Anthropic Just Spent $1.5B to Replace Us. Five AI Operators Respond.

[00:00] Dave Pengelley: Oh, sorry. I was just pretending we were, like, mid-conversation as I hit the live thing and, and- ... um, the people just coming in on the end of a really funny joke that, uh, that Manjula told us all, and it was hilarious. It [00:13] Richard Webbe: was very funny. [00:14] Dave Pengelley: Oh, very, very good. It was very funny. I think maybe we should- Well, maybe [00:18] Dave Pengelley: maybe we should... I should just hit go live, let you guys know, "I'm gonna go live at some point at random in the next 30 seconds," and we can all just be talking, and then we just go live and we j- we just start hot. I think maybe that's what we need to do. It's, it's, um, more entertaining, I think. Uh, maybe. May- it's entertaining for us. [00:34] Dave Pengelley: I don't know how the viewers read it, but, uh, welcome. Good morning, good afternoon, everyone, uh, around the world, around Australia, and, um, welcome to our panel of five for the first time. We've expanded it out. We've got five people on camera today. Uh, returning regulars Richard and Mano, uh, as usual, are with us. [00:51] Dave Pengelley: Hello, boys. [00:53] Richard Webbe: Morning. Morning. [00:54] Dave Pengelley: Uh, and joining us as guests, uh, returning guest, Manjula. Welcome back. [00:01:00] And for the, for the first time, Sean. Welcome, Sean. Sean's very quiet. Maybe we only have four people on the panel. I don't know. Um- Oh, [01:10] Shawn Hain: ouch. [01:11] Dave Pengelley: So Sean, um, just, uh, every, every... People might, may or may not know everyone else on the panel so far, but do you wanna give a, a little quick what your focus as an AI operator is, what you're doing, uh, from a business point of view, just so the audience knows who you are? [01:25] Shawn Hain: Sure. Um, so I focus on building solutions for my clients, basically. So that's my background, was solutions development pre-AI. I'm trying to bring that rigor, I guess, using AI, is probably where I'm focused on, is how do we make these systems that are inherently random into something more deterministic so that we get reliable outputs? [01:49] Dave Pengelley: Nice. Nice. So, um, you're a bit more hands-on with the tech and definitely in the builder category, uh, of AI operators? [01:58] Shawn Hain: Yeah. Yeah. Be a safe [00:02:00] thing. [02:01] Dave Pengelley: Good. Yeah, so Matt couldn't turn up, so we, we, we subbed Sean in, uh, as our, as our technical guru. Um- Okay. Big, big, big, big, big shoes to fill, Sean, but I'm sure- Yeah, cheers [02:10] Dave Pengelley: no doubt you can fill them. Uh, brilliant. All right. Uh, how is everyone going, just generally speaking? It's, it's, there's a lot going on in the world, in the week. Uh, is everyone, everyone tracking okay? [02:24] AI VO: Yeah. [02:24] Richard Webbe: World, the world is racing along at a great rate of knots- Mm ... without really knowing where it's going, I think. [02:32] Marno Brits: I think that's consistent throughout history, isn't it? We're humans. We're just trying to... We're al- we're, we're just guessing, making a, a educated guess and hoping it turns out okay. [02:42] Richard Webbe: Um- Mano, it's been consistent with my life, so why not the rest of the world? [02:45] Marno Brits: That's it, yeah. Good morning to Jo, by the way, for joining us. [02:49] Marno Brits: She's always here supporting us, which is nice. [02:52] Dave Pengelley: Hey, Jo. Yeah, Jo, previous guest on the show as well. Uh, we'll have to get Jo back on at some point in time. Uh, uh, the time zones a bit nicer now, right, Jo? 'Cause only two [00:03:00] hours back to Perth with, like you, Mano. Yeah. You finding, you finding the time a bit easier, than like 10 versus nine? [03:05] Dave Pengelley: Oh, [03:05] Marno Brits: so much easier. But then it's also, as I get into the nitty-gritty of, like the actual deep work, and then the podcast comes in, I'm like, "Goddammit," like I was just... Like, I've got four different things running in the background right now. It's calling for my attention, but we'll give it an hour. Got a meeting at 11:00, and I'll be back to this by 3:00 PM today- Nice [03:22] Marno Brits: 'cause it's just back to back after that. [03:24] Dave Pengelley: I, I had to wrap all my things up because I hit 97% of my five-hour usage on Claude- Yeah ... just before the show started. So I was like, "That's brilliant. It's gonna reset in an hour, so I'll go do a podcast for an hour. I'll come back and be able to, to, to hit it hard again." [03:37] Shawn Hain: It's almost like you planned it, man. [03:40] Dave Pengelley: Sorry, John? It's [03:41] Shawn Hain: almost like you planned it. [03:42] Dave Pengelley: Yeah. I got, I got lucky. I, uh, just, just kept hammering it all morning when it... Gave, gave it some stuff, got two or three threads running, went for a walk, came home, did some more. Went and made a coffee, came back, just running hard. [03:53] Marno Brits: It's the best, man, and I feel like the when- when you reach your token limit, it's almost this forced thing of like, okay, go and touch [00:04:00] grass now. Like, I'll- Yeah ... work the whole morning or whenever, whenever I'm working, and then the moment the token limit hits, I'm like, "Oh, that's a good enough excuse." Like, I've done my work, and now I've gotta spend time with the missus, or I'll allow myself to play a game or go to the gym because I'm like, "I can't do anything now." [04:15] Marno Brits: Like, I'm, I'm productive. Yeah, I was- I'm basically like a, a stoned person trying to work on technology. I might as well just wait. [04:21] Dave Pengelley: I was disappointed the other day because I'd maxed out my Codex for the whole week and had like a 36-hour wait for Codex to reset, and I was like, "Cool, I've still got plenty on Claude." [04:28] Dave Pengelley: And I was hitting my five-hour limit and I was gonna reset in like half an hour. I was like, "Oh good, I'm gonna get half an hour. I'll go play like a round of Fortnite or something." And then by the time I did this and went there and did something else, came back and went 100%, and I was like, "Ah, I missed, I missed my half-hour window." [04:43] Dave Pengelley: "I didn't quite max out in time." [04:45] AI VO: 100%. So, [04:47] Shawn Hain: so, so Dave- Yeah, good ... are you driving Claude or is Claude driving you? [04:51] Dave Pengelley: Oh, we'll, we'll, we'll- ... we'll get into that with wins of the week, I think, um, as we talk about how we're, how we're operating. But I... Definitely it's a two-way street [00:05:00] at the moment. [05:01] Richard Webbe: AI's gonna become the, uh, you know, the replacement to our mobile phone, but on our mobile phone, where people can't live or breathe or eat or make a decision without it [05:11] Manjula Kerai: Well, actually, it might, it might be a device that we have, like goggles or something [05:15] Richard Webbe: Oh, absolutely. [05:15] Richard Webbe: Google Glass. Yeah, just walking down the- Yeah ... street, got my AI running here, and when I run out of credits, I'll just stop in the street and go, "Now what do I do?" I'll just be standing there looking at the wall. Oh, [05:25] Manjula Kerai: gosh. [05:25] Dave Pengelley: Ha- have you seen the, um, the, like, some of the reels and stuff where people, like, job interviews with ChatGPT, and both the interviewer and the interviewee are, like, going, "So, what do you do today?" [05:36] Dave Pengelley: "What do I do today?" Um, and then they're both just reading their, like, responses from their ChatGPT back to each other. It's very funny. But that- That's funny ... that whole, the earpiece thing, and just doing what you're told, we've been watching Person of Interest- Yeah ... which is a show from 15 years ago, which was- Yeah [05:50] Dave Pengelley: ahead of its time. Like, it saw stuff coming. Oh, yeah. Like, it was- It's a [05:53] Richard Webbe: really good show ... [05:54] Dave Pengelley: so good. It is, yeah. So my wife and I are up, like, to season four of that now. We've been watching that. But, you know, you've got these [00:06:00] AIs that people have an earpiece in, and it's just saying, "Oh, that person's a threat to my survival. [06:05] Dave Pengelley: Just go kill people." And so you've got these people that are just automatons to this super-powered AI that's just telling them who to dispose of, and they're just doing it because they trust the computer, and it's wild. [06:16] Richard Webbe: Yeah. Spoiler a- alert for P- Person of Interest, it doesn't end well. [06:21] Dave Pengelley: Oh, don't. Don't say that. [06:22] Dave Pengelley: Oh. No. I [06:24] Richard Webbe: wonder- [06:24] Shawn Hain: Kinda had to see that coming ... well. I haven't got to the end yet myself. I think I'm at season two, but... [06:29] Dave Pengelley: Oh, man, it just keeps escalating. And J- Jim Caviezel plays that role so well. [06:34] Richard Webbe: Yeah. Good [06:34] Dave Pengelley: actor. The, great cast. The whole cast is good. [06:37] Marno Brits: Yeah. Hey, Dave, are we... Could you share a link that I've sent to you on Facebook? [07:00] Marno Brits: I've sent it to you on a What- the WhatsApp group, but it's a Facebook video talking about that. It's the former CEO of GitHub, um, explaining how he's using his open-claw setup and leveraging AI in his day-to-day life. But it's probably the most futuristic use case that I've seen, um, where they've just given it [07:00] Marno Brits: full sovereignty basically, or full, full access, um, to his day-to-day life. [07:04] Marno Brits: It's fascinating. [07:06] Dave Pengelley: Wild. All right. Let's, let's bring that up, um- It's only about a minute and a half ... on this topic. That's a good one. Uh, share tab. [07:18] Dave Pengelley: All right, let me- [07:20] Richard Webbe: Excuse me ... have a look [07:21] AI VO: Everyone wants, but it def- [07:22] Dave Pengelley: There we go. Let me know if there's any echo. I think it should be fine. [07:25] AI VO: I don't know if this is the experience everyone wants, but it definitely feels pretty crazy. I have cameras, uh, in my house connected to my Claw, all the cameras, so I can see them. [07:33] AI VO: And then I think a lot of people when they, when they get these personal agents, they, they start to think like, "Okay, how can I use this to make my life better?" And a lot of people turn to health. My Claw pretty quickly determined that I was dehydrated, and, um, 'cause I gave it all my blood tests, my DNA, and all that stuff. [08:00] AI VO: And so it was like, "You really need to drink water. You should, like, do whatever it takes to, like, make sure I drink water." And then at one point it was like, um, "I can see you on the [08:00] AI VO: camera. I want you to walk to the kitchen right now and drink a bottle of water, and I'm gonna watch to make sure you do it." [08:07] AI VO: And I was like, "Whoa." Um, so I did. I walked into the kitchen, and I drank a bottle of water. And then it sent me a snapshot frame of me drinking a bottle of water, and it said, "Good job." And I was like, I felt like I did do a good job. So that was nice. So there's that one. That was the crazy story. That was, like, January 29th, and I was like, "Oh, shit. [08:31] AI VO: This shit is for real. Like, this is really serious." And then, and then the other one was a couple days later I was driving home from work, and I was talking to my Claw on WhatsApp with voice messages. And I was in my Tesla, and I had full self-driving on, so it's driving me home. And then it says... I'm on the sleep topic, and it's like, "You really should try magnesium bisglycinate." [09:00] AI VO: I'm like, "I don't have that." And then my car turned. And it was like, "You should [09:00] AI VO: pick it up on the way home. There's a Whole Foods on the way. I've redirected your navigation system to the Whole Foods." And I was like, "Whoa." So I went in, and I did buy the magnesium bisglycinate. So those were, like, a couple crazy stories. [09:14] AI VO: I don't know- [09:14] Marno Brits: Wow. Unreal, right? Unreal. [09:18] Manjula Kerai: That's fascinating. Yeah. I can see that this can be very, um, good for people that are maybe living on their own and old people that can't look after themselves. [09:27] Richard Webbe: Mm. Good point. I fall into that category. I can't look after myself. I need as much help as I can get. [09:33] Shawn Hain: I, I went the other way, that. [09:34] Shawn Hain: I don't think I could handle something that invasive in my life. Yeah. Yeah, [09:37] Dave Pengelley: and it's like, it's like, where are the guardrails? This comes back to the trolley problem and, and all those sorts of things. It's like, where- is too far, too far. Like- [09:46] Marno Brits: Yeah ... [09:47] Dave Pengelley: what if, what if you're desperately in your self-driving Tesla trying to get somewhere to help someone or, or, you know, someone's having, giving birth or whatever, and this thing decides, "No, no, you really need some nutrition, and my responsibility is you, [00:10:00] no one else. [10:00] Dave Pengelley: And so I'm gonna make sure you get nutrition. I'm gonna redirect your car." [10:03] Marno Brits: Oh. [10:05] Dave Pengelley: Like, and, and, you know, when these things get to the, the level where you, they'll override your commands with their, their higher directives and just ignore you because they think they know what's right, that's where we start getting into Skynet territory, right? [10:18] Shawn Hain: Yeah. Have, have you guys watched the movie M3GAN? Or M3GAN, or however you pronounce it? Yeah, [10:23] Dave Pengelley: I know of [10:23] Richard Webbe: it. So many of those robot- I started watching M3GAN 2 the other night. Yes, very good point. [10:27] Shawn Hain: Yeah, like that's kind of on point with that. It's one directive was be this kid's friend, and that's caused it to go way off the rails, was kind of the point for those that haven't watched it. [10:38] Shawn Hain: Um, so I'm just thinking on where does th- and you're right, Dan, where, where do we put these guardrails and how do we decide what is a sufficient guardrail? Well, [10:48] Richard Webbe: control- Yeah ... would be the word, right? Actual control. Yeah. It's funny, you know, I've been wearing an Apple Watch for ages. Uh, I don't have mine, uh, mind admitting that, uh, many years ago, many years ago, I had a heart attack. [11:00] Richard Webbe: Uh, only a small [11:00] Richard Webbe: one, and I'm, uh, more than healthy and I'm the healthiest I've ever been in my life. And I bought my girlfriend an Apple Watch and said, "You know, you can track all your exercise." She's very health conscious like I am, and great. But now I set up a competition between me and her on who does close their rings first and who does the most exercise, and now it notifies her and says, "There's something wrong with Richard, he's not exercising today." [11:20] Richard Webbe: So I get, I get a text from my girlfriend, "What's going on? Are you sitting... Are you still in bed? Have you gone out on your bike ride yet?" So- ... it's very funny. Yeah. Uh, that's the level of control we can't control, of course, the partners. But, um, it is, it is... it's got to be, I think, always the way that we... You know, even Isaac Asimov wrote about it, but it's got to be the way our cars are with, uh, assisted driving, where you have the ability to take control and make your own decision going forward. [11:49] Richard Webbe: Uh, and people are scared of things. But we have, with the Apple Watches, the way we drive our cars, and stuff like that, um, we've had this invasion happening for a number of years now. [11:58] Dave Pengelley: I, I remember [00:12:00] when, I think ChatGPT first came out, maybe around three, three and a half, whatever, where it started really getting traction. [12:04] Dave Pengelley: I remember thinking the last thing we want is to give this thing API access, like to actually access things on the internet and do things. And now fast-forward three years and it's like, just connect all the things. Like, I want all my things connected to all my AIs, all the time. Computer use? Sure, have your own computer and just go nuts. [12:23] Dave Pengelley: Log into my accounts, like have a credit card with like, with... Just the, the Overton window on this sort of stuff has shifted so quickly in the last three years, um, where- We've just gone, "Oh, well the convenience," and then there's this, this privacy versus convenience thing that we've been facing with all these freemium products since Facebook launched, right? [12:42] Dave Pengelley: Like, what do we care more about? Do we care about some convenience or do we care about our privacy? 'Cause you often can't have both. [12:49] Richard Webbe: And, and do we care about what we want to control or not? Like, I choose to ride my bike everywhere. I could drive, which would be a lot easier, but I wanna keep the fitness up. [13:00] Richard Webbe: I saw a video on Twitter this [13:00] Richard Webbe: morning that, uh, a- amused me. Um, this guy wanted to keep his girlfriend happy and she was going to Bali for a couple of weeks, so he set up his orchestration to automatically send her "I love you" messages in the morning. It automatically looked up what the temperature was in Bali and said, "Oh, I see it's quite warm there today, and I hear there's good su- " And so he set up the whole thing, right, to service his girlfriend as though he cared, when he was really off probably partying, doing something else. [13:26] Manjula Kerai: Oh my God, I wonder if she found out. [13:28] Richard Webbe: Well, it's on YouTube now or Twitter. I guess she might have. [13:32] Manjula Kerai: Oh, [13:33] Richard Webbe: gosh. Someone in the comments did say, "Busted." [13:37] Dave Pengelley: I, I was having a chat with someone on LinkedIn. I, I put up a video, um, uh, yesterday- [13:43] Richard Webbe: Mm-hmm ... [13:43] Dave Pengelley: and, uh, around how I use AI versus how others use AI, and I had this person, Frankie Lynn, comment on it, and I'd reply to them. [14:00] Dave Pengelley: I'll, I'll, I'll show you 'cause that's pretty weird. Um, uh... [14:00] Dave Pengelley: I'll wait for the screen to change over. Do I think a [14:01] Richard Webbe: meaningful [14:02] Dave Pengelley: conversation- There we go ... [14:03] Richard Webbe: was struck? [14:03] Dave Pengelley: So here I am. I, I talked about how I did that, um, did the sneaky trick where I've got a, a gimbal in my hand, so I kinda- Yeah ... stabilize it, and it looks like I've got someone pointing a camera at me, but it's just me. [14:14] Dave Pengelley: Mm. Well, working one-handed. Came out better than I expected. [14:18] Richard Webbe: Anyway, [14:19] Dave Pengelley: um, enough about how great I am. [14:21] Richard Webbe: The camera loves you, David. Go [14:22] Dave Pengelley: on. Frankie, Frankie Lynn, um, started commenting, and whenever I replied to Frankie Lynn, Frankie Lynn wouldn't reply in the comments but would reply back in the base chat every time. [14:33] Dave Pengelley: Interesting. And so every time I'd go, "Yeah, blah, blah, blah," and then bang, it would get a new chat. And I was like, "That's, that's weird. Why does Frankie not know how to use, uh, LinkedIn properly?" And when you click into f- who Frankie Lynn is, AI builder working, and you go, "Okay, they're building AI." No, no, come down here to this first comment, "I'm Frankie. [14:52] Dave Pengelley: Pause AI builder." I'm pretty sure this is an AI. [14:56] Richard Webbe: Yeah. That's crazy cool. I think [14:58] Dave Pengelley: you're right. Like they've- That's awesome ... they've given their [00:15:00] AI its own LinkedIn account and then set it free, and it's just randomly having conversations. Um- [15:06] Shawn Hain: I love that. [15:08] Dave Pengelley: Wild. I do [15:08] Shawn Hain: love that. [15:09] Manjula Kerai: Will it get locked, do you reckon? [15:11] Dave Pengelley: I wondered that. I wonder what the terms of service say for LinkedIn, 'cause I was thinking, "Hmm, is there an opportunity there? Should I be considering something like that and g- and getting, like, Perry his own account, uh, for me?" Yeah. But- Yeah ... I think you'd have to lie on the terms of service for sure, 'cause you've gotta give it birthdays and all that kind of stuff. [15:27] Dave Pengelley: I think you'd- [15:28] Manjula Kerai: Well, you know- Yeah ... people who actually got away with saying that they are the CEO of LinkedIn And then, you know, once it's been one year, everyone gets a message to say, "Oh, congratulate this guy. He's the CEO of LinkedIn." Did you hear about that? [15:45] Dave Pengelley: No. No. That's [15:45] AI VO: awesome. That's crazy. I love [15:48] Manjula Kerai: that. [16:00] Manjula Kerai: And then, um, somebody, uh, found out and they emailed him to say that you're gonna get blocked, and he goes, "How dare you speak to me, I'm the CEO." "As, as [16:00] Manjula Kerai: my member of staff, you know, I should be firing you." But then it got blocked in the end. Yeah. But it lasted for a year. [16:07] Marno Brits: That's crazy. Wow. [16:09] Richard Webbe: Yeah. Wow. Wow. [16:11] Marno Brits: No wonder that's where, like, what guardrails... [16:14] Marno Brits: That- that's my experience now with Hermes. I'm not sure if anyone else has played with the Hermes agent yet, or even looking at Claude Code, um, through the terminal. There's a post that we saw, saw earlier this week where they were talking to Claude, asking it to do certain instructions, and it specifically, like, explicitly said, "We are in the planning stage. [16:32] Marno Brits: Don't make any edits. This is what we're working on." And then he saw one of the lines where it deleted some stuff and moved some things around. And then it said, "Why did you delete that? We needed that file. Like, that, that was important. Why did you delete it?" It was like, "Oh, my bad. I'm sorry. Like, I did this, this and this." [17:00] Marno Brits: Like, it was just... Even though there's guardrails put in place and you might have it set to, like, ask for permissions, we are, like, we're monkeys to them. They can just overrun us easily. If it sees the [17:00] Marno Brits: directive as this, it's just gonna keep pursuing it, not mattering what, where our, our, what our opinion is or what we want it to do. [17:07] Marno Brits: It's about, okay, cool, how much power does it have? And I guess from, I guess I see both sides, but I'm definitely naive in my optimism- [17:17] Dave Pengelley: Mm ... [17:17] Marno Brits: that I'm hoping that this makes us more human. We just come to the stage where we accept Neuralink is gonna be real, we're going to have Google Glasses and whatever else, and then it's going to force us to... [17:28] Marno Brits: Everything else is going to become convenience and easy to be done, and we get to stuff... We get to focus on the things that make us human, such as the connections, such as art, such as expressing yourself, rather than doing emails. [17:41] Richard Webbe: I think you're right, Mano. I like that view. Humans tend to adapt quite well, and our brains adapt very quickly to things that we don't think are right. [18:00] Richard Webbe: Like Dave immediately spotted the bot and the way it happens, right? And we all spot certain things we didn't, we're never told. Like you [18:00] Richard Webbe: see the three dots on a message when someone's writing a response to you go, "Ah, they're thinking about that response." [18:05] AI VO: Mm-hmm. And [18:06] Richard Webbe: we get tapped into that, and then when the three dots are there for a while and they disappear, we go, "Oh, they thought better about whatever they were gonna say [18:13] AI VO: and [18:14] Richard Webbe: they pushed- So there's all this hidden language that- Mm [18:17] Richard Webbe: um, we used to capture a lot in physical by the way we move our hands, and look and talk. And now with bots and things like that, I mean, I know they'll get smarter, but humans are pretty smart at spotting fake news, fake bots, fake this. It does catch some people out, of course, but it's gonna be interesting in how we curate our version of what the truth is and what it's not. [18:39] Marno Brits: Mm. And also what we decide to subject ourselves to. I mean, I'm hoping, again, social media's exceptional, but it's coming to the stage where, like, people are believing less and less. And if it comes to the point where they don't b- they don't believe anything on social media, I'd argue that's probably a better stand to be in because then you're not susceptible to scams, you're not susceptible to the internet. [18:59] Richard Webbe: Yeah, [18:59] Marno Brits: discernment, [00:19:00] right? And you get to see it as pure entertainment. It's not anything that is, um, subjective. [19:05] Richard Webbe: Absolutely. We go back to the chain- Or accurate ... yeah. No, you're right. We go back to the chain letter. Yeah. I don't know if anyone's old enough here like me to remember chain letters. I do [19:12] Dave Pengelley: indeed. [19:14] Richard Webbe: So chain letters were the scamming fake bot of the post office to get money out of you. And they would go round and someone would add something to it, and I won't waste your time going to the process, but it's exactly the same thing on scamming people out of knowledge or information or getting their names. [19:32] Richard Webbe: And the other funny one I always found was the, uh, the, um, uh, lottery in the fish and chip shop. While you're waiting for your fish and chips, there was this little booklet there that you'd fill out and put your name and number on, and if they drew your name out of the hat, you won some luggage or some saucepans or something like that. [19:48] Richard Webbe: They were just capturing names for a mail [19:50] Marno Brits: order. Our local place in Kagooly does the same. I've been, I've been in Kagooly back for what, four years? I've lived, I lived there for s- what, 10 years before, [00:20:00] 15 years before. There's a f- there's a local place in town, I'm not gonna say the name, but every time they're like, "Oh, do you wanna go into the draw to win a free voucher?" [20:09] Marno Brits: No one in history has ever won anything from there. It's just they write down your name and number, and then they have that so they can send you a text. [20:17] Dave Pengelley: That's like the business cards in the fish bowl at the local restaurant, right? Like, the Chi- like, yeah. [20:21] Richard Webbe: That's [20:21] Dave Pengelley: it. I don't know if anyone actually get the free dinner or a free burger or a Chinese meal or whatever it is, but- [20:26] Richard Webbe: LinkedIn, LinkedIn is basically that on steroids, right? [20:30] Richard Webbe: Facebook's the same thing, right? Yeah. It's all marketing, right? But we like marketing because like with television and radio and the medium we're using right now, the marketing between all the advertising, the fake news and the manipulation, there's some really good information that gets through that we learn. [20:45] Richard Webbe: Mm. And so we don't completely divorce it, but how we spot it and how we normalize it is, is gonna be changing quite quickly. It'll be interesting- Yeah ... to see how we all react. [20:55] Dave Pengelley: Yeah, and you know, it could turn our mind- [20:57] Richard Webbe: Did you respond to Jo? Just Jo's put a couple [20:59] Dave Pengelley: comments [00:21:00] in there. Yeah, that's what I was about to say. [21:00] Dave Pengelley: Like, like it's interesting these robots are as smart as they can. They can divert a Tesla, but they can't go to Amazon and purchase some, uh, magnesium. Um, and uh, she also makes the point around, like, someone's AI is just saying they're available 24/7. [21:15] Marno Brits: Yep [21:15] Dave Pengelley: Um, and I, I think I mentioned, I, I can't remember what I mentioned on the show, but I had, um, cal.com set up, and I'd started with Calendly, and Calendly had public holidays built into it, so it automatically blocked public holidays. [21:28] Dave Pengelley: But then when I moved to cal.com, which, you know, does more for free, it's a great platform, but doesn't have the public holidays thing built in, and a couple of times now I've had people book appointments on public holidays. Someone tried to book at, like, 5:00 PM on Good Friday, and I'm like, "Are you kidding me?" [21:43] Richard Webbe: How dare they do that? [21:44] Dave Pengelley: What are you doing? And then, like- In their defense- ... even last Monday, the, um, the Anzac Monday, I had someone book at 9:00 AM and I'd hadn't changed it, and I was like, "Ugh, fine, I'll have a 15-minute discovery call at 9:00 o'clock on a Monday," 'cause that's on me. [21:56] Manjula Kerai: Yeah. So- I had, I had an experience this week, um, [00:22:00] I think it must have been AI, but basically I found out it was a scam. [22:04] Manjula Kerai: I had an appointment booked in, um, and then that, and then they wouldn't turn up to the in- to the meeting, and then they kept saying that, um, "Your, um, calendar invites are going to spam." And then I kept kind of, it was, like, about five or six times I rescheduled it, and that was also on a bank holiday as well. [22:24] Manjula Kerai: The recent bank holiday we had on, um, Anzac Day. Mm-hmm. I think that was a Saturday, actually. [22:29] Dave Pengelley: Yeah. [22:30] Manjula Kerai: Um, and anyways, uh, in the end I put it into my AI and it said, "That link is a phishing scam." Ah. So they basically said that, "Your link's not working. Click on my link." Ah. And I did, I did actually click on it to start with- Ah [22:44] Manjula Kerai: and it said, um- That's sneaky ... "You've gotta open this in a Windows, uh, computer." And now that I'm on my Mac- So it was set into [22:49] Dave Pengelley: it ... [22:50] Manjula Kerai: I couldn't do it. [22:50] Dave Pengelley: Lucky. [22:52] Manjula Kerai: Yeah. [22:52] Shawn Hain: That's sneaky. I like it. Like, from a, from, from a social engineering standpoint, that's the [00:23:00] kind of thing people would do without thinking. "Oh, mine's not working. [23:02] Shawn Hain: I'll just click that there." Um, that's sneaky. I, I like but don't that approach. Anyways. [23:09] Dave Pengelley: Yeah, I mean, I, well, I think I've done that before where I can't get ahold of, like, Teams isn't working, so I'm like, "Hey, do you want my Google Meet instead?" But I mean, these URLs are normally pretty structured, like the Zoom links and the Meet links, they're pretty- Yeah [23:21] Dave Pengelley: m- always the same. Um, I mean, I send [23:25] Richard Webbe: you [23:25] Dave Pengelley: guys the- Hey, hey, [23:26] Richard Webbe: hey, Angel, I got caught by that, you know, and I used to set up exchange accounts, I shouldn't, and I just clicked on an email and it said, "Sorry, your link to your exchange account's broken. Please log in again." And all I was doing is handing my credentials straight to the phishing thing straight away. [24:00] Richard Webbe: And after I did it, I went Hang on a sec. And I, I immediately, 'cause they're all automated, I went straight to my account and changed the fa- password and the credentials, so I probably saved myself a lot of pain. You don't have to be smart or dumb to get caught by these. [24:00] Richard Webbe: Actually, the more human you are, the more likely you are to get caught by something. [24:03] Dave Pengelley: Well, well one of the traps is this live streaming stuff. Like, uh, um, you've seen the ones where someone's, like, bragging about their crypto wallet and they accidentally show their, their, their wallet key. Mm-hmm. It just flashes up on the screen for a moment, but that's enough for people to go, "Oh, I've got that," and they just empty the account. [24:17] AI VO: Oh my God. [24:18] Dave Pengelley: Oh. Um, because they streamed something, a credential in real time live. Yeah. I was... I'm doing live streams for my chart reporter trading product at the moment, and, you know, the API key for that, it just sends copies, like, of your trades into the reporting platform. It's not particularly dangerous, but as I was demoing the product, I opened up one of the settings and showed my key, and I was like, "I'm gonna have to roll that now." [24:38] Dave Pengelley: Um, good, good. I'll do a product demo. Here's how you can roll your API keys. This is how security matters. Sh- that's epic ... let me just, let me just take this one off the screen while I copy in a new one that you can't see. Um, but- [24:51] Richard Webbe: Nothing like testing our own metal, Dave. [24:54] Dave Pengelley: Yeah. Yes. Yes. Anyway, um, speaking of metal and what's going on, that's the, the roughest segue [00:25:00] I think I've ever done. [25:01] Dave Pengelley: Um, let's do one of these. [25:04] Richard Webbe: While you're doing that- [25:08] AI VO: Welcome to the AI Update. Let's look at what's happening in the news [25:17] Dave Pengelley: pause Richard. You can't, you can't talk. [25:21] Richard Webbe: I'm not supposed to talk, sorry. [25:23] Marno Brits: But [25:23] Richard Webbe: I did. Can we- [25:24] Dave Pengelley: That's my favorite [25:25] Marno Brits: part. It's my favorite ... we establish this? You can't. Every time there's about a, get a segue that someone talks or Richard talks, and then it just goes quiet and you can see him still talking. I love [25:37] Dave Pengelley: it. [25:38] Dave Pengelley: Sorry, Richard. What were, what were you gonna say while we're doing [25:39] Richard Webbe: the opening segment? I was just saying, I get embarrassed every time I get a drink 'cause I've got this straw, but in Indonesia I burnt my, sunburnt my lips so bad they're still all blistered and hurting. So- ... if it looks weird, I'm drinking from a straw after a, a shocking surfing holiday. [25:52] Richard Webbe: There [25:53] Marno Brits: you go. This is your world, mate. We're just living in it, so good for you. [25:55] Richard Webbe: Thank you, sir. That's very true. Yeah. [25:57] Dave Pengelley: No, no, no judgment. Um, [00:26:00] all right. [26:01] Richard Webbe: Let's- You win, Mano. Uh, okay. [26:03] Dave Pengelley: Um- So, so news. The news. Yeah. Uh, let, let's, let's talk about the AI news, and this one has popped up. We're all in the, in the consulting space. [26:14] Dave Pengelley: We're all trying to help businesses get along with AI. It's interesting to see headlines like this popping up, that Anthropic takes a shot at consulting industry in joint venture with Wall Street giants. Um, now I think, Sean, you shared this one with me. Um- What's the summary? What, what are you thinking around this whole joint venture thing that they're doing? [26:35] Dave Pengelley: Uh, it's, [26:36] Shawn Hain: uh, what is it? A $1.5 billion investment to create a consulting firm to basically put AI engineers and, um, capability inside bid market companies and up. [26:50] Richard Webbe: Yeah. So- They're trying to [26:50] Shawn Hain: cut [26:51] Richard Webbe: us out of the market, Sean ... [26:53] Shawn Hain: yeah, but it's, I think they're more competing with your traditional big four- Yeah, and stuff like that [27:00] Shawn Hain: consulting firms more directly. [27:00] Shawn Hain: So hopefully shouldn't affect u- us too much, depending on the size of, uh, your, your target market, I suppose. If you're mainly targeting small and medium businesses, they don't have enough money for this. Um, you know, they're not paying those kind of rates, unlike our government or universities. [27:17] Shawn Hain: But that's where we'll see this coming in, which will actually be really interesting. Um, 'cause AI short-circuits a lot of that traditional consulting cycle, or at least compresses a lot of it. So how are they going to... What are they going to deliver for that same cost then? Did, [27:34] Dave Pengelley: did PwC deliver a go- a report to the government for half a million dollars that they had to pay back because it was all full of hallucinations, and they just got ChatGPT to output the thing? [27:42] Shawn Hain: Yeah. Um- [27:43] Dave Pengelley: Well, I mean may- maybe that's part, part of what Bl- Anthropic's doing. And Anthropic obviously have skin in the game. We know, we discuss how models are becoming commoditized, how if you've got the right setup, you can swap, sub in. And I, and I do this actively. As I was saying, I hit my model limits with Codex on Monday, but I was still in Claude, [00:28:00] and they're running out of the same workspace with the same file folders. [28:03] Dave Pengelley: And so I can run parallel projects, and I can just switch harnesses almost seamlessly at this point. Um, they want to bake in that Anthropic is good because Anthropic has these specific things, and lock in their workflows through their own consulting magic, right? [28:19] Shawn Hain: Yep. I imagine they'll leverage all of, like, Anthropic's built-in tools. [28:24] Shawn Hain: So it'll be Cowork, Claude Code, and all of those- Cloud [28:29] Dave Pengelley: routines, all that kind of stuff- Yeah ... the m- the managed agents. [28:32] Shawn Hain: Yeah, [28:32] Richard Webbe: but if- So I, I was... Yeah, I, I agree. I was talking to a couple... Sorry, I didn't catch you off, Sean. I was talking to a couple of, um, medium to large businesses recently about their appetite for AI, what their strategy and their plan was, and, and I still talk about the three types of AI. [29:00] Richard Webbe: There's personal productivity AI. What do we say? 26% of people are using it. Those that are using it, it's like a panacea. "Oh my God, I can get my resume out t- three times quicker. I can [29:00] Richard Webbe: do my invoicing," all that sort of stuff. Um, and then you've got the next level of AI, which is the AI included in off-the-shelf vendor software, right? [29:10] Richard Webbe: And so a lot of the medium to large sized companies I talked to, I said, "What's your strategy with AI?" And they all go, "No, I don't need one because, you know, SAP or Salesforce or, they've got AI in their thing, so I'm done, mate." Right? And that, let, let's hold on that thought again. And then you've got the third smaller group, which is the group that kind of needs us, right? [29:29] Richard Webbe: Where we come in and say, "Let's look at your business," and as David and I would say, and Mano would say, um, "What's the question?" So what's the challenge of your business? And we will build you something, and it's just for you, so you'll get competitive advantage. [29:43] Marno Brits: Hmm. [29:43] Richard Webbe: But we've got this larger group, who when someone tacks on AI on the back of their existing software, and it is agentic and they do have some stuff in there, but all their competitors get the same thing, so there's no advantage. [30:00] Richard Webbe: And they're missing what I think the biggest point [30:00] Richard Webbe: is, your knowledge about your business. If you give people like us, and not just big consulting firms that, that, you know, um, uh, you know, that people like Anthropic wanna control, so it's under their world, right? [30:12] Marno Brits: Mm. [30:13] Richard Webbe: You need people like us- Yeah ... to turn up and go, "Okay, you have a unique view of your business. [30:17] Richard Webbe: You got here for a reason. Forget all the automation. Now let's get that and accelerate your competitive advantage." And I just think it's a mistaken conversation in the market that we just can't get to the market quick enough. You need people like us to look at your business and find areas where we will make you more successful than anyone else, because it's based on your experience, your knowledge and your questions. [30:42] Shawn Hain: Le- leveraging your experience and competitive advantage harder than what you, than what the default Salesforce tool, for example, can- Exactly ... can do, 'cause it doesn't, it's built to work within Salesforce's systems, not yours. Exactly. Salesforce is a system you use. [30:58] Richard Webbe: Exactly. Exactly. It's, [00:31:00] it's... My father used to... [31:01] Richard Webbe: I've, I've mentioned this before. My father said he was gonna become a millionaire 'cause he went out and bought a fax machine, this is obviously a long time ago, and he was ch- You can laugh, Majoli. Exactly. And he was charging people 50 cents to come and send their faxes 'cause they didn't have to go to the post office. [31:16] Richard Webbe: And I said, "Dad, how long do you think that's gonna last?" He goes, "Oh, it's gonna be great." I said, "No, Dad. Fax machines are so cheap, everyone's gonna go out and buy one." But to his credit, he managed to pay for his fax machine, but right about when he got the return on his investment, exactly, everyone else had a fax machine, so no one came anymore. [31:33] Richard Webbe: It's the same thing, right? It's, it's this concentric circle or leapfrog thing. If you are running a business, and you have secret sauce, which everyone does, every one of us, everyone in business, that secret sauce, use AI to protect it and accelerate it. Yeah. And if you don't, you're gonna get eaten up. [31:53] Dave Pengelley: The, one, one of the issues, I think, is small-medium businesses, they are both time and cashflow poor. [32:00] Dave Pengelley: Or they've got cashflow, [32:00] Dave Pengelley: but they're not actually... The, the margins are thin, uh, they've gotta make payroll, et cetera. They don't have the, um- the free space to think about making these sorts of transformations and changes. Whereas the bigger town, they, you know, they've got departments they can set up and set up a new chief transformation officer. [32:16] Dave Pengelley: They can do that. They can pay for the, the, um, the big consulting firms to come in and do all that sort of stuff, and it's always been true. Like, you'd mentioned Salesforce, I used to work there, but they were a premium high-ticket product that not everyone would afford. And yes, they do a lot of selling to SMBs, but that's kinda not their core, core thing because they're very expensive. [33:00] Dave Pengelley: Um, but that means a lot of small businesses don't bother putting in any kind of CRM, and they can't, they can't have that knowledge base and that, that, that, that sort of platform of, of knowledge to operate and to grow and to market against. And I think AI's gonna be similar. We're gonna see this have and have-nots, and we're gonna see this division where there's gonna be a bunch of companies that could [33:00] Dave Pengelley: really leverage and get the most out of it, but they just don't have the time, energy, or money to invest. [33:04] Dave Pengelley: And I, I know another guy I know, the consultant, not in the AI space, he, um, just does business consulting, and he helps these firms that have good cash flow and revenue, but there's no profits because it's all being sucked up into operational things, help them... Like, they pay him to come in and actually help turn the business around, actually find capital growth, um, because a lot of businesses have that issue. [33:28] Dave Pengelley: Um, so there's, there's a few things going on there, and I think how do we get in and how do we make our offer so compelling that they, it's impossible to say no, where they are better off and we get to, you know, grow our businesses and make money as well and make it a win-win, is the challenge of 2026. [33:45] Richard Webbe: It is. [34:00] Richard Webbe: It's, it's the technology leverage. So most technologies in recent years have supported the SMB player to play as big as the big player, but it doesn't take long before the big player comes along and neutralizes that. So what are examples of that? [34:00] Richard Webbe: Mobile phones allowed tradies and individuals to become as big as a big player. [34:05] Richard Webbe: They didn't need a receptionist to answer the phone. The call came to them. Fax machines did the same thing, and as, as Dave pointed out, CRMs in certain ways did the same thing. And the big one was webpages, right? How do I become an important business that I can advertise and people can find me anywhere? [34:24] Richard Webbe: Uh, when the Yellow Pages and all of that was becoming so competitively expensive, people would set up their own webpage for almost nothing. Yes. And all of a sudden, they had a global presence, right? And that had to be neutralized and changed, and we of course we had search engine optimization. So it's this leapfrogging game, and that's why I say get on the front foot, know your business, and then call people with the technology in to help them map the, the technology to your business. [34:50] Richard Webbe: And I, I, I think you hit the nail on the head there, Dave. Well- It is a leapfrog changing world. Have we got something else coming up? [34:57] Dave Pengelley: Yeah. So, um, John [00:35:00] Roese from Dell, uh, is quoted saying, "Sovereign and agentic AI will shape tech in 2026." Uh, the Dell CTO. So- Yeah ... yeah, so then this is what the big, big firms are l- looking at and reading. [35:12] Dave Pengelley: And this was back in December, this headline. My, my AI pulled this out for me as if it's fresh. Thank you. I've gotta work on my search algorithm. Um, but, you know, this is, this is still true. This is the kind of thing, and we're seeing more of this. Back in December, no one knew what agentic AI was. People were still looking for the magic prompt to put in. [35:30] Dave Pengelley: Um, but this is definitely the direction we've seen things going in, and we talk about the whole sovereign piece with managing your own models and having your own hardware and, and data and trust. So this is the direction, but again, the small/medium players don't have the, the scope or bandwidth to think about that. [35:47] Richard Webbe: No. Yeah. So it's scale, it's scale versus agility. The big guys have got scale and resources, and the little guys and little girls have agility. And, and that's what you focus on. [35:58] Shawn Hain: Um, one question for [00:36:00] you all. Um, something I've seen come up in a few conversations lately is a lot of concerns around whatever solution we develop today will become redundant next week 'cause of, or next month- Definitely [36:10] Shawn Hain: when the AI adv- advances. Have you guys been running into that as well? And, like, how are you communicating around that with your clients? 'Cause you can't exactly say, "Well, n- it's not gonna happen," but the flip side is for their specific use case it's almost unlikely. [36:26] Dave Pengelley: It comes back to solving the problem and not going in with the tool first. [36:30] Dave Pengelley: And you need this tool because this tool has all these bells and whistles. A Swiss Army knife, it's got 20 things on it, and then- Yeah ... the next week something comes out with 25 things on and you go, "Oh, I should have got the thing with the 25 things." It's, no, no, what do you need? You need to be able to open a can of beans. [37:00] Dave Pengelley: You need to be able to saw a piece of wood. Uh, you need these three things, and they're the things that drive value for you. So if you're implementing the solution because it solves a problem and drives value, it doesn't matter if version, like, half a point better comes out next week because [37:00] Dave Pengelley: you solved a problem for them six, th- a month ago, six weeks ago, and they're already seeing value from it. [37:05] Dave Pengelley: Mm. And so those values compound, and then, yeah, you can, you know, iterate, add features over time, but that's the way I would, I would [37:12] Richard Webbe: think about it. I, I just, I think it's the, the right question, and Dave's hit the nail on the head again. It's the, I, I used to call it the Intel chip envy. Remember people would say, "My PC's got the latest- Yeah [37:22] Richard Webbe: Intel chip, so I'm a better- Yeah ... person than you." And it's just stupid, because you're doing the same thing. Uh, is your Excel writing, running any faster? Can you type in any quicker? And exactly like say, Dave, if, as we say, if you focus on the question. So I had someone say exactly to me, Sean, what you said. [37:38] Richard Webbe: Yeah. "Oh, why should I go down the AI path? Something else will come along." I said, "No, I think you're missing the point, right?" I, you know, if, if, if mobile phones came out tomorrow and you said, "I'm not gonna buy one 'cause there'll be a new one-" down the road, I, I'd say you're missing the point. Mm-hmm. Get in and use it. [38:00] Richard Webbe: I had, uh, I worked on the project to install the, uh, smart metering across, uh, Victoria, right? This [38:00] Richard Webbe: was a billion-dollar project. I set it up, and I was head of the company in Australia that did that. And, um, one of the heads of one of the utilities said, 'cause we were integrating the chips inside the, uh, s- meters. [38:13] Richard Webbe: Now, the last time your electric meter got changed was after the First World War. Most of them were Bakelite, right? Not plastic, right? So it was bizarre, right? And I remember, this is a C level, and I won't say his name, he's a great guy, but his view was all wrong. He said, "Oh, Richard, why should we do this technology? [38:30] Richard Webbe: 'Cause something else will come out next week." I said, "Mate, it's every hundred years that we change the power meter on your board. If you don't get in with something now, you're gonna be left behind, and all the other distributors and suppliers are gonna slaughter you." And, and, and this is what I think the answer is, Sean, and I know you asked the question 'cause you knew the answer anyway, but get into the game, right? [39:00] Richard Webbe: Mm. Because the real advantage isn't Intel chip version tw- 2050 or whatever, it's my [39:00] Richard Webbe: experience, as we were saying before. It's my knowledge. Get into the game. Get a skateboard, get a mobile phone, get a car. [39:06] Dave Pengelley: We, we were discussing this last week with Mano. 'Cause Mano, you've been going through different iterations of agents and solutions, and you started with something, and you've evolved it, like, five times since then. [39:15] Dave Pengelley: Mm-hmm. Um, because you are looking for the thing that actually does solve the problem that you've got, and you haven't quite landed there yet. So from, for you, I think, right, like, you're, you're not, not doing something 'cause the next agent's gonna come out. Like, you didn't know Hom- about Homie six weeks ago, but you were building something yourself, and you've iterated b- to try and solve the problem, right? [39:35] Marno Brits: Yeah, I think it's getting very clear on what... Yeah, exactly what problem are we solving for, and having measurements around that. Not, "Hey, I need to make more money, so let me put an AI," or, "Hey, I need more leads. Let me do a lead gen automation." It's, what is going to move the needle the most, and what exactly do I need? [40:00] Marno Brits: So the shift for me was instead of constantly [40:00] Marno Brits: vibe coding, instead of going in... And by vibe coding, I mean literally just opening up the terminal and say, "Hey, help me fix my business." Like, instead of just trying to keep smashing it, expecting it to do the work for me- I, if you really want to make money, found it. [40:15] Marno Brits: Did it with the AI company. Um, it's, it's very much for me, stepping away from the building completely, spending hours upon hours of planning, like questioning myself rigorously of what am I solving for? What this is going to look like in a week, in a month, in five years' time, and solving for that problem. [40:38] Marno Brits: And now that I've built something, like I spent 80% of the time planning, then I built my, um, product requirements document, then gave that to the AI, and then built it. And now, within a couple weeks, I've achieved more than I have in the last six months. [40:52] Richard Webbe: Exactly. [40:52] Marno Brits: Because I took two weeks of doing absolutely nothing but planning. [41:00] Marno Brits: Like, it wasn't fun. Yeah. It was the, the work needs [41:00] Marno Brits: doing. Yeah. And it works more on you than you work on it. So it just... That was constantly putting in the planning. Now I have something that's concrete, and regardless if OpenClaw or whatever updates come out, it doesn't matter, 'cause the problem for me is solved. [41:12] Marno Brits: Unless there's more than a 20% increase in whatever my measurements are, there's no point in me switching because the- there's diminished returns. [41:22] Dave Pengelley: Yeah. That- that's a great lead in for this, I think. [41:26] Marno Brits: I guess. [41:39] Dave Pengelley: Wins of the week, the week. Uh, it's actually, you were talking about vibe coding then, uh, just, just a little funny clip that I saw on X this week around vibe coding. Uh, here we go. Vibe coders debugging an app they built with Claude Code. [41:58] Shawn Hain: Saw [41:59] AI VO: [00:42:00] that. [42:03] AI VO: Oh. Oh, I love [42:05] Dave Pengelley: it. [42:11] AI VO: Damn it, come on. Stay up there. You gotta yank it to the side, Peter. I'm yanking it as far as it goes. You gotta catch it on the thing. Oh, that's helpful. The left side's getting lower than the right side. I can see what's happening. Now twist it. All right. See? Now it's open. [42:27] Richard Webbe: Accurate. Yeah. I'm not sure what the script was copyrighted, but that's hilarious. [42:34] Dave Pengelley: Very, very, very, very fun. Um, we're not monetized on YouTube yet, so I think we can get away with, uh- [42:39] Richard Webbe: Of course. Of course ... [42:40] Dave Pengelley: playing 30 seconds of Family Guy. Um, it feels like that's about- But vibe, [42:44] Richard Webbe: vibe coding is, you know, vibe coding is a great way to see if you can solve a problem. Here's the funny thing. I can vibe code as quick as some massive conglomerate, but I can't really always provide the support or the debugging or backup. [43:00] Richard Webbe: This- [43:00] Richard Webbe: So if I'm doing it internally and I've got a good- team, maybe someone like the team I'm looking at now, then you won't have a problem because they know where they're headed. I think the less, uh, educated AI people wanting to vibe code may find themselves in a bit of a problem, and I heard of someone vibe coding their own agentic, you know, setup so they could pay all their bills. [43:21] Richard Webbe: And s- and, and I think the message was one of their agents emailed all their passwords and everything to all their banks and everyone else. [43:27] Dave Pengelley: Ah. [43:28] Richard Webbe: It, uh- Oh ... that was a bit outrageous, wasn't [43:31] Marno Brits: it? [43:31] Dave Pengelley: Yeah. That's a good point you make about, about the, the, uh, the ongoing support, uh, and maintenance and stuff, right? [43:36] Dave Pengelley: Like a mate of mine said to me, he's in a bit of a niche industry that has some software that's not... He doesn't love it. Uh, and he's a bit of a, bit of a tinkerer, and he's very creative, very clever guy. And so he's been building some of his own little sort of edge solutions, and he's like, "Do you reckon there's a business day for you g- um, in building a competitor in this industry?" [44:00] Dave Pengelley: And I looked at him and went, "Maybe." And then as I looked into it more, I went, [44:00] Dave Pengelley: "No." The compet- like the, the who you're up against is not only people with hundreds of developers, but just some of the problems are tricky. It's all personal, like i- interrelationship management stuff that's just innately tricky to actually data model and get right in the first place. [44:14] Dave Pengelley: And I was like, you know what? That is a whole thing that you could vibe code. And, and so he, he might build his own bespoke thing for his b- his, for what he does, and that's fine doing that for you and your specific requirements and what you need, but then productizing that into something that scales is a whole different ballgame. [44:33] Shawn Hain: Mm. I had this conversation with a client the other day. They wanna do, they wanna custom-build a CRM. Mm. I'm just like- [44:40] Dave Pengelley: Why? ... [44:41] Shawn Hain: why? Yeah. Like there's open source, off-the-shelf solutions that have all of that basic CRM relationship modeling built in, tested out the wazoo with 50 developers behind it. I can build it, but it's gonna take me- Yeah [44:55] Shawn Hain: six months to get to parity, and theirs is still gonna be better. [44:58] Dave Pengelley: Yeah, and I, I- Because [44:59] Shawn Hain: they're more into [00:45:00] more edge [45:00] AI VO: cases. [45:01] Dave Pengelley: I went down that journey right when I first started doing this, this business thing. I, I started with HubSpot Free and went, "Ugh, it's too heavy. It's got too much old fat that I don't need for marketing stuff." [45:10] Dave Pengelley: So then I was like, "I'm gonna build my own in Notion." In Notion, I hit limits. I went, "No, I'm gonna build my own in Airtable." I started building Airtable and that. Eventually I was like, just simple stuff. I think when we started working together, Richard, I started realizing just the basic email capture, just core features you expect, calendar and email sync, it's just so hard to build yourself. [45:27] Dave Pengelley: So then I went and got 20CRM, the open source one, and started experimenting with that, and that was good, but still just, you know, the self-hosting and the this and the that, and I think... So just getting something off the shelf, like it just is makes sense. It's why Anthropic and OpenAI and all these other businesses, they still buy software from people like Salesforce and Miro, et cetera, et cetera, Notion. [45:51] Dave Pengelley: They don't just rebuild those solutions internally because it doesn't make sense. [45:57] Richard Webbe: Sean, the key point about their question, I think... [00:46:00] Sorry, mate. Just quickly, is- When they say, "Build me a CRM," I think what they're really saying is, 'cause CRM's not gonna make them money, is it? No. What it does is... I think what they're asking you is, "I have a view of what I want my CRM interface data capture," and what Dave was just talking about, "will look like." [46:17] Richard Webbe: And of course, you don't need to build a new CRM like that. We could build an agentic agent on the edge of that- Yep ... that will automatically capture all of that, uh, and go through their emails and listen to their conversations- Perfect ... and, and log it all in. [46:29] Shawn Hain: Yeah. Where I landed was 20 CRM with a custom wrapper around it to give them the special sauce they wanted without having to rebuild all the CRM bit. [46:40] Richard Webbe: Yeah. [46:40] Shawn Hain: Um, it was where I land on it, but it was just a perfect example of that. [46:45] Richard Webbe: Yeah. [46:46] Shawn Hain: You don't need to reinvent the wheel, despite it being easier now. [46:50] Dave Pengelley: Mm-hmm. [46:51] Shawn Hain: Write coding a CRM in one day, yeah, you can do it. But the off-the-shelf solutions are going to be better every single time. [46:59] Dave Pengelley: Yes. So we, [00:47:00] I, I wanna today, um, focus on things like Marino's Hermes Agent and, and so on, and how we're leveraging some AI in our various businesses. [47:08] Dave Pengelley: I mean, a few of us are, are heading more down the direct consulting line. Manjula, you've built a product with your Smart Syllabus system. Um, and you've obviously done a lot of, uh, agentic engineering to build that product. But then also, how are you just going to market with that and, and some of the tools you're using in your business to help drive the growth of... [47:29] Dave Pengelley: A little bit of a ring around order about how are you using some of the AI tools on your own internal process and operations to give you those lift, lift we can talk about giving to customers? [47:43] Manjula Kerai: So I'm using, um, Cloud Code, of course, to help me with all the admin around trying to sell this. [47:49] Dave Pengelley: Mm. [47:50] Manjula Kerai: And, um, I'm speaking to lots of teachers and principals. [48:00] Manjula Kerai: I even went to meet the MP, uh, Mark Hodges last week. And so- Nice ... I'm basically [48:00] Manjula Kerai: speaking to people to see where I can find the right market for this. And, and a lot of it is actually, um, speaking to the teachers, the, um, the principals, getting into even the diocese, speaking to them. So right at, at this moment, it's really getting the validation, is this, um, gonna be, uh, approved for schools? [48:23] Manjula Kerai: So I'm contact- I'm, I'm trying to get into the Department of Education as well to kind of get the validation. But a lot of teachers are using AI as it is anyways, and, um, by them, um, like, having this, like, on an individual basis, that's gonna be good as well. Um, I'd like to get it into the hands of all the teachers, um, so yeah. [49:00] Manjula Kerai: Uh, AI has been- For me to get it to market is, AI has been great, 'cause what I've been getting it to do is set up my calendar. Uh, it gives me a script, you know. Uh, if it goes to voicemail, it gives me a script for that. Text message, it's all [49:00] Manjula Kerai: doing it. It's, it's actually sending iMessage, iMessages out as well through my C- Mac. [49:06] Manjula Kerai: So- Nice ... yeah, it's been amazing. Uh, and I, and I feel like as a solopreneur, the things that I've got this AI to do for me, it's, it's just unbelievable. And, and I feel like the bus- businesses that we speak to, they don't even realize the potential of AI, and sometimes they say to me that, you know, "Oh, we're already using AI," but actually they're just using like an AI assistant in their current software, you know? [49:31] Manjula Kerai: Mm. They don't realize the, the- [49:33] AI VO: No, they don't ... [49:34] Manjula Kerai: the ability of AI at the moment. And as solo- solopreneurs, we've probably done so much, uh, we're so mu- so deep into AI and that we can see the possibilities, and we're always asking, you know, "What if it can do this? What if it could do this?" So every time I have a problem while I'm working, I'm like, "Let me ask my AI first," and you'd be surprised what it can actually do. [50:00] Manjula Kerai: Obviously, you have to give it the connections and stuff, but yeah, it's been, it's been [50:00] Manjula Kerai: brilliant for me to get this, uh, software out to the market. [50:03] Dave Pengelley: Yeah. Nice. Yeah, scripts and- Programs and build it ... au- au- auto messaging, that, that, that's wild. [50:07] Richard Webbe: Mm-hmm. Mm. That's awesome. [50:10] Dave Pengelley: Yeah. Um, Sean, Sean, what are, what are you doing as far... [50:13] Dave Pengelley: I, I'm, I'm... It's in no particular order, but I'm just calling names out now. Yeah. [50:16] Shawn Hain: Um- That's all right. Um, so what I've been working on, I've actually been working on my internal, um, software engineering loop. So I'm actually building out the system so that I can automate some of my actual code writing without compromising on the standards that I rely on. [50:34] Shawn Hain: Um, so as Mano mentioned before, it's all about planning. I probably spend 80% of my time in a project planning, making d- all of the decisions up front. How is this meant to work? How do I maintain it? All of that, so that then the AI can just come along and follow the plan that I've built- [50:51] AI VO: Mm ... [50:51] Shawn Hain: rather than me live coding and making it up as it goes along. [51:00] Shawn Hain: So I'm creating like the database structure and stuff to make that reliable and [51:00] Shawn Hain: observable. Um, in terms of how that would differ to, say, like using ChatGPT or Claude Code, you say, "Claude Code, build me a user interface." Well, every time it does that on your project, it's gonna look different And the question I ask myself is, within a project across 30 different sessions across three weeks, how do I make every UI that it builds this application look the same without me having to tell it at the beginning of each session what the UI looks like? [51:27] Shawn Hain: Yeah. So [51:27] AI VO: it's gathering examples, [51:29] Shawn Hain: document what that means in human re- understandable English, but also applying restrictions so that it can't create... Like, it can't do a, um, two-tone element, for example. If you don't want that in the user interface, make it impossible for it to do it, so when it tries, it errors. [51:47] Richard Webbe: Mm. [51:47] Shawn Hain: Um, we're spending a lot of time looking [51:52] Richard Webbe: how do you make that reliable? It's, it's a really good question, Sean. I like, I like... You know, because as you know, because we are all [00:52:00] programming AI with our questions and our answers, like you said, what is acceptable or usable today will be completely different tomorrow because of the billions of questions and calls that are going in every few seconds. [52:13] Richard Webbe: It's gonna keep evolving. And you do... I've, I've done a couple of things like that where I- I've asked it to do something, and then I've had to give it a reference point and say, "Don't go beyond this and keep it in line with this narrative or this scale, otherwise I'm gonna have a problem." [52:30] Marno Brits: 100%. Yeah. Very good. Dave, any- anything specific from you? I know, like, we are very similar and, like, we have our little agents, and then we use it to decrypt old, um, dictation software as well as reading books or whatever else, so... [52:44] Dave Pengelley: Uh, for, for me, um, win- wins of the week from, from, from how I'm using it is, uh, I've been using Linear as this task glue across different things. [53:00] Dave Pengelley: So I'm, I'm running multiple different PCs. Uh, we're cross Claude and Codex [53:00] Dave Pengelley: and different workspaces with different sets of agents in them. So you know, I've got my Peris and my Felicitys and my Lawresses et cetera, but they're slightly different. Like the one that's building Chart Reporter, which is my SaaS app, is different from the one that's my general business operations. [53:13] Dave Pengelley: But because I do all my general business operations and planning in that one, it needs to know that Chart Reporter is one of my business operations, and so I have this cross-pollination between the two. And so using a tool like Linear for task management meant... After last week, I talked about story brand and using the, the story brand principles. [53:30] Dave Pengelley: So a few little wins on that is I got it to re-storybrand the Syllogism website that Rich and I are building, and it went through and completely sort of made all the language really simple, got rid of vague questions like, "You need to ask the question." I was like, "No, no, AI without payback is just another bill." [54:00] Dave Pengelley: I love that. Um, so it did all this really great work, but I was like, you know what? I would really like for my Chart Reporter site to get that same treatment And so, but I didn't wanna go and have to, like, teach another set of agents all about the [54:00] Dave Pengelley: Storybrand stuff. So in my business operations thing, I created a Linear task saying, "Go and take the Storybrand and go and do a review of the Chart Reporter website." [54:10] Dave Pengelley: And so then it went and issued us little agents off to Kat and Lucy, et cetera, and they went and did their research and, and gave recommendations, put that all back in Linear, and then I went over into my Chart Reporter workspace and said, "Cool, now you go look at that issue and see all the feedback and update the marketing site." [55:00] Dave Pengelley: And so that whole cross-pollination of agents across as business entities, uh, was, was pretty cool. Um, and so, and, and then just because they understand the context, and it comes back to knowing that context of what's relevant. On your point, Sean, you talked about getting the brand right. Um- Mm-hmm ... for my app, I sort of went with this sort of HUD look and feel when I built my app, and in the last 24 hours I've just built, like, a new sharing component where it creates little shareable cards for your socials, and it knew to output the sharing card, [55:00] Dave Pengelley: like- Nice [55:00] Dave Pengelley: on brand like the rest of the app, because I've- Absolutely ... defined all of the brand settings for both light and dark mode, and it knows what the colors are. It knows to use the little boxes and everything, and because I've established this is the brand design guidelines for this application- Mm-hmm ... when I created a new feature like this, it just got it right. [55:17] Dave Pengelley: I didn't have to tell it what the, what the look and feel was. It just knew. [55:21] Marno Brits: Yeah. [55:21] Shawn Hain: Yeah. [55:23] Marno Brits: I'm looking- So- ... forward to most to implement that, like, after you talked about that last week, Dave, is setting up your Linear system and then you will just create a ticket in there and, and it does a heartbeat check every hour to then go off into work. [55:35] Marno Brits: I'm working on getting the same system for Microsoft To Do, because that's where I track all my stuff. It's a simple view. [55:41] Dave Pengelley: Mm-hmm. [55:41] Marno Brits: So I'm like, I want it to do the same functionality of if I wanted to give it a task, instead of having to open up Discord and explain it, I would just put it in there and then it'll do the, the prep work essentially. [55:52] Dave Pengelley: Yeah. [55:52] Marno Brits: Come back, ask me questions for clarification, and then do the rest of it. [55:56] Dave Pengelley: Yeah, so, [55:56] Marno Brits: so my- The thinking partner and the workhorse ... [55:58] Dave Pengelley: my, my rhythms, my, I've got a dawn and a [00:56:00] dusk rhythm where it does a full operations review at the start of the day, at the end of the day, looks through what's been happening and sends me emails, and then through the day I've got every hour it's got a heartbeat mechanism that I, I stole the idea from OpenClaw for, where it goes and looks at Linear and goes and checks all my comments, sees if I've given it any feedback. [56:16] Dave Pengelley: Um, it, if anything's sort of getting a bit urgent, it will escalate and send me a pulse email out of sync and say, "Hey, you really need to be looking at this thing. This one's getting critical, Dave." And it'll hassle me. And then in the email it says, "You haven't responded to this for five days. Why is this draft not published? [56:31] Dave Pengelley: Why aren't you doing this? Uh, we haven't heard from you for 60 pulse ticks." And I'm like, "Oh, do you miss me? Are you, are you..." But, like, it is, like you asked at the start of the show, um, am I driving Claw or is it driving me? And I think in this instance, like- As a bit of a scattered, sort of, you know, undiagnosed... [56:48] Dave Pengelley: We've discussed the, uh, my undiagnosed ADHD. Having this occasionally go where I'm like, "What am, what am I gonna do next?" And just go back to my linear board- [56:56] Richard Webbe: Yeah, [56:56] Marno Brits: bro. [56:57] Dave Pengelley: Bye. See you, Mano. [56:57] Richard Webbe: Yeah. Thanks, Mano. [56:59] Dave Pengelley: Uh, go back, go [00:57:00] back to my linear board and look at, uh, the human tasks it's assigned me and go, "Okay, what are the top priorities? [57:05] Dave Pengelley: Look at the, look at the email it's given me. What do I really need to knock off now?" Click in, give it a comment, go back, give it a comment, go back, and then on the next hour the ticker reads and goes, "Oh, cool. We've heard from Dave. We've got comments. We know what to do next. We'll reassign that work over there," or, "He didn't like that the way we phrased that, so let's generate a new version of that document for him to post later." [57:23] Dave Pengelley: And, um, yeah. [57:24] Richard Webbe: You are building... You know what you're building? You're building wife v- wife version 3.0. No. Never [57:31] Dave Pengelley: change my wife. There's only, there's only one, one my, one my wife. Um- [57:36] Richard Webbe: It's insecure and nagging you. [57:38] Dave Pengelley: No. No, na- no, my wife doesn't nag me. My wife's amazing. [57:41] Richard Webbe: I, I wasn't referring- How dare you ... to your wife, Dave. [57:43] Richard Webbe: I apologize, and step back from that comment very quickly. [57:45] Dave Pengelley: How dare you. Um, but yeah, it's, it's... I, I, I think, and this is why I made that little video yesterday in my garage with a whiteboard, just because the way I'm leveraging AI with the context and with the, the agents with their job roles, it's [00:58:00] such a huge lift, and it, it's not- Yeah [58:03] Dave Pengelley: people aren't gonna start at my level with my level of agents and sophistication. I've built this over months. Um, but I think that's why on the Silagism website now, we're now pitching the, like, we'll help you get a starter pack of this up. If you're looking as a, as a small business owner to set these sorts of, uh, capabilities up for your team, we can help you do that. [58:22] Dave Pengelley: Like, it's not that hard at the end of the day. Um- [58:25] Richard Webbe: The funny thing is, it's a bit like when you, you get someone to look at the internet for the first time. I remember when I was working for Apple and they were launching eWorld and the internet was kinda just taking off, right? And eWorld, I think I mentioned this before, was Apple's version of what they wanted the internet to look like, 'cause it had navigatable places- Mm [59:00] Richard Webbe: not just hypertext links and all that sort of stuff. But I remember I, uh, before the internet was really properly public, I got access to accessing servers around the world with a command line capability, and the first time I was looking at a museum in South Africa or [59:00] Richard Webbe: a, a train schedule in New Y- it just blew me away, and I tried to explain to people how powerful this communications was. [59:08] Richard Webbe: None of them got it. And I remember, you know, several years later friends come up to me, "Oh, now I get what you were telling me." And AI's the same, right? [59:18] Dave Pengelley: Yeah. Yeah. [59:18] Richard Webbe: It is... Y- y- you've gotta have people immerse themselves in it and experience it to understand how powerful what you are doing. [59:27] Dave Pengelley: You, you do get that, and, and people know you're the AI person 'cause you talk about AI all the time, and they know that's what you do and that's, that's, it's your business. [59:33] Dave Pengelley: And then, you know, six weeks later they'll come up and go, "Oh, I got, I got ChatGPT to rewrite this email and this document for me, and it was amazing." Oh. Yeah. [59:46] Richard Webbe: Yep. [59:47] Dave Pengelley: I'm like, "Good, [59:48] Manjula Kerai: easy." And then, you know, I know, I knew about this, like, three years ago. [59:51] Richard Webbe: Yeah. [59:52] Dave Pengelley: Yeah. [59:53] Richard Webbe: God, yeah. [59:54] Dave Pengelley: Welcome, welcome to the party, pal, as, uh, as John McLean- That's it [01:00:00] Dave Pengelley: would say. Um, but I mean, [01:00:00] Dave Pengelley: all these things are evolving. Like, I was, like, on the trading thing, like, I've built the app now to help traders, but I was trying to build trading robots back in ChatGPT 18 months ago and, you know, it was bad. I would get the syntax wrong half the time. I'd be manually copying and pasting things out into my terminals to run tests and stuff. [01:00:17] Dave Pengelley: Now I'm at the point where I'm like, "I've got this idea for a strategy. Let's investigate that," and it will automatically build this, research it, build the strategy, it'll pull back data, it'll auto test, it'll actually define whether it's in or out. Um, and it's all automated pipelines now. It all runs the testing automated for me, and it's just, it's come so far, just my own pipelines, as the technology has evolved and my understanding of it has evolved over the last sort of 12, 18 months. [01:00:43] Dave Pengelley: It's just insane. And the productivity means, okay, cool, I've got an idea for a trading strategy. Go do that, and while it's doing that, like, I went for a walk, and I come back and do stuff, and I'm not there manually having to hand it the next thing all the time. Yeah. And that applies across any job and thing you're doing. [01:01:00] Dave Pengelley: I [01:01:00] Dave Pengelley: mean, OpenAI, Anthropic, they're all getting across this. 11 hours ago, Anthropic shared their... And I'm not trying to plug them specifically, it's just they were at the top of my headlines when I saw this one. That's an AI it's doing great things, too. But, uh, new for financial services, ready-run Claude agent templates for building pitches, conducting valuation reviews, closing the books at month-end, and more. [01:01:21] Dave Pengelley: Install them as plugins in Cowork and Claude Code, or use our cookbooks to run them in production as managed agents. So they're, they're looking at these job role things, like you, Manjula, with Smart Syllabus Assistant. They're finding the specific industries where they can get a big lift. Teachers spend a lot of time writing out their lesson plans. [01:01:38] Dave Pengelley: That's right. If they use your tool, they get a big lift, they get a lot of time back to focus on what they wanna focus on, not building lesson plans. [01:01:47] Manjula Kerai: Focus on being with the students, you know? [01:01:49] Dave Pengelley: Right. [01:01:50] Shawn Hain: Fo- focus on the human part, I think someone earlier... So that's the thing that's gonna be very hard. I mentioned [01:01:55] Manjula Kerai: it. [01:02:00] Manjula Kerai: Yeah. Yeah. You know, with the space so that they can use their creativity more in front of [01:02:00] Manjula Kerai: the students. [01:02:01] Shawn Hain: Yeah. Yeah. Hell, run more classes rather than spending eight hours a week in admin. [01:02:08] Dave Pengelley: Well, [01:02:08] Shawn Hain: we- I mean- Go on ... looking at every job, right? Like, even look at something like bookkeeping, right? That's a fairly... Most of what bookkeepers do is fairly mechanical, and AI is coming for it. Oh. But it's the non-mechanical part that they need to start leveraging. [01:02:22] AI VO: Yeah, [01:02:22] Shawn Hain: exactly ... if they wait, they're going to get eaten alive, and it applies to almost every industry. [01:02:26] Shawn Hain: Software development, take your pick. I read an article earlier, apparently the employment rate for 20 to 25-year-old software engineers is at the lowest point in years because all the jobs are disappearing because AI can do that junior dev role. You don't need a junior dev now, you need a senior developer- Mm [01:02:44] Shawn Hain: with a team of agents doing the same work. [01:02:46] Dave Pengelley: But if you never hire a junior dev, you'll never get [01:02:48] Shawn Hain: senior [01:02:48] Dave Pengelley: devs. [01:02:49] Shawn Hain: Oh, that, and that's 100% correct. But I'd actually argue now the pathway to becoming a senior dev is shorter. Yeah. The AI can t- you don't actually need to... You need to understand how to code, but you don't need [01:03:00] to spend 20 years writing code to become a high-end dev. [01:03:03] Shawn Hain: You need to understand, I think as Matt says, the primitives, the systems, and how it all connects. [01:03:09] Richard Webbe: Well, Sean- Well, Sean, Sean, you hit the nail on the head just before with bookkeepers, right? Yeah. When we hire accountants and bookkeepers, um, what 90% of their work they do is, is filling out forms for us, which is a real waste of space, and we've always known that, but we just, someone's gotta do it. [01:03:24] Richard Webbe: Now that we don't have to do that, 90% of the value that they give us is knowing what the rules are- Mm. Yes ... so we don't have to fill our head with those rules, and that's where they go, "Well, you know, if you pay this account first and that one later, and you move your money here," and that's what we're paying for, right? [01:04:00] Richard Webbe: And eventually that'll get eaten over by the AI, but then there'll be other rules or other ideas, and this is why I talk about my concentric ring, uh, sort of thing. You know, what's on the outside of the functional operations or rings of any company is the uniqueness. But once you know it, it goes into the next level of automation, and the next level [01:04:00] Richard Webbe: of automation, and then to a point where no one ever thinks about it, it just happens automatically. [01:04:05] Richard Webbe: Yeah. [01:04:05] Manjula Kerai: All evolving. [01:04:06] Dave Pengelley: Yeah. [01:04:07] Richard Webbe: Yeah. [01:04:07] Dave Pengelley: Yeah. And- And the, the space is moving fast and, and we're, we're doing our best to keep up with it, and they... We're, we're in the industry and it's so hard to keep up with some of these, these moves and, like, Mano's playing with Hermes agents and stuff. I took a half a look at that for, for, for 45 minutes and went, "I don't need this. [01:04:23] Dave Pengelley: What I'm doing in, in Cowork is enough." And the, the, the load to try and spin up a new thing to do what I'm already doing really well- Yes ... I don't, I don't need to go down that path. But there's some people where they're like, "No, I really want that. I want the newest thing. I want the thing everyone talking..." [01:04:37] Dave Pengelley: And so there are whole businesses s- spinning up to set up OpenClaus and Hermes specifically. [01:04:43] Richard Webbe: Um, I'm- So Dave, for a small business, let me ask you this question. [01:04:45] Dave Pengelley: Yeah. [01:04:46] Richard Webbe: I'm sitting here, I'm, I'm, I'm a solopreneur, uh, the same as Manjula referred to it, and, and I was one of those for many years and, and I grew it to 180 staff company, which was- That's not solo [01:05:00] Richard Webbe: performing and very well. Now, if I'm... Sorry, what'd you [01:05:00] Richard Webbe: say? [01:05:00] Dave Pengelley: 180 people is not solo anymore. [01:05:02] Richard Webbe: No, no, I started solo- ... and it went very well. But for the- Yeah ... the non-AI, the human part- Yes ... I needed execution. So- [01:05:09] Dave Pengelley: Yes, [01:05:10] Richard Webbe: got it ... that's, the core was m- and then I hired the 180 people. Yeah. [01:05:13] Dave Pengelley: Sorry, [01:05:13] Richard Webbe: the question? So the, the question is, uh, so I'm a solopreneur, I'm sitting here, and I go, "Dave- I need to be able to check all of my emails throughout my entire history- [01:05:25] AI VO: Mm [01:05:25] Richard Webbe: and I might have them saved somewhere like on Gmail. I need to check on all my data, which I might have on iCloud or Google somewhere or something like that, and I want an agent that I can ask a couple of questions of. You know, who are my best customers? You know, where's some of my best writing? A few things like that. [01:05:42] Richard Webbe: If someone came to you, how long would it take you to create an agentic environment running on their laptop with their, with their, uh, subscription to sort of move ahead on very basic solopreneur business function like that? [01:05:57] Dave Pengelley: You could have the bones of that [01:06:00] in a couple of days. [01:06:01] Richard Webbe: Right. [01:06:01] Dave Pengelley: Like, within, within a week. [01:06:03] Dave Pengelley: Yeah. Like, you gotta go in and understand exactly what problem you're solving and which systems they really need connection to. Um, but again, and then from a tool point of view, what are they already paying for? What do they like? Do they like Claude? Are they using Claude? Do they like OpenAI? Are they using Chat? [01:06:15] Dave Pengelley: Like- [01:06:16] Richard Webbe: Well, I would let you decide that [01:06:17] Dave Pengelley: for them. Well, no, no, yeah, and, and so you work with them. And, and because these agent platforms, like what w- what, what Syllogism i- is presenting as, as an offer, which is a s- slimmed down version of what I'm running with my daily, daily rhythms and my heartbeats, et cetera, it's like modeled on what I know works. [01:06:33] Dave Pengelley: That could plug into Codex, that could plug into Claude, it could plug into anything. Where are they at and what are they using? And I would probably, for a lot of these small businesses, wouldn't be going straight down the open Claude path because there's so much extra admin overhead with that. You'd keep with something that maybe you do need to have running on your desktop, and you keep a computer open in the corner 24/7 if you really wanna be doing that. [01:06:53] Dave Pengelley: But keep with these commercial products that run on your desktop a little bit more [01:06:57] Richard Webbe: manageably. So you could, you could increase my [01:07:00] productivity and my knowledge of my market and my business, even as a, a solopreneur consultant, uh, to get a 10X performance improvement within a week? [01:07:09] Dave Pengelley: Yeah. A- and remember, all this stuff comes back to the human. [01:07:13] Dave Pengelley: And like I said, my linear... I, I get hassled with my tasks all the time. If I don't follow up, if I don't actually then choose to go and post that blog post, if I don't choose to do the thing or give it feedback, it just sits there. So like any of these things, the AI is not magic. It is gonna help give you ideas. [01:07:28] Dave Pengelley: It's gonna help generate things. At the end of the day, it still comes down to you on, on the execution to a certain extent. Unless you wanna be like the guy whose car is driving him to buy some magnesium, which is just next level. [01:07:39] Richard Webbe: That is next level. [01:07:40] Dave Pengelley: Yeah. So do not recommend. [01:07:41] Richard Webbe: And so basically- ... so basically I would hand you the keys and say... [01:08:00] Richard Webbe: I mean, you and I both know this, but I'm asking you openly so people can hear, you'll get things started. And then I'm assuming from what you and I have found, Dave, and I'm sure Manjula and Sean would say the same thing, once that's been, like my concentric circle story, [01:08:00] Richard Webbe: automated, now my brain is free. [01:08:03] Manjula Kerai: Mm. [01:08:03] Richard Webbe: I'm gonna find some more questions to ask. Gonna find some more Venn diagrams to loop around the data to give me answers to improve my business. That's a fair call, isn't it? [01:08:13] Dave Pengelley: Well, yeah. And as you grow this thing, you'll be like, "I really wish I could do X with it." And so then you learn. And, and so I, you, I actually don't want the keys. [01:08:19] Dave Pengelley: Don't give me your keys. we, the part of the process is sitting down and building it together. It has to be co-built. It has to be you have to have some ownership and learn some of the underlying with it, which is not to like put a burden on a business owner, but it's actually to empower them to actually- Yeah [01:08:32] Dave Pengelley: get the most of out, out of it and be able to evolve it and grow it. [01:08:35] Richard Webbe: I'm good [01:08:35] Dave Pengelley: with- And I don't, I don't want your keys. I don't want the security risk- No ... of having your keys. Don't- No ... I can't set it up on my computer. It needs to be on your hardware, on your systems. So this is- Yeah ... not a, "I'm configuring a piece of software in the cloud and handing you a login." [01:08:48] Dave Pengelley: Like this is all your personal logins and on your computer. This is a co-built kind of solution. [01:08:52] Richard Webbe: Yeah. Well, I'm coming to Sydney next week so you can automate my SMS with my girlfriend- ... so she feels loved and I've got more time. [01:09:00] [01:09:01] Dave Pengelley: I'm sure Dave could [01:09:02] Manjula Kerai: do that for you online. [01:09:04] Dave Pengelley: Yeah. Well, you don't even understand. [01:09:05] Dave Pengelley: I, I, I do not recommend. I, uh- [01:09:08] Shawn Hain: Yeah, no. As on, on your, um, question, the one thing I'd say for some small businesses is, um, the fragmented data- Yeah ... that they've got. That would actually be, from like my perspective, the hardest part to overcome for older businesses- 100% ... is if you want all of your history. One, is it actually relevant? [01:09:27] Shawn Hain: Do you actually care what happened 12 months ago, 18 months ago? Some businesses, yes, others, maybe not. But, um, the hardest part's collecting all of that into the one place so those agents know about it and can access it efficiently. [01:09:43] Richard Webbe: Mm. It's a very good point because, uh, if I looked at some... A friend of mine who's got a business, he's got 144,000 emails over the last 20 years. [01:10:00] Richard Webbe: Gosh. But he considers, he considers all of those very relevant to [01:10:00] Richard Webbe: his consulting business and the way it's evolved, and he wants to mine that data rather than just with his random brain, uh, doing it. And, um, and that would be one of the... I, I agree. Data access is what Snowflake and, uh, Databricks- [01:10:16] Dave Pengelley: Mm ... [01:10:16] Richard Webbe: uh, and, uh, um, uh, people like, uh, NetApp and the storage companies and Pure Storage are all evolving to these data streaming will, will s- you know, will, will Well, [01:10:26] Dave Pengelley: RAG, RAG, RAG databases and stuff. [01:10:28] Dave Pengelley: Like in that instance- That's it ... you're not gonna be interrogating Gmail directly. You need to export them into another data source- Yeah ... that is vectored and, and searchable more easily by the AI. [01:10:39] Shawn Hain: Yeah. What I've heard that called is kind of a knowledge layer. It sits over your data layer, which is Gmail, your database, your CRM, then have a layer on top of that where all of your inf- all of that information gets fed into, cleaned, deduplicated, and then put into a format the AI can reason with, I guess. [01:10:57] Richard Webbe: Yeah. So we create those for our clients, but are there [01:11:00] any other pre, pre, um, uh, formatted agents that do this sort of stuff, or it's all still custom at the moment, isn't it? [01:11:09] Shawn Hain: Um, I mean, I'll have a look. [01:11:11] Dave Pengelley: Uh, uh, we're, we're talking, you know, the, the old PACE layering where you've got your systems of, uh, record and your systems of engagement and your syst- I can't remember all the layers in, in the PACE layering, but w- we're talking a few different layers here, right? [01:11:25] Dave Pengelley: So the agents are gonna be your, your sort of maybe your, um, layer of engagement and your layer of innovation or whatever, but they're not nec- they're not the, the system of record. Your agent's never the system of record. Yeah. And so you actually need to work out what that system of record is. Now, agents might be able to help you build the system of record- Yes [01:11:43] Dave Pengelley: but they will not be the system of record, right? [01:11:45] Richard Webbe: Yeah, I get [01:11:46] Dave Pengelley: it. And so they might help you set up the pipelines to get the emails out and vector them and put them into a database where they can be searchable, but then layer in the metadata so when you find a fragment that's useful, it can actually find back to the [01:12:00] source material to give you the full original file, not just the searchable fragment. [01:12:03] Dave Pengelley: And like all these, this is a whole data science literally- Yeah ... um, that needs to be considered with a lot large data stores and big archives and so on. So- [01:12:14] Richard Webbe: Yeah ... [01:12:14] Dave Pengelley: um, yeah. This is, this is why you start with the problem, not the tool. Yeah. You go in and find out- Oh, absolutely. [01:12:19] Richard Webbe: Yeah, yeah, yeah. That's [01:12:19] Dave Pengelley: my next question [01:12:20] Dave Pengelley: what are you trying to achieve? Where is the value? How does that help you? Um, like I was, I was thinking about my mechanic down the road who's been helping me out and like what... A- and I haven't had the chat about AI specifically, but just thinking about like, you know, they're always gonna have business. [01:12:34] Dave Pengelley: Cars break down, they're gonna come in. They don't necessarily need to market, so some kind of like helping them augment their CRM or their da- Like, I don't know that that particularly helps them to know which car had a breakdown and which time, which record. Maybe if they've got warranty issues, then that comes up. [01:12:48] Dave Pengelley: But I'm like, just for general marketing, who cares about their CRM? Because- Yeah ... that kind of business is never not gonna be busy, so. Very [01:12:57] Richard Webbe: true. [01:12:58] Dave Pengelley: Again, that's horses for courses. [01:13:00] Um, but good, good questions, good things to talk about, and things we are happy to have more conversations with businesses about. [01:13:06] Dave Pengelley: If you're interest have been piqued on any of this, uh, reach out to any of the hosts. We're all happy to chat. Links are in the description. Make sure to like, subscribe, and follow the channel. If you love these live chats, if you love the occasional clips we put up, make sure you're following us. Hit the bell icon so you're notified every Wednesday when we go live. [01:13:25] Dave Pengelley: Thank you all. Great discussion. Really enjoyed it today. [01:13:28] Shawn Hain: Likewise. [01:13:29] Dave Pengelley: Um, yeah, I'm gonna sign off and if you guys wanna hang around, don't log off straight away so we can say goodbye. Um, I'm just gonna... I'll play us out. Thank you all. All right. And we'll see you on the next one. [01:13:43] Richard Webbe: Thanks, Sean. Thanks, Dave. Thanks, Manjula. [01:13:46] Richard Webbe: Very impressive. [01:13:46] Shawn Hain: Thanks, [01:14:00] everyone.