Pengelley: So
[01:05:27] Richard Webbe: Yeah.
[01:05:28] Dave Pengelley: I I've like I said, I I keep I keep having random game ideas. I was out for a walk on Saturday afternoon and I was listening to a podcast and these guys just talked about like shoot for the moon. That's all I heard. Like they they talked about, you know, like they're talking about their own projects or whatever, and they're like, I can play it safe or just shoot for the moon. And so as I'm out for a walk through a park, I'm like, ah, moonshot. Oh, shooting for the moon. And so while I'm out walking, I'm like, hey Grockbot, I want you to prototype. Game where we chuck a cow over the moon and gave it a bit more details, but just the random creativity that happened, like where I just hear like shoot for the moon, and my brain's like, Oh, there's a game idea there. Can you can you gotta build some kind of cannon or catapult and like time with the nursery rhyme? The cow jumped over the moon, and it's sort of fair, like and and have something where you build this machine that chucks a cow, it'll be comical and funny, and you start by chucking a cow over another cow and then over a tractor, and you sort of build a levels up until you're actually at the point where you can reef it over the mood.
[01:06:30] Richard Webbe: It is interesting having the tools allows your creative mind to go to new spaces, right? I think of games like Leisure Suit Larry.
[01:06:38] Dave Pengelley: Of course you do.
[01:06:41] Richard Webbe: But you know, Grand Theft Auto. Who was what
[01:06:43] Dave Pengelley: Of
[01:06:43] Richard Webbe: you
[01:06:44] Dave Pengelley: course
[01:06:44] Richard Webbe: know, who
[01:06:44] Dave Pengelley: you
[01:06:44] Richard Webbe: would
[01:06:44] Dave Pengelley: do.
[01:06:44] Richard Webbe: sit
[01:06:44] Dave Pengelley: It's the
[01:06:45] Richard Webbe: down
[01:06:45] Dave Pengelley: same
[01:06:45] Richard Webbe: and
[01:06:45] Dave Pengelley: thing.
[01:06:47] Richard Webbe: sit down and say, let's deal drugs, find hookers, kill people, steal cars? But it's now the biggest selling game in the
[01:06:54] Dave Pengelley: Well
[01:06:54] Richard Webbe: world.
[01:06:54] Dave Pengelley: one of the most popular indie games that came out on Steam in the last twelve months was where you're a meth dealer.
[01:06:59] Richard Webbe: That's
[01:07:00] Matt Slager: Yeah,
[01:07:00] Richard Webbe: right.
[01:07:00] Matt Slager: um what was that called? Schedule one,
[01:07:02] Dave Pengelley: Schedule
[01:07:02] Matt Slager: I think.
[01:07:02] Shawn Hain: Yeah.
[01:07:03] Dave Pengelley: one, yeah, and you just go around and cook meth and you you break up the crystal like you see in breaking bad and bag it and that's all these
[01:07:08] Richard Webbe: Dive,
[01:07:09] Dave Pengelley: little moody
[01:07:09] Richard Webbe: this is not a
[01:07:09] Dave Pengelley: games.
[01:07:09] Richard Webbe: drug dealing podcast.
[01:07:12] Dave Pengelley: But people just
[01:07:14] Shawn Hain: Yeah, I was gonna say before like the one thing I've noticed that's really changed like how I approach stuff with the institutional knowledge you're talking about, Matt, is prototyping.
[01:07:23] Matt Slager: Yeah.
[01:07:23] Shawn Hain: Before that was like if I'm prototyping something, that's a months worth of work now
[01:07:27] Matt Slager: Yeah.
[01:07:27] Shawn Hain: today or two. You know, it depending on how polished I want it to be.
[01:07:32] Matt Slager: Yeah.
[01:07:32] Shawn Hain: I can throw together a bad prototype in 30 minutes.
[01:07:35] Matt Slager: Yeah, give me give me five versions
[01:07:37] Shawn Hain: Yeah.
[01:07:37] Matt Slager: and it helps steer the direction. Yeah.
[01:07:40] Shawn Hain: Yeah, yeah. Give me five wildly different visual versions of this and I'll tell you what I like about all five and we'll come to one solution. That's an option now, where before that was a team.
[01:07:50] Richard Webbe: Yeah.
[01:07:50] Matt Slager: Yeah.
[01:07:50] Shawn Hain: And just like how we work and how we think about stuff I think needs to change.
[01:07:55] Matt Slager: Definitely.
[01:07:56] Shawn Hain: And I think that's well I rally to what we were talking about earlier with the industry changes. Once industry catches up, there'll be jobs, but the jobs will be different. They'll be more creative, more
[01:08:09] Richard Webbe: I think they're there now. I think they're coming now. I
[01:08:11] Shawn Hain: Yeah,
[01:08:11] Richard Webbe: always, you
[01:08:12] Shawn Hain: yeah.
[01:08:12] Richard Webbe: know, my my brother's a musician, you know, um in the old days, musicians when isolated and that would take ages to come up with a song. You get a bunch of experienced musicians now who know their way around the tools really quickly. You sit them in a room, they'll come out with an Award winning song very, very quickly. And I see AI doing the same thing. The connection between your creativity and the output gets better.
[01:08:36] Matt Slager: Yeah, you're absolutely right. It's
[01:08:40] Dave Pengelley: Yeah,
[01:08:40] Matt Slager: kind of
[01:08:40] Dave Pengelley: I
[01:08:40] Matt Slager: speaking.
[01:08:40] Dave Pengelley: think I think thing the the what you're just saying about prototyping shorts, and like I iterate through lots of things all the time because it's so cheap and easy to iterate, right? Um and so before you go and you know set up Unreal Engine and start doing stuff, you have different options. And so I'm like, just just make a web, just make a web prototype of this thing. Last night, uh, this is this game is gonna go nowhere. But I was like, oh, I want to take the drop seven concept, but with mechs, I've got this thing about mechs at the moment, uh, where you can drop a mech down and it's got a power score against the other one and they fight and then it falls down and fights mechs beside it. And I played it, went, I'm not sure this is actually better than drop seven. I'd rather probably just play drop seven. So um, but I could quickly prototype this this this idea and go, is there any legs there? Maybe, maybe. Maybe not, I don't know. Um,
[01:09:28] Shawn Hain: Yeah.
[01:09:28] Dave Pengelley: but
[01:09:28] Richard Webbe: We
[01:09:28] Dave Pengelley: it's just
[01:09:28] Richard Webbe: are
[01:09:28] Dave Pengelley: so
[01:09:29] Richard Webbe: getting
[01:09:29] Dave Pengelley: quick and easy
[01:09:29] Richard Webbe: a
[01:09:29] Dave Pengelley: to
[01:09:29] Richard Webbe: really
[01:09:29] Dave Pengelley: try things like
[01:09:30] Richard Webbe: mate,
[01:09:30] Dave Pengelley: that.
[01:09:31] Richard Webbe: we're getting a really big insight to what goes on inside your head. It's
[01:09:35] Dave Pengelley: Giant
[01:09:35] Richard Webbe: fascinating.
[01:09:35] Dave Pengelley: robots,
[01:09:36] Richard Webbe: It
[01:09:36] Dave Pengelley: Jaegers.
[01:09:36] Richard Webbe: is fascinating.
[01:09:37] Dave Pengelley: I like Pacific Room. It's a good movie.
[01:09:39] Shawn Hain: Yeah.
[01:09:40] Dave Pengelley: Uh Joe's Joe's back, see if she's uh got a stable connection. She's
[01:09:44] Shawn Hain: Well,
[01:09:45] Dave Pengelley: outside.
[01:09:45] Shawn Hain: she's out of the car now.
[01:09:48] Dave Pengelley: Joe wasn't in the car. Joe was uh
[01:09:50] Shawn Hain: That's right, I was the other two. My apologies.
[01:09:53] Dave Pengelley: Yeah, their their car office is shut down.
[01:09:55] Shawn Hain: Yeah.
[01:09:56] Richard Webbe: Revolving labor door of AI, everyone keeps
[01:09:58] Dave Pengelley: Stop
[01:09:58] Richard Webbe: coming
[01:09:58] Dave Pengelley: it.
[01:09:58] Richard Webbe: and
[01:09:59] Dave Pengelley: This
[01:09:59] Richard Webbe: going.
[01:09:59] Dave Pengelley: is not a political podcast. Don't mention labor.
[01:10:08] Dave Pengelley: Good luck.
[01:10:09] Joanna Jones: Can
[01:10:09] Shawn Hain: Yeah.
[01:10:10] Dave Pengelley: This
[01:10:10] Joanna Jones: you
[01:10:10] Dave Pengelley: is not
[01:10:10] Joanna Jones: hear
[01:10:10] Dave Pengelley: a political
[01:10:10] Joanna Jones: me?
[01:10:11] Dave Pengelley: podcast. Um,
[01:10:12] Richard Webbe: Yes, we can.
[01:10:13] Joanna Jones: Oh
[01:10:13] Richard Webbe: Joe,
[01:10:13] Joanna Jones: great.
[01:10:14] Richard Webbe: say something.
[01:10:14] Joanna Jones: Okay. I'm I'm on a I'm I'm on the old iPhone.
[01:10:22] Richard Webbe: And
[01:10:22] Dave Pengelley: I'm not sure what I'm
[01:10:22] Richard Webbe: we're looking at that brick wall behind you. I'm not sure if you've been arrested. You're in your cell or whatever, but did you have anything you wanted to contribute? Because you were locked out for a bit there.
[01:10:33] Joanna Jones: I'm trying to put a better a better view on this, but no, I was listening.
[01:10:42] Joanna Jones: You were saying about
[01:10:49] Dave Pengelley: We can
[01:10:51] Dave Pengelley: you're breaking out, we can hear you, we can't see you. Oh.
[01:10:55] Shawn Hain: Maybe we
[01:10:57] Dave Pengelley: I
[01:10:57] Joanna Jones: biggest the thing.
[01:10:58] Dave Pengelley: I I have something for this.
[01:11:07] Shawn Hain: Love
[01:11:08] Richard Webbe: Well,
[01:11:08] Shawn Hain: it.
[01:11:09] Dave Pengelley: I had to use that for one of the operator stories because like while I was doing it, I was like, the one I published just last this week. Nathan, I was like, he's talking away and then his screen goes black like this because we use the same tool.
[01:11:19] Richard Webbe: I
[01:11:19] Dave Pengelley: And then I'm
[01:11:19] Richard Webbe: a good that's thing. think
[01:11:19] Dave Pengelley: like, Nathan, I've lost your camera. I've lost your camera. And then I just cut to that because it took us about five minutes to get him back online. So he just Little signal erupted video and then back. All right, you were saying something about innovation.
[01:11:31] Richard Webbe: Nostalgia.
[01:11:33] Dave Pengelley: Uh most people are like, what is what are those colours? What does that even mean?
[01:11:36] Matt Slager: Yeah, true.
[01:11:38] Shawn Hain: Yeah,
[01:11:38] Dave Pengelley: The
[01:11:38] Shawn Hain: it's
[01:11:39] Dave Pengelley: old
[01:11:39] Shawn Hain: been a while.
[01:11:40] Dave Pengelley: the old test pattern.
[01:11:41] Matt Slager: We're coming back next year, right?
[01:11:43] Dave Pengelley: I hope so. I hope so. But like as far as like the evolution of these tools and about how to drive Astro, I was using Dot um to drive Astro over the weekend um from uh g getting Hamilton to do it. I started by getting it all set up just directly in codecs. But then I was like, actually, I'm just gonna get Hamilton to drive this for me because uh I entered a game jam. Epic games were running an Unreal Engine game jam for four days. And they called it the AI Mega Jam. So I was like, AI? Yes. Because there is this huge pushback in the game industry about all the AI slop and people just sort of sloping games out there, which is why I'm trying to handcraft my story and narrative and do it a little bit differently and not just slot things out. But I had four days and so I got this was my final entry I'll show the video um from a win of the week win of the year this won't be judged till January so I probably won't win anything but I built this for me under my guidance
[01:12:42] Richard Webbe: Shadow Code. Oh, I knew you Masters of the Universe, your favorite character.
[01:12:51] Dave Pengelley: so you got a dodge around the drones
[01:12:56] Dave Pengelley: You can throw little distraction beacons. Uh oh, he saw me.
[01:13:00] Matt Slager: Yeah.
[01:13:04] Richard Webbe: Oh
[01:13:04] Dave Pengelley: Oh no.
[01:13:07] Dave Pengelley: But the
[01:13:07] Richard Webbe: have
[01:13:08] Dave Pengelley: idea
[01:13:08] Richard Webbe: you
[01:13:08] Dave Pengelley: they
[01:13:08] Richard Webbe: got
[01:13:08] Dave Pengelley: they they they themed it was the the theme was director's card. So I made it you gotta steal movie reels from the studio, avoid the security drones.
[01:13:17] Richard Webbe: a clever sort of visual mathematical um challenge?
[01:13:23] Dave Pengelley: And when you extract the real you get the movie poster to go with it. It so that way you got that little
[01:13:28] Richard Webbe: Have you
[01:13:28] Dave Pengelley: sense
[01:13:28] Richard Webbe: got
[01:13:29] Dave Pengelley: of
[01:13:29] Richard Webbe: uh
[01:13:29] Dave Pengelley: success.
[01:13:29] Richard Webbe: have you got uh copyright approval for Heyman yet?
[01:13:33] Dave Pengelley: He's just a Fortnite character, so he's not like it's Fortnite, so you can go as whichever skin that you have unlocked in Fortnite. I just
[01:13:40] Richard Webbe: Got
[01:13:40] Dave Pengelley: happened to
[01:13:41] Richard Webbe: it.
[01:13:41] Dave Pengelley: have paid for the He-Man skin and he was the one I've been using in Fortnite lately. So when I was doing the testing, he man was running around the
[01:13:48] Richard Webbe: What
[01:13:48] Dave Pengelley: movie
[01:13:49] Richard Webbe: what
[01:13:49] Dave Pengelley: studio.
[01:13:49] Richard Webbe: wasn't he your childhood hero?
[01:13:51] Dave Pengelley: Wasn't he everyone's childhood hero, Richard? Even even grown men in
[01:13:58] Matt Slager: So I have a y your uh your respect's probably gonna uh drop for me. Dave,
[01:14:04] Richard Webbe: Uh oh.
[01:14:05] Matt Slager: but uh I've actually never watched any He Man and um whenever I hear it or see it it always reminds me of the little rascals.
[01:14:12] Dave Pengelley: That's it, he's gone.
[01:14:14] Matt Slager: So yeah.
[01:14:19] Matt Slager: Yeah.
[01:14:22] Richard Webbe: Matt,
[01:14:22] Matt Slager: For some
[01:14:22] Richard Webbe: don't
[01:14:22] Matt Slager: reason
[01:14:22] Richard Webbe: feel
[01:14:23] Matt Slager: like
[01:14:23] Richard Webbe: bad.
[01:14:23] Matt Slager: it.
[01:14:23] Richard Webbe: I'm the same. I've never watched Mar Masters of the Universal He Man. I would see a flash of it on the telly but I'd left that age group. I was always a Spider-Man man, so Matt, you and I are in the same category
[01:14:34] Dave Pengelley: Have you watched
[01:14:34] Richard Webbe: there.
[01:14:35] Dave Pengelley: the new movie that came out this year that's on Prime?
[01:14:38] Shawn Hain: It's
[01:14:38] Richard Webbe: The
[01:14:38] Shawn Hain: not
[01:14:38] Richard Webbe: new
[01:14:39] Shawn Hain: bad.
[01:14:39] Dave Pengelley: It's
[01:14:39] Richard Webbe: the
[01:14:39] Dave Pengelley: good,
[01:14:39] Richard Webbe: new
[01:14:40] Dave Pengelley: yeah.
[01:14:40] Richard Webbe: Masters one
[01:14:41] Dave Pengelley: Yeah, the
[01:14:41] Shawn Hain: Yeah.
[01:14:41] Dave Pengelley: new
[01:14:41] Richard Webbe: with
[01:14:41] Dave Pengelley: one that
[01:14:42] Richard Webbe: the
[01:14:42] Dave Pengelley: came
[01:14:42] Richard Webbe: Australian
[01:14:42] Dave Pengelley: out this year.
[01:14:42] Richard Webbe: guy who plays the lead.
[01:14:44] Dave Pengelley: He's not Australian, he's he's English,
[01:14:46] Richard Webbe: Oh,
[01:14:46] Dave Pengelley: isn't
[01:14:46] Richard Webbe: I thought it
[01:14:47] Dave Pengelley: he?
[01:14:47] Richard Webbe: was Australian. Oh. Okay. Well, I watched it because you recommended it. It wasn't a bad film.
[01:14:53] Dave Pengelley: Yeah.
[01:14:54] Shawn Hain: Not bad.
[01:14:55] Dave Pengelley: Yeah. HR, like he was working in HR at the start of it. It's good, entertaining. Um, I thought I thought it respected the source material well enough, and but it also acknowledged how ridiculous the kids' cartoon was. So, yes.
[01:15:08] Shawn Hain: Yeah.
[01:15:09] Richard Webbe: Yeah.
[01:15:09] Dave Pengelley: Um,
[01:15:10] Shawn Hain: Maybe if you think about most cartoons from back then, they're they're all pretty ridiculous.
[01:15:15] Dave Pengelley: they're they're they're hard to watch. Some some of them are really like you try to watch the old Ninja Turtles cartoons. Like, I've watched a bit of like He-Man like more recently, and it's okay. It's like it is what it is, but you watch some of the original Ninja Turtles. Turtle stuff that we were addicted to when I was a kid, and I'm like, this is so hard to watch. Like,
[01:15:33] Shawn Hain: Oh,
[01:15:34] Dave Pengelley: oh my
[01:15:34] Shawn Hain: I
[01:15:34] Dave Pengelley: goodness.
[01:15:35] Shawn Hain: I ran into some old Captain Planet stuff the other day. It was painful.
[01:15:42] Richard Webbe: Go
[01:15:42] Dave Pengelley: At
[01:15:42] Richard Webbe: back and
[01:15:42] Dave Pengelley: least.
[01:15:42] Richard Webbe: watch the old Spider-Man stuff, because my last name was Webb. I was cast to be a Spider-Man lover going back many years, and that stuff's a little bit painful.
[01:15:51] Dave Pengelley: At least Captain Planner was fighting pollution, not This is not a political podcast.
[01:16:02] Dave Pengelley: Oh man, everything crosses over. It's so hard. Um that's not no.
[01:16:08] Richard Webbe: Oh
[01:16:08] Dave Pengelley: Um
[01:16:08] Richard Webbe: you know what's funny? Just on the this is not a political podcast, I went and voted in the low sensory voting booth the other day. Yes, so
[01:16:18] Shawn Hain: Oh,
[01:16:19] Richard Webbe: you might
[01:16:19] Shawn Hain: yeah,
[01:16:19] Richard Webbe: like
[01:16:19] Shawn Hain: how
[01:16:19] Richard Webbe: that.
[01:16:19] Shawn Hain: was that?
[01:16:20] Richard Webbe: So yeah, so it's designed for neurodiverse people, uh which I'm a little bit But it's also designed for carers of neurodiverse people. And because I have some members of my family in that category. So I thought, okay, um, it's a day early and I don't have to queue up with all the other idiots. So that was across the road from my dentist, so I went and voted there. But what was so funny is no one had any idea how to how to vote.
[01:16:44] Matt Slager: But what you know when the people all stand outside with the flyers, were
[01:16:47] Richard Webbe: Yeah.
[01:16:47] Matt Slager: they uh were they handing you earmuffs
[01:16:49] Richard Webbe: No,
[01:16:50] Matt Slager: instead?
[01:16:50] Richard Webbe: I I yeah, no, yeah, very good point. No, I walked in with a pretend blind stick. Um no, I didn't do that. Have done that. Um, but
[01:16:58] Shawn Hain: Oh
[01:16:58] Richard Webbe: yes,
[01:16:58] Shawn Hain: yeah.
[01:16:58] Richard Webbe: they were actually very friendly, and there was one particular person from a particular party that I think we're lucky to get one percent of the vote, and they were sitting there. I won't say which one, of course, it's not a political
[01:17:10] Dave Pengelley: Non
[01:17:10] Richard Webbe: podcast,
[01:17:11] Dave Pengelley: playable
[01:17:11] Richard Webbe: but they
[01:17:11] Dave Pengelley: castle.
[01:17:11] Richard Webbe: were sitting there crocheting, and such a lovely person. And I went up to her and I said, You're doing well representing, you know, that party here and stuff like that. And I looked, and she was the actual candidate, so she had a photo there, and I said, Um Well done to you because without you there would be no democracy. So whether you get lots of votes or little or how tough it is, keep doing what you're doing. And that's how I feel. I thought it was great.
[01:17:36] Matt Slager: That's cool. Was it was it Pauline?
[01:17:41] Richard Webbe: Mate, based on the numbers last night, I don't think it was a it was a balling person. Yes, it was um, but it was very interesting.
[01:17:50] Matt Slager: So you weren't wearing your uh your pin then?
[01:17:53] Richard Webbe: No, I didn't.
[01:17:55] Matt Slager: We we are still live.
[01:18:01] Shawn Hain: Yeah.
[01:18:01] Dave Pengelley: So, in I mean, uh the the point you make is really relevant. I mean, democracy is in the balance, and this is not a political podcast, but uh I in a previous life actually built this tool and I actually own the domain, how to dot vote, and I've I've just since dispensed with it because it was expensive and I I'm not doing it anymore. But um, and uh yes, this is a polling video, but this is just a good educational video, regardless of who you vote for, that explains the preferencing system. So this was for The federal election. And this is a completely apolitical website that just sort of says who you're voting for, House or Senate. You could pick, you know, where you're at, and you can see all the candidates. This was the exact order that the sheet was going to come into when you went into the booth. And so then you could see and go, I'm planning on you know donkey voting. I'm just gonna go like this. You could generate that as a little image to take with you on your phone. Uh, but if you want to know more about who someone was, like who is Ed Houston. It just googles HUC Labor Canada, and you get the details for it. So
[01:19:00] Richard Webbe: Well
[01:19:01] Dave Pengelley: um,
[01:19:01] Richard Webbe: done. Well done.
[01:19:02] Dave Pengelley: yeah,
[01:19:03] Richard Webbe: One one thing I noticed interesting, well, one thing we should think about, and and not who they're voting for, but there
[01:19:10] Dave Pengelley: I a good idea. that's think
[01:19:10] Richard Webbe: is a phenomena with music and fashion and everything that people always like to go for the winner. And there's actually quite a large percentage of voters who are like they're at Melbourne Cup Day and they just want to vote for the winner. They don't care who the winner is. just internally want to feel like they picked the winner rather like gambling rather than saying I like that person's policy. And I'm just I was thinking with AI and people saying who should I vote for? I wonder how that might affect the outcome.
[01:19:43] Dave Pengelley: Maybe. I don't know if they're just asking the LLM. I mean, hopefully the LLM's ask them some questions around what sort of what they like, or is going from their memories and their own personal personality aspects that's shared with them.
[01:19:55] Matt Slager: I'm pretty sure, like, if you were to ask Codex, you would get flagged for a cybersecurity risk.
[01:20:06] Shawn Hain: No, ask Fable, it'll just stop working, it's fine.
[01:20:11] Dave Pengelley: But but talk about the evolution of the tools. Like that was I built that back in um a couple of years ago using Replit. And because that was a tool that made the most sense, Replit could stand things up and run the full hosting and have a back end database, and it brought in all the the electric. Role CSV files from the AEC and was able to format them up and iterate through the look and feel. And then, you know, um, when I decided we need a research, I added like two prompts and got the little question mark button that let me Google out to find them. The original idea was I was going to try and build a paid monetized solution for parties to set up their own like how-to-vote cards uh on my platform and print QR codes and that point to my platform and they could run all their candidates and manage all their preferences and their how-to votes out of My platform, but two years ago, trying to do that level of database, personalized individual profiles, paid accounts, account management was overwhelming. Like, I like I couldn't get my head around the architecture of that. And certainly the AI AI wasn't just gonna do it for me. Nowadays, I'd go, Oh, yeah, I could probably build that exactly what I had in mind back then. Um, but at the time I was like, I'm I'm running on a deadline, the election's coming out, and I uh and plus, as we're discussed, building the tools, one thing the go to market. How do I get in front of party campaign organizers and how do I distribute this thing and who do I sell it to and who's gonna pay for it and all those kinds of things? Uh, the actual hard bit, like having the ideas is easy, but does it solve a real problem? Do you know who actually cares about the problem and who's actually gonna spend money to fix that problem?
[01:21:43] Matt Slager: Hmm. I think that that literally that last section you just said there is gonna be the theme of 2027 because we can build technically so this is the conversation I was having with uh a friend yesterday, it was actually Blake, yeah, who who joined in the audience today, who, you know, we talk about value a lot, like what is value, you know, and what is perceived value and how do you actually go about those things. So we believe that in the world of making stuff, like what you're just describing, Dave, like the floor to entry, like the barrier to entry is dropped. You know, you absolutely can just go and make stuff now. It's not a problem. But at the same time, there's a ceiling, right? And that ceiling is, you know, now you have a production grade thing in effect that can protect against nefarious agent swarms now,
[01:22:29] Dave Pengelley: Yeah.
[01:22:29] Matt Slager: you know, all of that kind of stuff. The ceiling has risen. So the idea of actually Selling an idea to someone, I think, is going to be crazy difficult because everyone can just make their idea, but what we'll actually be selling is hardening their idea and main maintaining it and managing it and you know making it better and improving
[01:22:50] Richard Webbe: Yep.
[01:22:50] Matt Slager: it.
[01:22:51] Richard Webbe: Yep. Yep. Just out of interest. That's so true, Matt. I really I really identify with that. Just Um, I did that just for fun. Who should I vote for on the LLM? And it was great. It came back advising me of what the top issues from a generalist point of view in Victoria is. And I went, ah. And it said, which one of these are important? Joe, you're back. I'll be quiet. Talk to us, Joe, talk to us.
[01:23:19] Joanna Jones: Oh, my hair should be hanging out in all sorts of spots
[01:23:22] Dave Pengelley: No,
[01:23:22] Joanna Jones: at the moment.
[01:23:22] Dave Pengelley: you still
[01:23:22] Joanna Jones: I've
[01:23:22] Dave Pengelley: look
[01:23:23] Joanna Jones: literally
[01:23:23] Dave Pengelley: amazing.
[01:23:23] Joanna Jones: been upside down in this laptop. Oh, thanks, Dave.
[01:23:29] Joanna Jones: And it's not a political podcast.
[01:23:32] Dave Pengelley: It's not a political podcast.
[01:23:34] Matt Slager: So what's what's
[01:23:35] Shawn Hain: Oh,
[01:23:35] Matt Slager: happening is your your reception is just not good today.
[01:23:39] Joanna Jones: I believe so. I've now put my little uh router right beside my head and my computer. So it seems to be working.
[01:23:47] Shawn Hain: yeah.
[01:23:49] Richard Webbe: Don't tell uh don't tell our climate minister and our prime minister about putting radiating things near body parts. They had a bit to say about I think it's not really a podcast.
[01:23:58] Dave Pengelley: My goodness. I don't think we are coming back next year with this rate.
[01:24:04] Matt Slager: Must be uh yeah, sponsored
[01:24:06] Shawn Hain: I'll I'll leave
[01:24:06] Matt Slager: by
[01:24:07] Shawn Hain: that
[01:24:07] Matt Slager: it.
[01:24:07] Shawn Hain: alone. Sorry. Um not a political podcast.
[01:24:12] Dave Pengelley: That's what we're gonna rename ourselves to. Um that that's gonna be the after hours actual political podcast.
[01:24:18] Matt Slager: I guess the w how long are you going for today, Dave? Are we are we
[01:24:22] Dave Pengelley: Look,
[01:24:22] Matt Slager: intending
[01:24:22] Dave Pengelley: I thought
[01:24:22] Matt Slager: to wrap
[01:24:23] Dave Pengelley: I thought
[01:24:23] Matt Slager: up?
[01:24:23] Dave Pengelley: being the last one of the year, I was happy to run late. We are obviously this is a long show uh we've got a we've got a bunch of people viewing we've had people jumping in and out i think it's been uh good but we uh we can we can start tightening it down to the pointy end i respect everyone's got work and things to do and i thank everyone for their time and energies this year again um yeah i think on reflection that it's been a crazy year how much ai has changed the fact that at the start of the year i was like oh yeah i'll pay for some anti-gravity and i think i was paying For some Chat GPT back then as well. Like I'd upgraded the business account because I was gonna consult with companies and help them with their setup. So I figured I'd better know what the business accounts look like at the same time as well. Um, but then you know I went to Claude for a while, was paying, you know, um money at Claude, and they went, ah Claude, they're costing me too much, so I need to downgrade. So I ditched Claude, and then I've tried Minimax of all things for a little while, and I shopped around to some of those cheap models for a little bit. Um, and then Grok started looking like it was. Coming better. And so I bought the the $30 a month Groc plan because it came out like as a $10 a month special for the first three months. I was like really penny pinching on my models for a little while. But now I haven't got back to Claude. I don't have Claude access other than through news portal with Hermes. And I've got my Grokbot and I've got my Hamilton AI dot. And for me, the the big thing is trying to work out which jobs do I give to which agent teams, which ones have the context they need, which ones are running in their own computers, which ones can I leave a job, which ones Are better mobile accessible, which ones like
[01:25:59] Shawn Hain: Yeah.
[01:26:00] Dave Pengelley: there's no necessarily right or wrong answer. It just depends on how much effort I've spent um with them preparing them for that task.
[01:26:07] Matt Slager: Similar to what Phil said earlier, like that's just environment design. And yeah, I I uh I'm constantly tuning and tweaking mine as well. Um it is an interesting thing though, like at this point, if if there's people out there that you know haven't dabbled yet, you know, they're still yet to pick up their first tool. Um I I'm still cute, I love asking people now, like what what is the beginner's choice and why, and like you know, those kinds of things. Obviously, there's so Many um layers of ifs, you know, you know, depending on what you want to do
[01:26:39] Richard Webbe: So
[01:26:39] Matt Slager: and all these
[01:26:40] Richard Webbe: so
[01:26:40] Matt Slager: kinds
[01:26:40] Richard Webbe: so
[01:26:40] Matt Slager: of things.
[01:26:40] Richard Webbe: Matt, that's a a a really important question. So we're in our own bubble. Um I and you guys I'm lucky you guys invited me into your bubble because it wouldn't have normally
[01:26:49] Dave Pengelley: I
[01:26:49] Richard Webbe: been
[01:26:49] Dave Pengelley: think
[01:26:49] Richard Webbe: my
[01:26:49] Dave Pengelley: you
[01:26:50] Richard Webbe: bubble.
[01:26:50] Dave Pengelley: invited yourself,
[01:26:50] Richard Webbe: But
[01:26:50] Dave Pengelley: Richard, from memory.
[01:26:51] Richard Webbe: yeah, that's true.
[01:26:56] Richard Webbe: I learn all this stuff from you guys about the technical side and where to use it. And then it dawned on me that you know the biggest use of AI is copilot, or you know, it's either Google or Microsoft enabled AI that's shoved under our nose at no cost. And that's the biggest use. And I think most beginners are learning there. I had someone who knew nothing about AI come to me the other day and go, Oh, I'm loving this copilot thing because you know it's on my phone and on my computer, and I can harness all the data from our computer. And of course, we're a couple of layers because That with an abstraction of using independent AI LLMs and so on, which brings me to a really important question Is there a political financial bias to AIs that, and it's not a political podcast, the AIs that you pay for versus the ones you get for free? So, you know, everyone's trying to monetize the investment in this technology. So if Google gives me AI, is the answers there going to go towards the AI's, the customers. Does it pay Google for sponsorship or is it going to still be open and wide? And am I getting a more truer answer by paying for pure and applied LLM?
[01:28:07] Dave Pengelley: Well we'll we will discuss this like if you go directly to the API, you get different than if you use their out-of-the-box harness. So if you use the Gemini web interface or the ChatGPT interface, that's different than going directly to the back end of
[01:28:21] Richard Webbe: Yes.
[01:28:21] Dave Pengelley: the thing and talking directly to the model with all out all its layers and things. Picking aside on this, but politically, it is interesting that Albanese's announcement this week around the new AI safeguards and bureaucracy and department of uh government salaries that's going to help safeguard us from AI potentially puts a risk what sort of AI we're going to be able to have access to domestically in Australia. Like the labs might have to put geocaching and tagging on that puts a whole bunch of extra guard routes. And harnesses across any models we get access to that even further restrict the unbiased nature of the results we can get. Like there, there's there's huge things to look at in the new year as far as what this legislation, if it goes through, means for Australians' access to AI compared to anywhere else in the world.
[01:29:16] Richard Webbe: And are the answers biased if someone else is paying for them other than us?
[01:29:23] Shawn Hain: I think the I don't know if the answers would be biased, but I Yeah, and shit, if you're getting something for free, then you're the product.
[01:29:30] Richard Webbe: That's it.
[01:29:31] Shawn Hain: Is but my internal thing. If I'm not paying for this, then they're making money off of me somehow. Or it's something like Railway or something where it's like he will give you a really cheap or free version to encourage people to bring in the larger stuff.
[01:29:45] Matt Slager: Yeah.
[01:29:45] Shawn Hain: Um
[01:29:48] Shawn Hain: they're making money somehow out of you.
[01:29:50] Matt Slager: With like for fear of going too technical down and making everyone glaze, um, this is similar again to what Phil said earlier about capturing the SOPs and having it in a in a knowledge managing situation. So Like that, your question there, Richard, let's just use the scenario of going to Google and you you type your normal Google search that you've always done before. But now there's the AI results. You know, Gemini's in there trying to give you extra assistance.
[01:30:17] Richard Webbe: Yeah.
[01:30:17] Matt Slager: Now, your your question is, is that output going to bias me in some way? You
[01:30:23] Richard Webbe: Yes.
[01:30:23] Matt Slager: you would have to find out what is the environment that that agent system works in. You know,
[01:30:29] Richard Webbe: Yeah.
[01:30:29] Matt Slager: what is
[01:30:30] Shawn Hain: Yeah.
[01:30:30] Matt Slager: their system instructions? You know, what tools do they have access to? You know, how How are they actually advised to use those tools? Because you know, it could literally just be that that agent is not biasing you at all, but the data that that agent has access to is biasing you.
[01:30:47] Richard Webbe: Absolutely. It's like um I I we all know that if we do a search on Google, it's going to push its sponsorship customers to us. First. So if I'm looking for a new oven, the people that pay Google the most, and I know that. If I'm watching the Tele and Channel 9 News is reviewing the new products for Christmas, I know they're reviewing products that have paid Channel 9 to promote their products. And that's okay because, as we were mentioning earlier, I'm prepared for that bias. It makes sense, and my brain can filter out the biased answers. But I think several people said earlier today, um, I think you as well, Major, it is done. Dangerous to think it's a pure answer. And that's what I think we need to educate the people on. What is a pure answer and what is a gilded lily?
[01:31:37] Joanna Jones: I've actually got a workshop on Saturday and it's literally for basics. So it's called AI Simple Accelerator, and it's for people that are either business owners or professionals that I don't feel confident in a room with people talking about AI or the introduction. Into their workspace. And so we're just starting at grassroots, like literally starting at the beginning. And it's um a six hour workshop,
[01:32:03] Richard Webbe: Yep.
[01:32:04] Joanna Jones: face to face. And it's only it's uh it's maxed out at eight, just so that I can spend good one-to-one time with them because they're all bringing their own
[01:32:12] Richard Webbe: Yep.
[01:32:12] Joanna Jones: problem to the workshop.
[01:32:13] Richard Webbe: Yep.
[01:32:14] Joanna Jones: And it might be something really simple, but um it might be something more complex that needs to be then referred on. But yeah, just trying to find because there's people that are within these businesses that they might have AI or they're using other people in the organization are using it but they don't feel confident to ask the questions or to really get stuck into or why you're using that. I think we're this would be better. So I'm just um yeah that's on Saturday coming up so it's a pre-Christmas
[01:32:42] Richard Webbe: Well
[01:32:42] Joanna Jones: bash.
[01:32:42] Richard Webbe: done, Joe. I think that's negative because so many people have written AI slot books about how to master AI and they've not put a friggin' word down and they've just told AI, write me a book that tells people how to use you and printed it. And put the picture and face and name on the front that means Jack Shit.
[01:33:01] Dave Pengelley: Tell
[01:33:01] Joanna Jones: Yeah.
[01:33:01] Dave Pengelley: us how you really
[01:33:02] Joanna Jones: Well
[01:33:02] Dave Pengelley: feel,
[01:33:02] Joanna Jones: people
[01:33:02] Dave Pengelley: Richard.
[01:33:02] Joanna Jones: want experiences. People, yes, Richard, tell us how you really feel. But people genuinely now want to build trust with someone that they um know and they like. And it could be that they prefer to work with my jewula and not me. It's you know it just it's or whether it's Sean or whether it's Dave or Matt or you. So people find and gravitate to the people that they like. And that means that you do have to be yourself.
[01:33:26] Dave Pengelley: Hmm.
[01:33:27] Joanna Jones: You have to present visibly as yours, you know, just be yourself. And if people don't take up what you're you know putting out there as far as your work um and your education, then they'll choose somebody else,
[01:33:39] Dave Pengelley: That's
[01:33:39] Joanna Jones: and that's
[01:33:39] Dave Pengelley: that's
[01:33:39] Joanna Jones: great because it sort
[01:33:41] Dave Pengelley: that's
[01:33:41] Joanna Jones: of filters
[01:33:41] Dave Pengelley: very much this
[01:33:42] Joanna Jones: it
[01:33:42] Dave Pengelley: podcast.
[01:33:42] Joanna Jones: out.
[01:33:42] Dave Pengelley: We all are all who we are. And that's kind of the whole point. It's
[01:33:45] Joanna Jones: Yeah.
[01:33:45] Dave Pengelley: that sort of behind the scenes pub chat view that we've always gone for, that authenticity, and and that's a personality driven show as much as a news and AI show. And if people want to hear us prattle on, because they enjoy The banter, well, that's great. If they don't, okay, we're not their cup of tea, that's fine.
[01:34:00] Matt Slager: Joey, with
[01:34:01] Joanna Jones: Yeah.
[01:34:01] Shawn Hain: Yeah.
[01:34:01] Matt Slager: your your workshop, is it full already?
[01:34:05] Joanna Jones: It's up
[01:34:05] Matt Slager: So
[01:34:05] Joanna Jones: to
[01:34:06] Matt Slager: you
[01:34:06] Joanna Jones: six
[01:34:06] Matt Slager: got two more
[01:34:06] Joanna Jones: at the
[01:34:06] Matt Slager: thoughts?
[01:34:07] Joanna Jones: moment, so I've got two
[01:34:07] Matt Slager: Wow.
[01:34:08] Joanna Jones: spaces.
[01:34:08] Dave Pengelley: In
[01:34:08] Matt Slager: So
[01:34:08] Dave Pengelley: in
[01:34:08] Matt Slager: where
[01:34:08] Dave Pengelley: Perth.
[01:34:08] Matt Slager: do people
[01:34:09] Joanna Jones: Yeah.
[01:34:09] Matt Slager: go if they want to sign up?
[01:34:11] Richard Webbe: Good question.
[01:34:12] Joanna Jones: Well, they can look uh come oh right, okay. Um because I've been doing a lot of it with face-to-face conversations and you
[01:34:19] Dave Pengelley: Look
[01:34:19] Joanna Jones: for
[01:34:20] Dave Pengelley: up
[01:34:20] Joanna Jones: yapa.
[01:34:20] Dave Pengelley: Joe on LinkedIn and DM
[01:34:21] Joanna Jones: Um
[01:34:21] Dave Pengelley: her.
[01:34:21] Richard Webbe: Yeah,
[01:34:21] Dave Pengelley: Connect and
[01:34:21] Richard Webbe: look
[01:34:22] Dave Pengelley: DM
[01:34:22] Richard Webbe: up
[01:34:22] Dave Pengelley: with
[01:34:22] Richard Webbe: join
[01:34:22] Dave Pengelley: Joe.
[01:34:22] Richard Webbe: or linked the index.
[01:34:23] Joanna Jones: yeah. Thank you, Dave. You're on to it. Yeah, just look
[01:34:27] Dave Pengelley: Yep.
[01:34:27] Joanna Jones: me up on um
[01:34:27] Dave Pengelley: Yep.
[01:34:27] Joanna Jones: LinkedIn. You can either look up emu Appleai or uh Joanna
[01:34:31] Dave Pengelley: Yep.
[01:34:31] Joanna Jones: Jones,
[01:34:31] Dave Pengelley: She's or
[01:34:32] Joanna Jones: either
[01:34:32] Dave Pengelley: or go
[01:34:32] Joanna Jones: way.
[01:34:33] Dave Pengelley: go to
[01:34:33] Joanna Jones: And
[01:34:33] Dave Pengelley: AIOperatorspod.com,
[01:34:33] Joanna Jones: just DM me.
[01:34:34] Dave Pengelley: hit the guest tab, find Joanna, and there's a link on there.
[01:34:38] Joanna Jones: Yeah, that's a
[01:34:39] Shawn Hain: My
[01:34:39] Joanna Jones: better
[01:34:39] Shawn Hain: hair.
[01:34:39] Joanna Jones: way.
[01:34:41] Richard Webbe: Guys, be prepared. Dave's going to charge us to come on next year because his marketing is so good.
[01:34:46] Joanna Jones: Yeah, I know I I
[01:34:48] Dave Pengelley: You just
[01:34:48] Joanna Jones: yeah.
[01:34:48] Dave Pengelley: you just all have
[01:34:49] Joanna Jones: Chick
[01:34:49] Dave Pengelley: to buy a mug.
[01:34:49] Joanna Jones: ching.
[01:34:50] Dave Pengelley: I just need everyone to be turning up with with their branded sort of uh this is not a political podcast mugs.
[01:34:55] Joanna Jones: And so what's the time frame between receiving? Are we going to be able to get these
[01:34:59] Dave Pengelley: Yeah,
[01:34:59] Joanna Jones: before
[01:34:59] Dave Pengelley: yeah,
[01:34:59] Joanna Jones: Christmas?
[01:35:00] Dave Pengelley: a hundred percent. Um I think it just tells tells you the time frames on the site when you get order.
[01:35:07] Joanna Jones: Okay, so the kids' uh satisaks are already filled. They've got Richard's book in each of
[01:35:11] Dave Pengelley: And
[01:35:11] Joanna Jones: them
[01:35:12] Dave Pengelley: and
[01:35:12] Joanna Jones: and
[01:35:12] Dave Pengelley: and
[01:35:12] Joanna Jones: um
[01:35:12] Dave Pengelley: a t shirt.
[01:35:13] Joanna Jones: AI operators' mugs,
[01:35:15] Dave Pengelley: Yep.
[01:35:15] Joanna Jones: t-shirts. I'm done.
[01:35:17] Dave Pengelley: Excellent.
[01:35:17] Joanna Jones: Thank you.
[01:35:19] Dave Pengelley: Uh and there'll be more, there'll be more as we uh I've sort of mind some of the podcasts and and mine some of the things like this is not a political Podcast was one of the first things the AR pulled out when I read through the transcripts and said, You guys say this a lot. This should be a meme. Stick that on a mug.
[01:35:36] Joanna Jones: Oh, that'd be a great
[01:35:37] Dave Pengelley: Yeah.
[01:35:37] Joanna Jones: t shirt.
[01:35:38] Dave Pengelley: So
[01:35:39] Joanna Jones: And do the t shirts come in white, Dave?
[01:35:42] Dave Pengelley: no, though because the design is made to contrast. Do you want a white t-shirt? I can I can look at making a white one. I'll
[01:35:49] Richard Webbe: I
[01:35:49] Matt Slager: I
[01:35:49] Richard Webbe: live
[01:35:49] Matt Slager: love
[01:35:49] Dave Pengelley: see.
[01:35:49] Matt Slager: it.
[01:35:49] Richard Webbe: in Melbourne. I live in Melbourne. We only wear black t shirts, so don't change
[01:35:53] Shawn Hain: Yep.
[01:35:53] Richard Webbe: it for me.
[01:35:54] Shawn Hain: Yep.
[01:35:55] Joanna Jones: Doom
[01:35:55] Dave Pengelley: Yeah.
[01:35:55] Joanna Jones: doom doom
[01:35:56] Dave Pengelley: Apparently
[01:35:56] Joanna Jones: doom
[01:35:57] Dave Pengelley: like 'cause I've
[01:35:57] Joanna Jones: doom
[01:35:57] Dave Pengelley: looked at the print
[01:35:57] Joanna Jones: doom
[01:35:57] Dave Pengelley: on demand
[01:35:58] Joanna Jones: do
[01:35:58] Dave Pengelley: thing and the most popular color t shirt that people buy is black.
[01:36:01] Richard Webbe: Yep.
[01:36:01] Dave Pengelley: I think
[01:36:02] Shawn Hain: They
[01:36:02] Dave Pengelley: because it
[01:36:03] Shawn Hain: contrast
[01:36:03] Dave Pengelley: just
[01:36:04] Shawn Hain: so well with almost everything.
[01:36:06] Dave Pengelley: Yeah. It goes
[01:36:07] Joanna Jones: yeah, but I was born in the seventies. Sorry, did I hear seventies?
[01:36:11] Dave Pengelley: you wanna tie
[01:36:11] Joanna Jones: So we
[01:36:11] Dave Pengelley: so
[01:36:12] Joanna Jones: like
[01:36:12] Dave Pengelley: do you want
[01:36:12] Joanna Jones: light.
[01:36:12] Dave Pengelley: to tie dye one? I can't do that for
[01:36:14] Joanna Jones: Oh
[01:36:14] Dave Pengelley: you, no. It's
[01:36:17] Joanna Jones: Dave,
[01:36:17] Dave Pengelley: not a thing.
[01:36:18] Joanna Jones: serious?
[01:36:22] Joanna Jones: This is not a political podcast.
[01:36:24] Dave Pengelley: There's no with
[01:36:24] Joanna Jones: Richard.
[01:36:24] Dave Pengelley: there's not a drug making podcast, Richard.
[01:36:29] Joanna Jones: Oh no, I can tie dye my own, but
[01:36:31] Dave Pengelley: Okay.
[01:36:31] Joanna Jones: thanks, Dave. If I get
[01:36:32] Dave Pengelley: Yeah.
[01:36:32] Joanna Jones: a white one, I can
[01:36:35] Dave Pengelley: Uh
[01:36:35] Joanna Jones: that was a Saturday afternoon event
[01:36:38] Dave Pengelley: my
[01:36:38] Joanna Jones: on a
[01:36:39] Dave Pengelley: my daughter
[01:36:39] Joanna Jones: regular
[01:36:39] Dave Pengelley: doesn't mind doing
[01:36:39] Joanna Jones: basis.
[01:36:40] Dave Pengelley: a bit of tie-dying. She's she's twelve. She loves
[01:36:42] Joanna Jones: Oh
[01:36:42] Dave Pengelley: it.
[01:36:43] Joanna Jones: unto my own heart.
[01:36:45] Dave Pengelley: All right. Well, thank you so much, everyone. It's been a
[01:36:48] Shawn Hain: All
[01:36:48] Dave Pengelley: wonderful
[01:36:49] Shawn Hain: right.
[01:36:49] Dave Pengelley: twenty twenty six. Thank you, ladies, for joining us today. Thank you, Sean, for dropping in again. That's been a pleasure. And to to to uh Uh Kate who came in and out.
[01:36:58] Joanna Jones: I don't like what I
[01:36:59] Dave Pengelley: Thank
[01:36:59] Joanna Jones: don't
[01:36:59] Dave Pengelley: you to
[01:36:59] Joanna Jones: want
[01:37:00] Dave Pengelley: everyone
[01:37:00] Joanna Jones: to do.
[01:37:00] Dave Pengelley: who's been part of the show, all our other guests. We're going to continue on for the next few weeks. So I've got three operator stories to record this week already booked in. I'm still open for recordings next week, and then I'm taking December off. And so we've still got about a month of operator stories out. So that'll probably run much run pretty much up until Christmas being released on Mondays. And we'll see how we go with field notes on Fridays. But
[01:37:26] Shawn Hain: Uh
[01:37:26] Richard Webbe: Merry
[01:37:26] Shawn Hain: we'll
[01:37:27] Richard Webbe: Christmas.
[01:37:27] Shawn Hain: make this for everyone.
[01:37:28] Richard Webbe: Merry
[01:37:28] Manjula Kerai: Christmas,
[01:37:28] Richard Webbe: Christmas.
[01:37:29] Manjula Kerai: everyone. Okay, so there's
[01:37:31] Joanna Jones: Merry Christmas,
[01:37:32] Dave Pengelley: I don't
[01:37:32] Joanna Jones: everybody.
[01:37:32] Dave Pengelley: have a special N of E graphics. We're just gonna go the usual play
[01:37:35] Manjula Kerai: a lot
[01:37:36] Dave Pengelley: out.
[01:37:36] Manjula Kerai: of
[01:37:38] Dave Pengelley: Thank
[01:37:38] Manjula Kerai: things.
[01:37:38] Dave Pengelley: you all.
Related episodes
Keep going
Chapters
Jump to a section
0:00Cold open
6:03The last live panel of the year
12:39Christmas shopping bots and the limits of AI slop
18:29The AI jobs debate at the Christmas table
30:06Picking the model for the job
36:15Which AI habits are worth keeping?
42:14Ask me one thing at a time
48:38What Shawn cut from his AI framework
1:00:07Start with the business outcome
1:07:10How AI changes what is worth prototyping
1:12:38A game prototype in action
1:17:57An election explainer and the question of bias
1:24:19Looking back at a year of changing tools
1:30:26Where bias enters an AI answer
1:36:44Year-end thank-yous