Stop Asking When You Can Have an Agent. Ask Why You Need One.

[00:05] Dave Pengelley: My um my camera looks like it's funny, all our lighting is so different. All our lighting and color correction. [00:12] Richard Webbe: How do I [00:13] Dave Pengelley: We [00:13] Richard Webbe: fix [00:13] Dave Pengelley: could be [00:14] Richard Webbe: that? [00:14] Dave Pengelley: we could be three different species of human. [00:16] Matt: Yeah, my like paleness is emphasized with mine a [00:20] Dave Pengelley: Yeah, [00:20] Richard Webbe: I'm [00:20] Matt: lot. [00:20] Dave Pengelley: Rich. [00:20] Richard Webbe: very I'm I'm very pink. [00:22] Dave Pengelley: Yeah, you got you got the whole your camera's making you look very pink. Have you got a big light in front of you? Turn the light off. [00:30] Richard Webbe: Um oh [00:30] Dave Pengelley: Yeah, [00:30] Richard Webbe: sorry, [00:31] Dave Pengelley: because [00:31] Richard Webbe: I've [00:31] Dave Pengelley: your camera's [00:31] Richard Webbe: got the [00:31] Dave Pengelley: overexposing and it's compensating. [00:33] Richard Webbe: I've got a a window and I'm trying to block some of that light out. [00:40] Dave Pengelley: That's better. [00:44] Matt: I'm just leaving mine. You can deal with paleness. [00:49] Dave Pengelley: I'm very, I'm very I'm I'm I've got a big light on me, but I've I look darker than you guys. Um it's [00:54] Matt: I'm [00:55] Dave Pengelley: weird. [00:55] Matt: probably gonna get my back wall painted soon, so you know. [01:00] Richard Webbe: Sorry, [01:00] Matt: It's not [01:00] Richard Webbe: as we talk about our skin pigmentation, are we live? [01:03] Dave Pengelley: Yeah, yeah, yeah. [01:06] Dave Pengelley: This is the show, mate. This is [01:08] Richard Webbe: That [01:08] Dave Pengelley: this is [01:08] Richard Webbe: is [01:08] Dave Pengelley: what we're [01:08] Richard Webbe: very funny. [01:09] Dave Pengelley: about. Um [01:10] Matt: a good one. [01:11] Dave Pengelley: we we always well, we we do have a tendency or I do have a tendency to lead off with you know technical inside baseball talk around our cameras and microphones and headphones and setups and everything. [01:22] Richard Webbe: It's good [01:23] Dave Pengelley: Um [01:23] Richard Webbe: cheese. [01:24] Dave Pengelley: yeah, I've got windows in front of me too, but I close them so that way the back wall doesn't light up too much and I just The colored light on the back wall [01:31] Richard Webbe: Yeah. [01:32] Dave Pengelley: versus like other things. I've got this monstrosity above me where it's just a single bayonet fitting, but I bought this like triple adapter for my bayonet fitting, and I've got three lights plugged into it. Two to put color on the back wall, and one just to put this little shine on the back of my head just to help help me stand out. So [01:49] Richard Webbe: Excellent. [01:50] Dave Pengelley: yes, yes, that's uh all earned, all the uh colours and everything. Um Um but you know webcams webcams do a good job too. [02:04] Matt: Yeah, I mean that's all I've used forever. So [02:07] Dave Pengelley: You [02:07] Matt: I'm [02:07] Dave Pengelley: got [02:07] Matt: getting [02:07] Dave Pengelley: the cool [02:07] Matt: away [02:07] Dave Pengelley: blow [02:07] Matt: with it. [02:08] Dave Pengelley: going on, you got the the the depth of field [02:11] Matt: Yeah, [02:11] Dave Pengelley: effect [02:11] Matt: well that's [02:11] Dave Pengelley: behind [02:12] Matt: actually [02:12] Dave Pengelley: you. [02:12] Matt: that's fully just Mac OS camera settings. So [02:17] Dave Pengelley: Yeah, [02:17] Matt: yeah, [02:17] Dave Pengelley: it's just [02:17] Matt: like [02:17] Dave Pengelley: real time rendering it, it's not in the camera. [02:20] Matt: Yeah, yeah. I don't know. I feel like I'm just cheating. Probably should get something good one day. Change the angle, maybe get a light. Like I said, paint the background, make it proper studio stuff. But there's [02:31] Dave Pengelley: My [02:32] Matt: this [02:32] Dave Pengelley: my [02:32] Matt: thing to do. [02:33] Dave Pengelley: yeah, my my my biggest thing that I I would like if I had a different setup or a different place to be is push the camera further back so that I wasn't, you know, like two foot from the lens. Uh just helps with your eyelines and stuff, and you can position a monitor to see like if I'm talking straight down the camera now, I'm not looking at you guys. I can just see you guys out of my peripheries, but I'm talking to the camera. [02:53] Richard Webbe: I have set at your table. So, what's new in the world for us today? [02:58] Dave Pengelley: Well, well, um Richard's got somewhere to be. He's in a hurry today. [03:02] Richard Webbe: No, no, I've just I wanted to get off the camera conversation now. Our audience would be getting very bored about me. [03:09] Dave Pengelley: No, they love it. They love they love hearing about how podcasters run their podcasts. That's that's part of the the vibe. Um I get feedback from that all the time. People are like, oh Dave, like uh tell us more about how you run a podcast. Don't just do the pod no, I'm just kidding. Um I understand, I understand. You you you're kind of a big deal Richard and and I apologize I've been I've [03:32] Richard Webbe: I [03:32] Dave Pengelley: been [03:33] Richard Webbe: have [03:33] Dave Pengelley: wasting [03:33] Richard Webbe: not a big [03:33] Dave Pengelley: your time [03:33] Richard Webbe: deal. [03:34] Dave Pengelley: with with all the all of this camera talk um because [03:40] Dave Pengelley: you are a [03:42] Richard Webbe: Oh [03:42] Dave Pengelley: published [03:42] Richard Webbe: dear. [03:43] Dave Pengelley: writer in the Australian [03:44] Richard Webbe: I have. [03:45] Dave Pengelley: um [03:45] Richard Webbe: I did that for a number of years writing about all the experts in the industry who uh who are a lot of fun at Warwick Pye is an old friend of mine who was used to be head of Telstra International and then he was uh head of Seoul Communications and uh uh Very smart guy. And he's mooting house. And he found this article that I wrote about him. I did about 20 or 30 articles about industry leaders. And it was really interesting in those days because you got to learn so much. Like we do on the podcast with our discussion, you get to learn so much by just these general questions that you wouldn't normally ask people. And they're very honest and open when they know they're going to be published. So all the best to Warwick if he's watching. [04:26] Dave Pengelley: That's great, man. Well done. Well done. Um, having a race. Yellow column is no small feat. [04:32] Richard Webbe: Yeah, I I failed English and I uh I never thought it'd be any good, but there was a guy, um Peter Little, I think is his name, who was one of the publishers for or editors for the Australian, and he encouraged me and I said, I don't know if I write very well. But of course, my new book coming out soon on how to screw up selling. Thanks to you. [04:54] Dave Pengelley: That looks great, mate. That looks awesome. I'm uh excited excited for uh the edits to find Analyze so we can publish [05:01] Richard Webbe: Yeah, [05:01] Dave Pengelley: it. [05:01] Richard Webbe: yeah. I've [05:01] Dave Pengelley: Um, [05:02] Richard Webbe: got s yeah, I've got some edits to let you know on that, but [05:04] Dave Pengelley: yeah. [05:04] Richard Webbe: thank you, sir. [05:04] Dave Pengelley: No, some great stories in there. Great stories. And I mean that that gives me a bit of a segue to something that I'm planning and I haven't even spoken to you guys about. But I've been trying to work out as I've done a few other podcasts and bits, bits, and pieces, and been interviewed and spoken to other people. Um, and got me thinking that there's a space for what I'm thinking we'll uh call a separate thing that we'll do, not the live stream, called operator stories, where we'll get people in. And just uh get them to talk about not it doesn't even have to be specifically AI focused. Oh, AI is probably going to come into the conversation organically because they're operators and they know their industry and their and their subject matter expertise, and we can talk about the AI lens on them. But I think there's some really clever people out there with some really clever stories, really interesting ways of thinking about things. Um, that and those sort of principles can then come into the agent tech future as well. Um, especially, especially if there's a lot of people that have, you know, um. found themselves between full-time employment, whether they're underemployed or unemployed or whatever, they've they've gone through restructures and stuff. And I I know personally myself, like when you spend time in that space, you start getting some self-doubt and confidence issues about did I know anything? And I know myself when I jump on some of these podcasts, and there's one that hasn't been published yet. I did last week, where I ended up basically talking sales leadership. And I've never worked as a sales leader, but I worked at Salesforce in sales for long enough to know how to do sales leadership. Uh And so I ended up like presenting all this information around, you know, sales strategy and leadership techniques and customer negotiation and value selling and all this kind of stuff. Um, and it's uh just made me think there's there's a space for us to share those stories for all these people that have knowledge and to share that with our audience. [06:50] Richard Webbe: I have a lot of people who would love to come and talk on that too. [06:53] Matt: I think it's a good idea. 'Cause uh there's sometimes like no connection for somebody, like you know, if they could listen to us ramble and talk about all our different things and doesn't make sense. They're like, oh yeah, that's cool. But [07:04] Dave Pengelley: Yeah. [07:04] Matt: until they have that experience themselves, uh, you know, that might click with somebody else, you know, their unique take. [07:12] Dave Pengelley: Yeah, [07:12] Richard Webbe: Yeah. [07:12] Dave Pengelley: and and this podcast I went on to um give it a plug. It's called the SEO lab. Uh it's um Yeah, yeah. I I posted a screenshot after I did it. Um they've they're gonna do some editing before they post it up. They've got a pipeline, they're they're doing a whole pre-recorder thing versus our live bit. That's great. And um But it's called the SEO lab, and we barely spoke about SEO at all, which was another thing that got me thinking, okay, once you start building a podcast, like that's the branding, but [07:38] Richard Webbe: Yeah. [07:38] Dave Pengelley: you can you can innovate within that. And I think [07:40] Richard Webbe: As [07:41] Dave Pengelley: you [07:41] Richard Webbe: I would [07:41] Dave Pengelley: know. [07:41] Richard Webbe: say, what is the problem you're trying to solve with your uh podcast? [07:44] Dave Pengelley: Well, yeah, and I thought [07:45] Richard Webbe: Stay [07:45] Dave Pengelley: about setting [07:46] Richard Webbe: on [07:46] Dave Pengelley: up [07:46] Richard Webbe: track. [07:46] Dave Pengelley: setting up another podcast. I was like, I don't want to manage another YouTube channel. Um, this sort of fits within, I think, that that operator language. Uh, and I think like because like you said, Richard, with your interview there, like like with Borick, like you Interesting stories, interesting people you're interviewing for the paper. This is the modern version of that in many ways, the podcast storytelling interview [08:07] Richard Webbe: It [08:08] Dave Pengelley: thing. [08:08] Richard Webbe: is. It is. And it's it's unique this medium because we can ask lots of questions that other people, you know, what's our industry all about? AI to technology. No one wants to ask the obvious questions, even though they're so obvious, because they don't want to be seen to not knowing. And that's what makes podcasts so great. You can whack the head headphones in your ears while you're bike riding or driving your car. or traveling somewhere and all of those things you ever wanted to ask they get asked they get discussed openly without any fear and [08:36] Matt: And [08:37] Richard Webbe: I think [08:37] Matt: if [08:37] Richard Webbe: it's [08:37] Matt: they [08:37] Richard Webbe: a great [08:37] Matt: don't, [08:38] Richard Webbe: and [08:38] Matt: then [08:38] Richard Webbe: if [08:38] Matt: you [08:38] Richard Webbe: you [08:38] Matt: can always [08:39] Richard Webbe: don't [08:39] Matt: put it in the chat. [08:40] Richard Webbe: exactly in the chat right now I agree [08:44] Dave Pengelley: Yes, chats and comments, we love them. They're our lifeblood. I think we're bleeding out. Um [08:53] Dave Pengelley: all right. Uh [08:54] Richard Webbe: so [08:54] Dave Pengelley: yes. [08:54] Richard Webbe: what's that theme this week [08:55] Dave Pengelley: So [08:55] Richard Webbe: other than podcast [08:56] Dave Pengelley: no no no I've got a segue, I've got a segue. [08:58] Richard Webbe: oh [08:58] Dave Pengelley: I've [08:58] Richard Webbe: good [08:58] Dave Pengelley: did [08:59] Richard Webbe: good [08:59] Dave Pengelley: lately. Uh so it's interesting. You you are a published writer and uh we talk about AI things all the time. [09:05] Richard Webbe: Yes. [09:05] Dave Pengelley: I saw this come up in my feeds this week on that topic. Uh, that Substack are now putting in this tagging, in this scanning for AI on their articles. And so apparently you you can scan the article, or I don't I couldn't see it on my own Substack settings, so maybe it's not out yet. But uh yeah, then you can tag your content as like fully. [09:30] Richard Webbe: How can they tell? How can anyone tell? Again, ignorant comment from the commercial guy. Can you can you get a bit of you know writing and have AI look at it and say, yeah, this wasn't written by a human? [09:42] Matt: It it's like a it's a spectrum. You know, like there's there's very obviously generated and then there's very not because it's just patterns. So if there's no patterns, then it's likely a human. It's the same thing when they they do those browser tests. You know, when you go to open up a website and it's like, are you a human? and That in that moment it actually tracks your mouse cursor, you know, and a human will go like [10:06] Richard Webbe: Wow. [10:07] Matt: and then click the button, whereas [10:09] Richard Webbe: Yeah. [10:09] Matt: uh a robot situation will go for like from here to here, you know, [10:13] Richard Webbe: Yeah. [10:14] Matt: direct. [10:14] Richard Webbe: Oh, [10:15] Matt: Um [10:15] Richard Webbe: that is very clever. I like that. [10:17] Matt: yeah, [10:17] Richard Webbe: I've never [10:18] Matt: it's [10:18] Richard Webbe: had [10:18] Matt: just [10:18] Richard Webbe: that explained like that. Well done. [10:20] Matt: yeah, it's just pattern recognition. It's like I've been trying to think of like many different ways to create my own content for various platforms, you know, like on LinkedIn. In like on X scripts for YouTube, that sort of thing. But the the stuff you get out is always so like revoltingly AI generated. So [10:41] Richard Webbe: Oh [10:41] Dave Pengelley: Here's [10:41] Matt: you [10:41] Richard Webbe: yeah. [10:41] Matt: need [10:41] Dave Pengelley: the [10:41] Matt: to [10:41] Dave Pengelley: thing, [10:42] Matt: go [10:42] Dave Pengelley: it's [10:42] Matt: through. [10:42] Dave Pengelley: not this, it's this, but here's the kicker, and that's when it that's when it hit me. All the generic isms. [10:47] Richard Webbe: It's [10:47] Dave Pengelley: That's [10:47] Matt: Yeah. [10:47] Richard Webbe: dreadful. [10:48] Dave Pengelley: wild that you gotta [10:48] Richard Webbe: You [10:48] Dave Pengelley: try and [10:49] Richard Webbe: know, [10:49] Dave Pengelley: filter [10:49] Richard Webbe: you [10:49] Dave Pengelley: out. [10:49] Richard Webbe: know, we spoke about this a few months ago, uh how kids don't like AI. They've had it marketed to them. It's this evil thing that's taking over. So all the kids have naturally gone in the other direction. Oh, I don't like AI. And I have two very creative children. And they throw a fit, you know, they're 11 and 14. They go, No AI, Dad. No, we know you love AI. You're I'm I'm the evil overlord now because I love AI. And then I showed them the book the other day and said, Look, your dad's written a book. What do you think? And they go, I bet you AI wrote it. You're such an AI person. We don't want to read it because it's all AI. Go, no kids, I actually wrote this. But they wouldn't believe me. [11:30] Matt: Don't I don't see the problem with using an AI system like that. Let's say that you used AI to help draft or scaffold, you know, put your ideas together, and then you took those ideas and kind of like rewrote it into your own words with your own anecdotes, your own little like intonations. That's no different to then you know learning from a teacher or some sort of mentor or coach, and then you're taking their ideas that they [11:55] Richard Webbe: Absolutely [11:56] Matt: inspire [11:56] Richard Webbe: true. [11:56] Matt: you with. Yeah. [11:57] Richard Webbe: Absolutely [11:58] Matt: But just [11:58] Richard Webbe: true. [11:58] Matt: blatantly copying. And pasting it, like that's definitely bad. [12:02] Richard Webbe: The cut and paste economy, I agree. I read an article about that last night, the cut and paste economy, it's a worry. [12:08] Dave Pengelley: Well, what was really cool, and and so I've I've I've been working with Richard to help package that up and get it up on the Kindle store. And uh I took his his raw document manuscript and I gave it to Hermes and said, Hey Hermes, we want to put this up on Kindle. Can you help us format this up as a proper book? And he goes, Cool. What size book layout are you using? What are you what paper? And and so, yeah, okay. I went and measured some of my own books and went, Yeah, okay, we'll go to the seven by Or whatever, or the six by eight, or whatever that is. Um, and then it went and created a Word document with all the pages in the right size and like made all the formatting standardized and made sure chapters had page breaks and were on new pages and just did the whole thing for me. It's like saved me out like hours and hours and hours of work. Then you've got to go through and go, okay, yep, that sentence slipped on that line. We'll fix that, we'll change that. You do the tweaks, but the bulk of that editing that would have taken me half a day, I just did while I was off doing other stuff. It was amazing. [12:59] Richard Webbe: Yeah, it should be good. Because I'm I'm borderline dyslexic, son of an intelligent person, by the way. I did use actually, you know what I did? What I did that was AI that I liked. Once I'd I'd I'd written it, I gave it to AI and said, could you tell me how good is this book? And I said, you know, compared to most, you know, instructional, it's called, you know, how I screwed up selling. So it's basically it comes in the business genre storytelling books zone. And it came back, and it was very nice. It gave me a 7.7 or 8.7 something out of 10. And it uh it actually read the now, you know, obviously it's not that objective, but I thought it was pretty good. And it did pick up on a couple of spelling mistakes or a couple of phrasing issues that made no sense. And that's, you know, words used wrong spot. But that's the only AI I use to do it. Because for me, storytelling is just that. It's my story. I want to tell it [13:54] Dave Pengelley: Yeah. [13:54] Richard Webbe: my way. [13:55] Dave Pengelley: Yeah, you can tell reading through it that the that AI did not make those stories up. [13:59] Richard Webbe: No. No one just quietly, like when I went to the nightclub, no one could make up my stories. [14:06] Dave Pengelley: Indeed, indeed. But there are people that are, you know, to your point, Matt, just you know spinning out AI generated content like there's no tomorrow. And Substack um are worried about it. I love this quote here. Uh Tom's posted this up. Substack is combating what they call clawed fishing, which is not flagged as an anti-AI move. They say that not everything made with AI AI is slop, but readers should be armed. The knowledge that what they're reading may not have much or any human thought. Substack co-founder Chris Best said, we don't want to wait until your SubSec app turns into LinkedIn before we start to learn and make progress. [14:43] Richard Webbe: That's very clever. That's very clever. [14:45] Dave Pengelley: That's so good. [14:46] Richard Webbe: You know, intellectually, intellectually speaking, when you talk about content and AI and stuff like that, it's you know, content is about branding, isn't it? It's about this is my intellectual capacity, here's the content I create. created. You can associate with me. So if we're going to engage, then you know I've got this knowledge or I can share that knowledge with you for monetization, for relationship, for building a deal for whatever reason and stuff like that. And I think that works, you know, perfectly. Now if we try and apply the mathematical version of that, I go back to that fun story when they first had automated programs buying and selling shares on the New York Stock Exchange and the stock exchange crashed and they had to turn them off because everyone copied. And I get back to the Venn diagram And uh the the the name of the company that Dave and I started called Syllogism, which is like a definition of fuzzy logic. You just get the same net average of news out, all be the same if everyone's using AI to write the content. There'll be no opinions, there'll be none of that. And I read a great article on um IQ and that humans need to disagree to be creative. And while people are getting all upset with this, you know, do the right thing and let's not call People out, and you know, all these laws stopping people using hate thing. If you don't disagree, we will create nothing and we'll become a bland world of nothingness. So AI content that I've seen is very quickly the same everywhere. It seems to run through a seam for about a week through all our news sources, and then it's just useless, and no one would give it any value. It's the new stuff, like the stuff we come up with on this program, that's really important. Do you like that plug? That was a good plug, wasn't it? I just sort of [16:30] Dave Pengelley: So [16:30] Richard Webbe: Segments. [16:30] Dave Pengelley: I I wish I wish the algorithms would feel the same way. Um, you know, could use LinkedIn, use video to grow on LinkedIn, and I post original video of original thought and original content, which has not had a an ounce of AI scripting on it, and you get 50 views, and you're like, everyone's lying to me. But anyway, let's [16:48] Richard Webbe: It's [16:48] Dave Pengelley: let's [16:49] Richard Webbe: momentum [16:49] Dave Pengelley: proceed. [16:49] Richard Webbe: dive. [17:02] Dave Pengelley: Speaking of momentum, let's stop yours. Um it's I think it's become a running gag that I I cut someone off talking every [17:09] Richard Webbe: He's always [17:09] Dave Pengelley: week. [17:10] Richard Webbe: right because I'm most cool. [17:12] Matt: That um that whole like you know understanding the algorithm thing though, it's the same as what you've just said, Richard, where it's like if everyone has all of the secrets to the algorithm, then how do people actually succeed? You know, I don't I I don't know the answer to that because. Because X, you know, like Twitter will completely share their entire algorithm. It's completely open source. And you can see exactly what gets demoted and downvoted and what will get promoted. But it doesn't necessarily mean that if you know that, you're just gonna win. Like it's interesting. I don't know what the answer is, or I don't know [17:52] Dave Pengelley: And [17:52] Matt: if anyone will. [17:53] Dave Pengelley: I mean I I I have my own personal ethical morality which comes into play where I don't want to do the DM secret and I'll send you the secret or or comment, comment the word secret, and then I'll uh DM you the secrets. It's like that is just pure engagement bait. That is just that [18:11] Richard Webbe: There [18:11] Dave Pengelley: is [18:11] Richard Webbe: is no [18:11] Dave Pengelley: just [18:12] Richard Webbe: secret. [18:12] Dave Pengelley: gaming the system and rigging things to get engagement. And the algorithm, for whatever reason, they haven't processed that if it says in this thing, comment this one word, when people comment that one word, that's not real engagement, that's just fishing. Um That should be a super simple thing algorithmically, deterministically, to be able to filter out and block, but they don't. And so then people do that, and the more they do it, the more everyone else does it because that's what gets results. And so then you go, like, oh, I don't like doing that because I think it's rubbish. Oh, good. Play on hard mode, Dave, and get nowhere. So eventually you cave and you start doing it. And then you're like, ah. [18:52] Matt: A couple of [18:53] Richard Webbe: Yes, [18:53] Matt: years ago, [18:53] Richard Webbe: I [18:53] Matt: like that the whole the whole reason for many chat existing was basically just to facilitate those things. Yeah, [19:00] Dave Pengelley: Yeah. [19:00] Matt: that was one of my first freelance jobs was wiring in many chats and allowing people to do like Instagram posts with that and have it all fully automated and triggered. But you're right, it's pretty lazy now and it's pretty gross now. Like it doesn't feel like good content. It doesn't feel like you want to participate in that either. You're like, well, no. [19:18] Dave Pengelley: Yeah, I judge I judge people when I see when I see people like it pops up in my feed because you know there's a second or third level connection that's done the post, but someone that I'm connected to comments the secret word and I look at them and go, Ah, you want Of them. You're making it worse for all of us. There's no prompt secrets. [19:35] Richard Webbe: Humans [19:36] Dave Pengelley: You [19:36] Richard Webbe: are [19:36] Dave Pengelley: don't [19:36] Richard Webbe: and [19:36] Dave Pengelley: need ten special prompts. [19:38] Richard Webbe: look, it's a very interesting point you make, and I think humans are incredibly adaptable in both directions. They're adaptable to have their attention captured, but they're also adaptable to turning off and producing your own subconscious filter to bullshit. It comes up, right? And so um uh, you know, I remember you know, you read articles on many. Different things like how to meet someone or how to open a conversation or how to be a really good presenter. And pretty soon, once that article's hit the theme, everyone's doing it. And within a very short period of time, all of a sudden it is useless and ineffective. So creativity, I think, with AI is still number one with marketing. It's still number one with sales, with strategy, it's still number one. Because when everyone starts doing the same thing, our brains are very quick to spot those patterns and go, you know what? But I'm like Dave said, I'm not going to engage. It's not interesting for me. [20:34] Matt: Hmm. You're right. And then there are those patterns though that are less um prickly. I don't know if that's the right word. Where you you your your radar. Yeah. You know, think of just the biggest companies in the world. Think about when like Apple or somebody will release a new marketing campaign. You know, it feels very Apple, obviously, because that's their pattern. But there's none of those like, I don't know, real gross, prickly things, I feel. So like That's where AI could come in. It could be okay, let's analyze what all these big recent campaigns were. What are the patterns? What are the triggers and the mechanisms? [21:08] Richard Webbe: Yep. [21:09] Matt: And how can I apply that to my campaign? And [21:12] Richard Webbe: Yeah. [21:12] Matt: then you just do it more legitimately yourself using their structures rather than hey, create me a campaign, you [21:19] Richard Webbe: Yeah. [21:20] Matt: know. [21:20] Richard Webbe: The company in the group that's first always win. There's the me's and then the me twos. So the me twos will get there, and then the me threes and the me fours, who are usually Lower resourced are not as quick to follow, they always end up down the tree and probably worse off than if they just came up with a bad campaign. [21:39] Dave Pengelley: Yeah, from a campaign point of view, yeah. I think from products, sometimes a second or third, if they innovate well, um, they wait like Apple is a is a classic example of that. Microsoft were trying to do personal handheld computers from the late nineties, but the technology wasn't there to make it a real viable good user experience. Steve waited till you know in the late twenty ten, drop your iPhone and everyone started losing their minds. [22:04] Richard Webbe: Yeah, [22:04] Dave Pengelley: The technology [22:04] Richard Webbe: yeah. [22:05] Dave Pengelley: caught up to the vision. [22:06] Richard Webbe: Mm-mm. Yeah, yeah. It's it's certainly you've got to paint the vision. I always used to in my talks, I always talk about um what marketing's like watching little children, little babies. And I think I might have said this before. But a baby or a human can't walk unless in their mind's eye they can imagine themselves taking a step. But little Babies watch us walk and they can't imagine themselves doing it. And that's why it takes a while before they learn how to walk. And then once they do, they go, Oh, I know how to do that. And their brain very quickly says, Oh, my feet can do that. And then you see them walk. And it's usually a very quick, amazing, and wonderful process when you're a father or a parent. But um, [22:47] Dave Pengelley: Yeah. [22:47] Richard Webbe: yeah, and I think marketing is similar to all of that stuff. It's all how our brain works. And people come out with something new and everyone gets excited. But as soon as you copy it, you go, Yeah, whatever. I'm over it. We adapt pretty quickly. [23:00] Dave Pengelley: I think because so much of this content is so easy to create, we devalue it and we just we dismiss it very quickly. Like the AI imagery when it first came out, you go, Oh, that's really cool. You could make a a dog on a skateboard, like chucking a penguin [23:14] Richard Webbe: Or [23:14] Dave Pengelley: through [23:14] Richard Webbe: a cat [23:14] Dave Pengelley: out [23:15] Richard Webbe: on [23:15] Dave Pengelley: the [23:15] Richard Webbe: a [23:15] Dave Pengelley: moon [23:15] Richard Webbe: robo [23:15] Dave Pengelley: or [23:15] Richard Webbe: vacuum [23:15] Dave Pengelley: whatever. [23:16] Richard Webbe: cleaner. Sorry, that [23:16] Dave Pengelley: And [23:16] Richard Webbe: was real. [23:17] Dave Pengelley: and [23:17] Richard Webbe: Come [23:17] Dave Pengelley: now [23:17] Richard Webbe: on. [23:17] Dave Pengelley: you see that or you see marketing copy and you see brochures, AI generated brochures, and you know that's not like you back in the day you'd get a publisher template and you know that was a template. A designer didn't do that, and Canva then came along and made everything templated as well, and you could Spot some of that. But now with AI, they they all have the same look and feel to the brochures, the gradients, the colors, and and just everything looks very AI gen, the pills, the information pills everywhere. Um, and so we go, oh well, that we devalue that. And maybe it's good content, maybe it actually is interesting, but I think our brains are already biasing against it because we go, whoever made that probably didn't even put any content into it. They probably just got AI to do the research, think about the thing, put it together, publish it out. And so because no human Was involved, I actually don't care. So it's funny, this whole pendulum, like, will AI replace the world? I think we are gonna see people wanting human content, wanting human things, wanting physical things that physical people made because they're rejecting this sort of lack of humanity that the AI stuff's creating. [24:20] Richard Webbe: And our brains are very quick to spot it. Hey, look, it's first to market, I think, always still has the advantage. Like you said, sometimes, Dave, the tier two people will add something clever to something. I remember I was sitting in the office of a particular person, and I I reckon it was one of the first in the world, but I was in IBM in the finance branch, and I went and saw the guy that brought out the first home loan calculator. Had never been done before, had never been given, you know, you used to have to go see your bank manager, and he'd sit there and go. [24:52] Dave Pengelley: Was it Russie John? [24:54] Richard Webbe: No, well before him. And so it's very interesting because uh at the time I think pretty Was ANZ that came out with a home loan calculator, their home loan wins and their back wins just went through the roof. Of course, now we go online, everyone's got a home loan calculator, and it's a known quantity. But people identified back in those days with ANZ being the bank for helping people because it gave you the information that everyone used to block you getting in the old days, right? There weren't that many templates. It it coincided with, of course, the launch of Excel and things like that. Of course, when I buy Excel, I'm not an accountant. Want to get Canva, I'm not an artist or an advertising person, and that's the point. The point is we're given tools, but the creativity, the human essence must be introduced. [25:39] Dave Pengelley: Yeah. Yeah, I mean I'd I'd always say to my designers when I had a team that I know when something doesn't look right, but I don't know how to fix it. That's what I need them for. [25:50] Richard Webbe: Yes. [25:50] Dave Pengelley: Like um and I think you know all the AI slot that you're seeing on websites and everything is because people don't really know what to ask for. They don't know how to fix things, they just hope the magic Black box can fix it all. Uh and we discussed this like when I was trying to build my valid agenda site a few weeks ago, and I've got that fly through with the 3D grids and meshes and stuff. That took a while to get to because I didn't necessarily have the right language to use. I didn't know how to articulate what I wanted in design language for the AI to know. So it was trying to, you know, decompress it and translate my random errant ramblings into what I actually wanted to create. Um, but yeah, it's interesting what you talk about the home loan calculator. Um, I went to an event last night. [26:32] Richard Webbe: Here we go. [26:33] Dave Pengelley: Yes. Matt, you love you love this Matt. Got some swag [26:38] Richard Webbe: Damn, [26:38] Dave Pengelley: from [26:38] Richard Webbe: damn, damn, [26:39] Dave Pengelley: Oh, [26:39] Richard Webbe: back, [26:39] Dave Pengelley: there [26:39] Richard Webbe: back. [26:39] Dave Pengelley: it is. [26:40] Richard Webbe: Verso, oh, [26:42] Dave Pengelley: Yep. So I um where's the thing? There we go. So I I got a tote bag, I got a water bottle, I got a pen. I got this is the most random swag in the world, right? I've got this little thing. I'm pretty sure this is just a key off a keyboard with the version. [26:59] Richard Webbe: yeah. [27:00] Dave Pengelley: icon on like i i i haven't opened up i think it's it doesn't do anything it's not even a light it's nothing it's just a key really weird swag maybe they're trying to take on the openai keyboard thing [27:14] Matt: No, [27:14] Dave Pengelley: um [27:14] Matt: that's just like have you never seen mechanical keyboards like the like building a custom keyboard? [27:21] Dave Pengelley: you forget how odd i am that i [27:25] Dave Pengelley: i have i have pulled apart and cleaned coffee out of and reconstructed and and [27:31] Matt: I've got boxes of keys here. [27:36] Dave Pengelley: why [27:37] Richard Webbe: Oh, [27:37] Matt: For building [27:38] Richard Webbe: that's [27:38] Matt: custom [27:38] Richard Webbe: impressive. [27:38] Matt: keyboards. [27:40] Dave Pengelley: so I d I didn't realise that's a that's the thing the kids are into now building custom keyboards. My my bad. Um yes, my [27:48] Matt: Yeah, [27:48] Dave Pengelley: bad. [27:48] Matt: it's [27:49] Dave Pengelley: But [27:49] Matt: it's a [27:49] Dave Pengelley: no [27:49] Matt: full [27:49] Dave Pengelley: I [27:49] Matt: thing, like where you you [27:50] Dave Pengelley: yeah. [27:51] Matt: you customized your switches for certain feel. [27:53] Dave Pengelley: Yeah. [27:54] Matt: Yeah. So like in the in the guts of that, like that'll actually sit on top of a a certain [27:58] Richard Webbe: Function [27:58] Matt: type of [27:59] Richard Webbe: K. [27:59] Matt: switch. [27:59] Dave Pengelley: Yeah. Yeah, [28:00] Matt: And [28:00] Dave Pengelley: yeah. [28:00] Matt: then you can use different greases and lubricants to change the feel and different packing inside your actual keyboard uh frame to dampen and change the [28:10] Richard Webbe: Looking [28:10] Matt: sound. [28:11] Richard Webbe: like a true gamer, man. [28:13] Matt: And am I gonna ruin all of that and say to all of the keyboard crowd out there that I actually just use one of these? [28:22] Dave Pengelley: I've got my loggy at the moment. [28:25] Matt: Yeah, I don't know what it is. I think since using a MacBook, I just mini caplets just. Work with my dropping style [28:31] Richard Webbe: Oh, [28:31] Matt: now. [28:31] Richard Webbe: the old uh Apple one, there you [28:33] Dave Pengelley: Yeah, [28:33] Richard Webbe: go. [28:33] Dave Pengelley: I I when I'm when I'm using my Mac, I do like the uh the Mac keyboard, the flatness of it, the spacing. [28:38] Richard Webbe: I bought my son a keyboard the other day and he gave me a rundown, Matt. You guys would know this about all the different types of keyboards, the feedback, the click, the mechanical versus the electrical. He can now change the colours on all these key keys and he's just into it completely. And he's oh dad, no, that's too slow. I can't play that game if you give me that keyboard. Being an old service engineer that used to fix keyboards. [29:03] Dave Pengelley: Yeah, right. [29:04] Richard Webbe: Yeah, um, I was flabbergasted by the variation. [29:08] Dave Pengelley: Wild, wild now. And and you know, back back when they did what before PS2, right, Richard, when they were like the five point din, [29:14] Richard Webbe: Yes. [29:14] Dave Pengelley: the uh yeah yeah. [29:16] Richard Webbe: Yes. [29:17] Dave Pengelley: Yes. [29:18] Richard Webbe: Do you know IBM? IBM is a, because I'm, you know, I'm always into business cases and stuff like that. IBM was really great at creating return on investment and total cost of ownership analysis. And they got a group of people. To use their keyboard versus other keyboards. [29:33] Dave Pengelley: Yeah. [29:34] Richard Webbe: And they even created a financial return on investment for using an IBM keyboard. And it had to do with the nanometer pressure back on the fingers, on how quick you could get. They just went into complete detail. I was most fascinated [29:46] Dave Pengelley: Wow. [29:46] Richard Webbe: on how the pressure of the key changed how quickly someone could work on a computer. And it had a fiscal value. [29:53] Dave Pengelley: Yeah. Look, I I I I get it. Yeah. I mean I just the spacing. Like I can type on like the It's got the similar sort of air gap between keys, slightly like the magic keyboard. If I go to a regular old school mechanical keyboard where they're all jammed together with those keys and there's no spacing between them, all of a sudden it's like I'm like the typewriter arms like hitting each other. I'm like, ah, like if my fingers are not trained for that tight spacing anymore. And so it slows me down. So I fully get that, you know, press depth, etc. as well. Mind you, I'm a sucker for the old click-clack. The tick tick tick tick tick tick tick. I do miss that of the old keyboards. [30:28] Richard Webbe: Yeah, I've always to be asked. To be a futurist again and say what's going to happen, I would say the least valuable piece of technology, unless you're a gamer, in the future, will be a keyboard. [30:41] Matt: Boy, what's your thinking there? [30:43] Richard Webbe: Well, I saw someone the other day with this bulbous thing on the head. We're all just going to be talking to our devices. I think in a hundred years the typing, touch typing capability is going to be gone. Because people go, Why are you taking your time typing that in? Just talk to your computer for God's sake. [31:00] Matt: Really interesting sort of take or that I've experienced recently of that. So I I type pretty quick. Like I I had to learn how to [31:08] Richard Webbe: Oh [31:08] Matt: type hilariously. [31:09] Richard Webbe: yeah, [31:10] Matt: I [31:10] Richard Webbe: yeah. [31:10] Matt: learned how to type fast, actually playing games, which is kind of funny. So, like, you know, in a lot of games where you have the ability to chat, you know, sometimes you need to get your message out quickly. [31:19] Richard Webbe: Yep, yep. [31:19] Matt: Other games, you're on you're on voice comms, and it is what it is. Now, when we're dealing with these agent things now, obviously you can continue to touch type and send your message off quickly. Like you just said Richard, you could just use your mic and transcribe or dictate what you want. Now what when I work, I do a mixture because if I just dictate it up, if I just use mic, I actually end up getting really, really stressed. And like my blood pressure rises and I end up having to walk away for a while because speaking you you naturally put tone into it. And when you put tone into it, your body feels it. And then when Most of the time when we're dealing with these AI systems, we're kind of speaking to them in a frustrated way. You know, [32:05] Dave Pengelley: Just [32:05] Matt: you're not listening [32:06] Dave Pengelley: beep, [32:06] Matt: to me, [32:06] Dave Pengelley: beep, [32:06] Matt: you're not doing [32:07] Dave Pengelley: do [32:07] Matt: what [32:07] Dave Pengelley: it, [32:07] Matt: I [32:07] Dave Pengelley: beep. [32:08] Matt: want. And yeah, and you bring in the uh the swearing, the expletives, and you start getting angry and you start yelling at the computer, which is the funniest thing in the world to even consider. [32:19] Richard Webbe: It [32:19] Matt: And [32:19] Richard Webbe: is so true. [32:20] Matt: genuinely though, like when I when I try and use just voice to text, I end up just getting so angry. And but it and I have to every now and then just stop. Stop and go back to typing, and when I type, it's so much more peaceful. [32:34] Richard Webbe: So I did do some research on this a while ago, and um I think I might have mentioned Max Walker, the cricketer of Australia's one of Australia's greatest selling authors, was a good friend of mine for many years before he passed away. And I said to him, I always wanted to do a lot of writing, and he used to give me hints. And it was very sad I've written my first book and he wasn't here to help me with it. But he he had some research, and there was a couple other people he'd worked with. When you write, you A physical pen when you write using a keyboard, when you write using your words or voice, you're actually using very different parts of your brain. And it actually has a different emotional, cognitive, right, and reactionary response to you. So you're 100% correct. But we tend to fall to the lowest common denominator. And that's why I think eventually, because you know, writing with your voice is like you're being put on stage. There's a lot to get right. Whereas when we're writing, we take Oh, I typed that wrong. And while you're doing that, your brain's calculating the next few words and working it out. So it's it's what you say is right on the money, and it's a very clever thing to look at as you go forward. [33:40] Dave Pengelley: Yeah, I was reading something the other day that was saying, um when you speak things, speaking is good for um just generating your ideas and getting your ideas out, whereas typing is better for structuring them and formatting them and that does activate different parts of the brain. And I know that's one of the reasons why and and I wrote a sub stack on this last year around voice being a big winner in 2026 with all the AI stuff, because of that ability to to ramble and to activate different parts of your cortex and ideation in that soundboarding frame versus the opposite of what you said. You said you got to be really think about performing and what you're saying. I I the other way around, it's like, oh, there's a permanence to typing and I have to think about exactly the sentence structure and I need to write the words and I want to get the sentence to be Grammatically accurate and correct, and I need to to punch those keys in the right order versus I'm just gonna say some stuff. [34:36] Richard Webbe: Hmm. One's editable, that's where the brain reacts differently. So when you just say stuff in your frontal cortex, you don't think it's editable because once it's out there, it's out there. Whereas when you're typing, your brain does actually know I can backspace and do the delete button and change a couple of things. But they're all valid because all of it's editable now. And if you're talking like with an otter or something like that and capturing voice, Voice with meetings. Often I see I say something and it appears on the screen and then it reads it and comes back and it all comes out wrong and then it reads it, comes back, and readjusts it and makes it correct. I like autocorrect when I'm talking to a screen. [35:16] Dave Pengelley: Yes. So anyway, I went to a verse elevator and got a key. [35:22] Dave Pengelley: But beyond that [35:24] Richard Webbe: Is it the key [35:24] Dave Pengelley: like [35:24] Richard Webbe: to life? [35:26] Dave Pengelley: uh rabbit holes your law here. Um so the They had a few guest speakers. They had uh Molte Ubal, their CTO, was in town, and he talked a bit about what Versel is doing from their product strategy. But they had a couple other uh SaaS guys get up and talk, and one of them was from Flywheel.co. Um, you've probably never heard of them. They, you know, do some stuff where you spend money on your ads, and they'll basically, I think it's almost short-term loans where they go, Okay, well, we know that you're you'll make four dollars for every dollar spend, and so we'll then The four dollars up front. Um, if you spend the dollar through us or whatever. I don't fully understand the model. But that's not the point. [36:06] Matt: Interesting. [36:06] Dave Pengelley: The point, [36:07] Matt: I need to look that up. [36:08] Dave Pengelley: yeah, yeah. Um, what he talked about was uh go to market engineering. So he was an engineer and he was talking about using your engineering skills uh for go to market because you know, as we as we say, building is arguably the easy part at this point. It's the go to market, it's the getting it out there, it's the sales motion. And he talked about this concept of creating. Using engineering and engineering tools, assets that you can give away for free because they will promote traffic to your product. Um, and the example he had is they wrote an MCP plugin that would aggregate data from, you know, all the various meta ads and Google ads and the various ad platforms and give you a single endpoint where you could ask and interrogate questions across that. And they gave that all they were using that for themselves, but then they published. Set out as a free resource. And what that meant was when people Googled MCP for meta ads, they were coming up very quickly as the top recommended thing. They were coming up in the Google AI overviews. And so that drove traffic to their actual paid offering. Um, and [37:15] Richard Webbe: Self [37:16] Dave Pengelley: so [37:16] Richard Webbe: fulfilling [37:16] Dave Pengelley: it's [37:16] Richard Webbe: prophecy in a technical sense. [37:18] Dave Pengelley: and so when it comes to building these different applications and so on, like for your customers, what are the things that they are searching for and looking for that you Can give them something for free. And this is what Homozy says as well, right? Like give them so much value for free that they'll feel obliged to pay you at some point. Um, but thinking about from an engineering and application point of view is really, really interesting. Um, and something that I'm I I took away from that event, going, okay, that's good. And beyond, you know, having an ROI calculator on your website for if you adopt AI, but what's a real tool? What's a real thing? Is it a chat GPT plugin? Is it something like what's something someone can grab, deploy, and Install use, get immediate value from that, you know, when they're searching for that tool, they might find you as well. [38:06] Matt: Hm, or or why they're searching for that tool in the first place? Like what problem are they actually trying to solve? [38:11] Dave Pengelley: Yeah, [38:12] Richard Webbe: So [38:12] Dave Pengelley: yeah. [38:12] Richard Webbe: if you use your um your offer of uh information, knowledge or a tool to people so that they come to your website and create the research you need to win them over as customers, it's it is like a self-fulfilling prophecy. Just quick minor segue. I saw a very interesting article. On the ABC the other day. I forget the name of the journalist. I do, he's got a great beard. I do love listening to him. He does these little half-hour historical overnotes. And he was talking, he started off talking about the dead bodies on Mount Everest. And either him or someone he knew sat down, and they thought they sat down next to someone, but they were sitting next to a dead body that had been there for a long time because no one has the energy to bring them down. Now, the guy whose dead body it was, and I forgot his name, created a website. That said it's the and became the most accurate predicting website in the world. What a clever, clever concept. So what this guy did is he avoided having to be listed as a gambling website, but he got the audience that were betting on whether something was happening to give their opinion. [39:19] Matt: Was it polymarket? [39:21] Richard Webbe: Might [39:21] Dave Pengelley: Okay. [39:21] Richard Webbe: have been. Might have been. But [39:24] Matt: Polymarket's [39:24] Richard Webbe: it was. [39:24] Matt: one of the big ones right now. [39:25] Richard Webbe: Yeah, well, well, this guy a few years ago was doing quite well, but he wasn't. Clipping enough of the ticket. So he wasn't making enough money. So eventually he embezzled two million dollars from his own company and they were after him. And then he was found dead at the top of Everest, sitting upright like a chair. But I just thought what I I'd never I love thinking of new concepts like you guys do. I think what a great concept. I go and put a website on there and go, everyone tell me their opinion on this, and then get the same group to bet against it, and away you go. Now, if you do that, the majority will always get it right. And most he talked about a thing called um cognitive group intelligence is better than individual cognitive intelligence. That is, if you get a thousand people and ask them a question, your average correct answer is going to be here. But if you ask one person, it's very varied. And he was using that to make money, but because it was so accurate, it didn't make money because everyone got the right answer. But yeah, I'd never heard of this concept being actually published before. It just existed in my mind. And I found it completely fascinating. And from a marketing point What a great thing to do. Yeah, clear. [40:32] Matt: I don't know if it's like my brain just seeks patterns constantly. But as you're speaking, then it makes me think of the concept of a jury, you know, like that you need a court in order to make a decision. It makes me think of AI subagents, you know, when people actually do um proper enterprise grade reviews and penetration tests and stuff, they'll spin up like 50 to 100 subagents all doing the same task. Because there's going to be the deviation and the spread [41:02] Richard Webbe: Yes. [41:02] Matt: of results. So [41:03] Richard Webbe: Yes. [41:03] Matt: therefore, you'll find like the things that one won't find. [41:07] Richard Webbe: Yeah. [41:08] Matt: So yeah, it's very interesting. And I think it's the same sort of pattern that you'll see in a lot of things if you really look. [41:14] Richard Webbe: Yes, yes. And and that's how marketing works. Doing it well, repeating it, and having the majority follow you. Doesn't matter if the minority don't follow you because you're not losing money because the majority is following you. And you're getting that capture is kind of back to what Dave was saying. Give them something of value that they want. And and the world will be the path to your door. [41:33] Matt: That kind of concept too, like just I don't know, this is entrepreneurial founder brain. [41:39] Richard Webbe: That's [41:39] Matt: Um [41:39] Richard Webbe: what we're here for. Go for it, buddy. [41:40] Matt: the the idea of uh the MCP thing that that people are searching for and that's the free thing that you build to give away um to act as like a lead magnet or a flywheel, you know. Um that that kind of concept, um, you know, like how do you think of what to build and how what are those things going to be? Is the kind of thing that we're going to be doing more and more moving forward. Like if you have a product, your product is not one thing. It's a, it's a, it's a combination of parts. And those parts should be their own individual thing. You know, think about it like Lego, like they're like blocks to a whole. You'll see more and more people talking about like building blocks or building components and then distributing or enabling distribution of those blocks. I'm gonna keep using the word block because it just seems to be the adopted term at this point. So, like some of that stuff, you know, that that polymarket style predicting thing, you know, that could become a block. And then depending on your industry, you know, you could plug that block into your particular product to then get that kind of crowd prediction stuff. But you know, it needs to be distributed in the first place. So that, yeah, it's very, very interesting. Like everything that I build now, I try and make sure that it's like Like agent first and built in blocks so that [43:03] Dave Pengelley: Yeah. [43:03] Matt: I can, you know, repeat and use that in other things. [43:06] Dave Pengelley: And [43:06] Richard Webbe: Yeah. [43:06] Dave Pengelley: and I mean, like last week we had Angus here and he's building FIP.app and same similar premise, like he's giving away a few free uses, which for you know, the general tie kicker, that's all they're ever gonna use, and they probably wouldn't pay for it anyway. Um, so he just gives that to them because it's low cost for him to deliver. And then but the people that actually go, This is great, I want to use this regularly, they pay like fifteen bucks or ninety nine for his whole suite, like. It's it's a pretty decent model. Um, but it's you know, get them there with a free thing that that drives value and then they'll find everything else that you do as well. Um yeah, it's uh it's hard when you think about, you know, business consulting and workshops, like what does that look like when you're not a product company? Uh, but I think on the product side of things, I'm thinking about that for my chart reporter app, like what I can do as far as that. And I thought, well, maybe, yeah, so all these different, all the different value ads. What are the free things you can do that drive value for people that people are looking for? Um That you know, solve their immediate need, but pivot them towards you as a solution for their actual larger pain or issues that they might be having. [44:10] Richard Webbe: And and that comes back to um, I know we've spoken about this before, that comes back to what are the things people really need outside of products? And I think today what they want is not the answer, they want the question. And no one knows in their industry what the question is. The question is the thing that's going to protect you from disaster. It's going to insulate you from mistakes. It's going to help you advance your business, your personal life, and everything. And we've got so much thrust us at us now. The question of what am I going to do next, or what do I need to do, is the hardest question. What is that question? The answer's not hard, right? I've got to do this. I'll find anyone who helped me do that. But what is the actual question? And I think that's the new world. That's the world of marketing and that. I need to get inside my market. And find out what are they asking and help them hone that question so they insulate it from the rest of the world changing. [45:07] Dave Pengelley: The the question is when can I have an AI receptionist? When can I have a chatbot? When can I have an agent? [45:12] Richard Webbe: Well, they're all productivity things. [45:14] Dave Pengelley: I [45:14] Richard Webbe: That's the [45:14] Dave Pengelley: know [45:15] Richard Webbe: easy [45:15] Dave Pengelley: no, but I [45:15] Richard Webbe: answer. [45:15] Dave Pengelley: mean that that that's that's where people are though. They're like, oh, when when when do I get this? Or or how do I get this? Not why do I need it? And that's what we keep coming back to, right? All these solutions are out there, but they might be solving problems that aren't really a problem for you. They don't actually relieve any issues, they don't help your business. They're [45:30] Richard Webbe: No. [45:30] Dave Pengelley: just, you know, a thing you can do. [45:33] Matt: And on the on the flip side, they they find the problems that exist that you've been ignoring. You know, just because you want an AI receptionist doesn't mean that you can just drop it in and it'll work because you you all of a sudden now realize that you need ways for that receptionist to do things, you know, [45:49] Richard Webbe: Yeah. [45:49] Matt: to interact with your business depending on what you want it to do. [45:52] Richard Webbe: Yeah. [45:52] Matt: And if you haven't set any of that up, be prepared to need to set any of that up because yeah, it's not. Just always drop in. [46:01] Richard Webbe: Yeah, [46:01] Dave Pengelley: Yeah, if [46:01] Richard Webbe: if [46:01] Dave Pengelley: you're [46:01] Richard Webbe: you [46:01] Dave Pengelley: just scheduling everything on a paper calendar book, then your AI agent receptionist isn't going to be able to schedule meetings for you. [46:07] Richard Webbe: No, no, it's it that's exactly if you if and that's always a business consulting, if you accelerate or legitimize or automate a part of your business, um other parts of your business will fall apart unless you've got an end-to-end story uh and you know the problem you're trying to solve. Uh so the problem is not a receptionist, the problem is getting customers to come in, be scheduled. Be supported and looked after, and the reception is just one minor part of that whole end to end story. [46:35] Matt: With the people that I've worked with, with voice agents, the the smallest, like the absolute base level voice AI solution is it's it's fancy voicemail and it's and it's routing. So you know, you think about a router that most companies will have, like uh f to to contact accounts, press one. To contact this company, press two. Like, you know, it's the the number menu. So instead of having that boring number menu, it's like, hello, you've rung this person. Who do you want to speak to? Uh oh yeah, cool. No, I'll put you through. Like that kind of router is pretty easy. And then the glorified voicemail is who do you want to speak to? About what? Okay, cool. I'll send them a message and they'll they'll get back to you. [47:18] Dave Pengelley: Yeah. [47:18] Matt: You know, there's no tooling there other than like an email tool and uh a call transfer tool. [47:24] Dave Pengelley: Yep. Yeah, I I was um I think Kelly I think Kelly who spoke it out are relevant. Something like that that they've set up for their business where it's basically glorified routing and email. Um, but she said that the time saving that routing in calls going to the right person or leaving a message for them to follow up later has been huge. So there's definitely value in that. But yeah, I remember heading into Christmas. I tried doing a quick offer where I was gonna sell $99 AI receptionists for just two weeks across the Christmas break. And so trade's and stuff gone away, they would divert their phone to go to this voice agent that would, you know, personalized answer the phone. Um, talk to a customer, grab their details down, assess, analyze the the call transcript, and then send them like is it a desperate urgent email? Do they need to escalate it and actually call the call the person straight away? Or you know, not even just send them an email, but actually build like a little airtable list of or or a Google sheet. I think I was just using to simplify it down. But then when they get back, here, here's your Google sheet with your 50 calls to follow up, all transcribed, all prioritized, all summarized to what they actually want to ask you the question to do. Rather than you going, oh okay, I'm gonna have to listen to this thing for the next half a day processing [48:34] Richard Webbe: Exactly. [48:35] Dave Pengelley: through all the voicemails. So [48:36] Richard Webbe: Exactly. [48:36] Dave Pengelley: huge value add there for people that want [48:39] Matt: Let [48:39] Dave Pengelley: to [48:39] Matt: me guess, [48:40] Dave Pengelley: use. [48:40] Matt: like people still didn't see the value even at nine nine dollars. [48:44] Dave Pengelley: Yeah, I didn't couldn't get a thing, couldn't get a single inquiry. It killed [48:47] Matt: It's [48:48] Dave Pengelley: me. [48:48] Matt: crazy. [48:49] Dave Pengelley: I [48:49] Richard Webbe: Stuff. [48:50] Dave Pengelley: even [48:50] Richard Webbe: I [48:50] Dave Pengelley: set [48:50] Richard Webbe: remember [48:50] Dave Pengelley: up a whole interactive demo on my website where you could put in your phone number and I would call you um to put to give you the pretends thing so you could actually hear what an AI voice sounded like. [48:59] Richard Webbe: it. [48:59] Dave Pengelley: And then when you finish. It would send you an email of the call you just had with that AI thing. Like I had a full end-to-end, like arguably overly automated for a product that had no customers. But you know, I was deep into the whole NAN rabbit hole and doing all those sorts of things back in December. Um [49:14] Matt: I think you'll find it all come back because like we've spoken about before, there's a lot of these all these technologies that that us in this like little bubble uh experience. And you know, when you're inside the bubble, you still you feel so bar far behind still. But In reality, like the rest of the world still gonna take like another 12 months to get there. So [49:34] Dave Pengelley: Yeah. [49:35] Matt: when, [49:35] Dave Pengelley: It's [49:35] Matt: you know, let's say six [49:36] Dave Pengelley: a a a bit bit little little little of of [49:36] Matt: months ago, when everyone [49:37] Dave Pengelley: bit [49:37] Matt: was like, you need to get into voice AI, it's the easiest thing to implement for people and the highest value lever. Um, yeah, it probably was like six months ago [49:46] Dave Pengelley: of [49:46] Matt: and we're figuring it out. But in another six months is when the rest of the world's gonna be like, okay, I actually need that now. [49:52] Dave Pengelley: a a a bit bit bit little little little of of a a a bit bit little little of of of a a bit bit bit little little little of of [49:52] Richard Webbe: Yeah. [49:54] Dave Pengelley: that really went all in on it will have productized it and streamlined it and it's gonna be harder to break in as a As a person, if you didn't niche it. And so [50:01] Matt: Yeah. [50:02] Dave Pengelley: the people that went, Yeah, I'm doing AI receptionists for martial arts dojos, and I'm just gonna pound every martial arts dojo in the in the country. Um, you're gonna get an uptake on that. Um, my my problem is I'm I I struggle to niche down because I know I can provide value across so many domains. Going, I'm just gonna be the voice AI guy just doesn't appeal to me, like because I know that's just like a little tiny chink in the bigger [50:23] Richard Webbe: Halistic [50:24] Dave Pengelley: story. [50:24] Richard Webbe: for you guys. [50:25] Dave Pengelley: Yes, but I mean overwhelm. Richard shared this graphic for us. Um There is a lot of different things you can't a bit hard to read, but LLM, agentic, rag, embedding, MCPs, AI security, observability, memory, AI agents, automations, vector databases, and everything else that goes [50:41] Matt: That's [50:42] Dave Pengelley: with [50:42] Matt: quite a [50:42] Dave Pengelley: just [50:42] Matt: good graphic. That's like I I feel like I need you to send me that so that I can use it as a little bit of a not not an ego stroke, but like a do I know what these things are. [50:51] Richard Webbe: Sending to you now and I can't remember who I stole it off, but it popped up with some flash on LinkedIn or Facebook or something that I was looking at [50:58] Dave Pengelley: a [50:59] Richard Webbe: in social [50:59] Dave Pengelley: little a a bit bit bit little little of of of [50:59] Richard Webbe: media. [50:59] Dave Pengelley: a a a bit bit little little little of of bit a a a bit bit little little of of of [51:00] Richard Webbe: thought, oh, that captures the categories I want. And and actually just so you know, and of course, as as you both would know and our audience would know, in each one of those sectors, there's probably four times as many tools and companies as they've listed there. There was a [51:12] Dave Pengelley: Yeah. [51:12] Richard Webbe: couple of obvious ones that were left out from the and it's changing monthly, right? And it's it's outrageous from that point of view. Let me send it to you, sir, because I know um you're across all of these things. I've got it mainly not for the names, got it mainly for the categories, because I like being able to categorize. Sorry, Dave. [51:30] Dave Pengelley: No, I was gonna say let's wrap up with five minutes of news because [51:33] Richard Webbe: Oh. [51:34] Dave Pengelley: we we like to do that. [51:35] Richard Webbe: Well, [51:35] Dave Pengelley: Yeah. [51:36] Richard Webbe: we know. [51:37] Dave Pengelley: So we didn't get there. [51:44] Dave Pengelley: At the start, are you talking about open weights and open models and open stuff? I don't know if that was on the call or before the call. I lose track [51:50] Richard Webbe: I was [51:51] Dave Pengelley: sometimes. [51:51] Richard Webbe: confused and I was looking to you two to explain to me frontier AI, open AI, [51:56] Dave Pengelley: Yeah. [51:56] Richard Webbe: open weight AI, and what what it all [51:58] Dave Pengelley: Well, [51:58] Richard Webbe: means. [51:59] Dave Pengelley: obviously there's Been a lot in the news this last week. Um, with there being um this whole open letter that was put out around open-weighted uh AIs and supporting them or wanting government to ban them or whatever. And um, most of the big players, your anthropics and Microsoft's and not not anthropic, um meta Microsoft. Uh I think um Nvidia, OpenAI came to the party. Everyone going, Yeah, we should allow open weight models, that's a good thing. But Anthropic have uh come out in some level of opposition against it. They're not a big fan of the open weight model movement. Um and I did like this from Technium, the Hermes guy. We don't support bands on open source models because that wouldn't go far enough. Um [52:44] Matt: They're just a joke. Like, I I don't understand. Like, yeah, they make good models, but then they put them in a cage and gimp them to like the point where they basically don't function anymore and then expect everyone to follow and adopt their own beliefs. Like, I don't understand. Like, we we spoke about Kimi K3 last week. And there's been like literal companies like probably even people within Anthropic, for example, you know, just to go this far, where their own model won't do a thing that they need as part of their frontier work that they're actually doing. It could be security work, it could be, you know, anything. And let's let's say that you know, Claude's putting up the big uh red stop sign and saying, nah, can't do. So then they go to Kimi instead and pipe. Me through something like Open Code, and all of a sudden now they're finding these things that they actually genuinely needed to do to do their work. You know, how does that help anyone? Like there's always going to be the positives and negatives to every like argument. And I feel like I can empathize with the people that are trying to make these decisions, but how do you do it? [53:53] Richard Webbe: You're [53:53] Matt: You know, [53:54] Richard Webbe: a little [53:54] Matt: how [53:54] Richard Webbe: bit damned [53:54] Matt: do you [53:54] Richard Webbe: if you [53:54] Matt: actually [53:55] Richard Webbe: do, and damned if you don't. [53:56] Matt: yeah, exactly? I I feel like there is No answer other than just making it open and encouraging everyone to understand like everything about it. I I think I've used the term or the analogy of crossing a road on [54:11] Richard Webbe: Yeah. [54:11] Matt: this podcast many times, you know, and [54:13] Richard Webbe: Yes. [54:13] Matt: and swimming. Like you teach [54:15] Richard Webbe: Do [54:15] Matt: your [54:15] Richard Webbe: it [54:15] Matt: kids [54:15] Richard Webbe: again. [54:16] Matt: cross a road and you teach [54:17] Richard Webbe: Yeah. [54:17] Matt: them how to swim. You don't make the water any less dangerous. You don't make the cars, you know, any less hard when they hit you. Um [54:26] Richard Webbe: You're [54:27] Matt: but [54:27] Richard Webbe: just watching [54:27] Matt: yeah, [54:27] Richard Webbe: out for what's coming. [54:28] Matt: yeah, it's the edge. Education of like how to actually be literate in this world. [54:34] Dave Pengelley: The um the CTO from First Sell last night, Multe, he was saying uh talking about this this thing they've open sourced called DeepSec, which is a way to write get an agent to run a full security review review across your whole repo. Um, and he did some benchmarking. He came out with did made a deep benchmark or something as well, where he tested different models or their effectiveness versus cost on running a deep sec across your repo. And he said Fable 5 or whatever won't even run it. It just, you know, like you said, it's it's been gimped so hard, it won't even actually do the security review for you because it's too risky to run a security review. Um, dumb. But uh Opus 5 will run it, and it was 10x the cost of [55:20] Dave Pengelley: soul extra high and didn't do as good a job. [55:25] Dave Pengelley: Yeah. Um Grock, uh, I think Kimmy was about a 10. The cost of soul um and did about half as good a job. So it was cheaper than the much, much cheaper, but did half as good a job. But then you know Grok did surprisingly poorly, very cheap, but Groc 4.5 did very poorly on his uh on his benchmark assessment of doing this security review. So yeah, interesting. Um [55:48] Richard Webbe: I'd be interested in the view from from you guys. A lot of the reading I'm doing, as you know, from a technical point of view, I'm catching up to you guys. But I noticed a lot rather than in the old days it was the the Compliance and the government going to the private enterprise saying you've got to fix this, you've got to do this, you've got to do that. I'm now noticing with A, a lot more is it the private enterprise going to, you know, Groc, uh, Anthropic, uh Chat, everyone open AI, going to the government and asking them to put compliance and restrictions and cybersecurity protections in. And it's kind of a uh turning the whole model on its head. [56:29] Matt: Yeah. It is interesting. What are you [56:31] Dave Pengelley: So [56:31] Matt: showing us here, Dave? [56:32] Dave Pengelley: this is the chart I was talking about. So this is you know cost uh of um opus five max and it scored just over 30, whereas 5.6 sol extra high was like significantly cheaper and down and way and actually higher in score. This is the uh deep sec harness thing I was talking about, deep sec bend [56:53] Matt: Yep, yep. [56:53] Dave Pengelley: there. Um so [56:55] Matt: That's [56:55] Dave Pengelley: yeah. [56:56] Matt: cool. [56:56] Richard Webbe: We just had a a comment from Doug, of course. Thanks, Doug. He said AI is not a tool, so don't use that lens. It's an open weights this is intelligence at scale. [57:12] Richard Webbe: Thanks, [57:12] Dave Pengelley: Yeah, it's [57:13] Richard Webbe: interesting [57:13] Dave Pengelley: it's [57:13] Richard Webbe: viewing. [57:14] Dave Pengelley: it's it's uh intelligence is a tough word as well, like it's formatted knowledge. Um, because intelligence is how you do it. And we we have the reasoning layers with these models and and they do reason. And so we're sort of simulating intelligence. I mean, we call it artificial intelligence. It's artificial, right? That's the whole point of it. Um, that is what we're discussed. Like, it doesn't have that ambient context and knowledge to know so many things. Like, I was trying to get it, like, I was trying to get it to help me ideate some um some joke memes around like techniums post around we don't support bands on open source models because they don't go far enough. I was like, ah, that feels like a kind of like tongue-in-cheek joke around, you know, the the the underlying menace of we don't support this thing because you know, we want something much, much worse. And just Chat GPT just on the web chat was woeful. It just sort of gave me completely different joke setups. And like, no, no, that's not the setup. The setup is this, and then the implied menace. And then it was like just doing direct menace. I was like, that's not it. You don't get it. Like it was funny. Like it's like that simple concept of [58:17] Richard Webbe: IOI [58:17] Dave Pengelley: this humorous [58:18] Richard Webbe: doesn't really get [58:18] Dave Pengelley: setup. [58:19] Richard Webbe: yeah, I doesn't really get irony. [58:22] Dave Pengelley: Yeah. [58:22] Matt: Which is kind [58:22] Dave Pengelley: So [58:22] Matt: of [58:23] Dave Pengelley: anyway. [58:23] Matt: right. [58:23] Richard Webbe: Yeah, it's totally ironic. [58:26] Dave Pengelley: Um, Matt, was there any uh you you sent through some great. News points. Um, we will smash through some of those quickly. GPT's got some new live transcription things. [58:38] Dave Pengelley: That's [58:39] Matt: That's yep, [58:39] Dave Pengelley: um [58:39] Matt: it's enough said. [58:41] Dave Pengelley: yeah. [58:46] Dave Pengelley: Um, if I can't run a locally, I don't care. I'll keep my cheap uh free parakeet model or whatever I'm using for doing the uh podcast transcripts for now. But uh yeah, I think that'll well we talked about voice um receptionists. I think new fast. Faster better models will help make those calls more streamlined. That'll also mean you'll get like more um dodgy spam calls that are gonna be more convincing, [59:11] Matt: Yeah, [59:11] Dave Pengelley: which [59:11] Matt: like seriously, the the the whatever GPT is doing with their live stuff, as in, you know, when you go to the Chat GPT app on your mobile or [59:20] Dave Pengelley: yeah. [59:20] Matt: on the desktop, whatever, their new live mode is is so like good. It's very, very convincing. But I think running that through a voice agent. doesn't really have the economics of scale [59:33] Dave Pengelley: Yeah. [59:34] Matt: right at this point. It's still very expensive to run those. But yeah, yeah, well, as you said, you know, if we can have those at a fairly decent cost, um it's [59:44] Dave Pengelley: Yep. [59:44] Matt: very convincing. It's [59:45] Dave Pengelley: Let's [59:45] Matt: very [59:45] Dave Pengelley: see [59:45] Matt: good. [59:45] Dave Pengelley: the old Star Trek computer. Uh um yeah, I I like the live mode and I used it again the other day when I was out for a walk and I was I in five point five I think they'd completely stuffed it and it was just monitor. Monotone and just rubbish, but they brought the tone back into the voices and it's much more friendly to talk to again. Um, Doug. [01:00:08] Richard Webbe: What do you like about the live mode, Dave? [01:00:11] Dave Pengelley: Hang on, we've got a comment from Doug. The context is the next step. We're all looking at inference to save token costs, but then we find every company needs to build its own living ontology. The what's where and why the world's been used. Example HR legal healthcare use the same language and words. Yeah, I mean very, very different things at times. Not to disagree, just adding to help. Great chat. Yeah, thanks, Doug. I think. Yeah, context is a thing. And like, like I talk about, and I use that term ambient context, humans, we just know so much stuff, and we can adapt and understand that you know one word can mean different things depending on if you're standing in a hospital or a police station or if you're doing software engineering. Um the AIs don't always know that in like [01:00:51] Richard Webbe: We have [01:00:52] Dave Pengelley: intuitively. [01:00:52] Richard Webbe: a little we have a little alarm bell in our brain that we sometimes ignore, but we all use. And remember, I think it was Socrates. He said the definition of intelligence is someone who knows how much they don't know. AI rarely knows how much it doesn't know. It just reacts on what it does know and assuming that that's the boundary it must be working in. And that can be a dangerous context sometimes. [01:01:14] Dave Pengelley: Yeah. It's especially and and Matt, you talk about the s sick of fancy all the time. Like with OpenAI today, I was, you know, talking about this operator stories concept for a podcast. I'm thinking about, you know, names for a new podcast and what I could call these different things. And I'd say, I'm thinking something like this. And go okay, but be careful of that. And be okay, run a collision check. Are there other brands using that thing and go this? I'm like, oh, actually, that makes me think, could we do this? And he goes, Oh, yeah, that that's actually a much better direction to go on for all these reasons. Blah blah blah. I'm like, yeah, actually, that's not working for me. What about this one? Oh, indeed, that that's probably the best direction yet that you should go on. Blah blah blah blah. Like that that [01:01:49] Matt: There's never friction. There's never [01:01:51] Dave Pengelley: over-agreeableness, [01:01:51] Matt: like [01:01:52] Richard Webbe: Yeah, [01:01:52] Matt: Yeah. [01:01:52] Richard Webbe: yeah, good point. I like that [01:01:53] Dave Pengelley: like [01:01:54] Richard Webbe: there's no friction. [01:01:55] Matt: Uh [01:01:55] Dave Pengelley: um, [01:01:55] Matt: it could [01:01:55] Dave Pengelley: like [01:01:56] Matt: I I can't remember if this was before the show. Richard, you said something about like the the whole means of progression is through disagreement and you literally do not progress. [01:02:11] Dave Pengelley: But to be to be fair, it did say I still think this m particular title is better than those ones for these reasons. Like it didn't completely just bow down. But you know, if I fundamentally pivoted it'd go, that's a great pivot and that's great for these reasons. Like it was very, very yes man, very um affirmative. But yeah. [01:02:30] Matt: Comment [01:02:32] Matt: I was just gonna tie that back to what Doug said in his comment. It's the same thing, you know, it's the the the idea of curating and controlling that context layer and actually having sort of second order thinking, you know, it's not just you trying to answer instantly. Sometimes you may need to clarify, you know, is this HR legal or healthcare? You know, how do you know? And then what is that ambient context to allow that so that you choose [01:02:56] Richard Webbe: Yeah. [01:02:57] Matt: that right language in the right situation? [01:02:59] Dave Pengelley: I I completely I completely stuffed up an interview once a few years back where they asked me a question and I didn't establish context. I thought it was an individual compute contributor role, not a leadership role. And so all their questions I answered around my capability to execute, to get things done, to to work through things and do other bits and pieces. And my feedback was, yeah, we actually want someone who's a bit more of a leader for this role. I'm like, shut up. Oh, one question at the start of the interview around like what is the scope of this role? And if I'd known it was a leadership role. I would have pivoted on my decade plus of people leadership and talked about my capacity to manage and lead, but I didn't. And so all they heard was one half of my story that I thought they wanted to hear because I didn't have the right context. [01:03:45] Richard Webbe: Well, here's the funny thing, and it's always the same. You feel like you stuffed up the interview. I'd say they stuffed up the interview. I mean, we know you're a good leader, and anyone who knows you knows that. So the funny thing is the fact that they asked you a question and didn't publish the context well. Well enough so that you answered as though it was an individual contributor role. They're the ones that missed out, not you. Do you know what I mean? We tend [01:04:06] Dave Pengelley: Yeah. [01:04:07] Richard Webbe: to be very hard on ourselves with that sort of [01:04:09] Dave Pengelley: Tell [01:04:09] Richard Webbe: lens. [01:04:09] Dave Pengelley: it to my mortgage repayments that are they're the ones that missed out. Anyway. Um back back to the quick news of the the the lightning news that we're going through because we're well over time. There's a new MCP standard coming out. This sounds exciting. Um [01:04:25] Matt: It's it's it's boring, but it's very good because if you guys recall or if any of our long-term listeners recall like I was super anti-MCP at the beginning. I'm like, just it should just be a CLI tool. Like MCP is just like a dumbed down abstraction of the CLI tool. But it's getting very good now. And the the biggest issues with MCP in the past was it was just crammed. Here's here's Doug's uh comment again, it was just crammed into context. So you know everything was shoved in there. So you would have all of your HR legal on healthcare all shoved in there as use MCP definitions. But now it's very much lazy loaded. Like skills, where [01:05:01] Dave Pengelley: Yeah. [01:05:01] Matt: you know you have an index and you only load the very the relevant context when that situation requires it. So I don't know, it's a cool thing. I think MCP is going in the right direction. And I've built some stuff recently with the new MCP protocols, not this one, the previous one. And it was actually a delight. And it was quite nice to see what's possible now. And it's kind of changing my opinion a little bit. [01:05:22] Richard Webbe: What [01:05:22] Dave Pengelley: Yeah, I think [01:05:22] Richard Webbe: are [01:05:23] Dave Pengelley: I [01:05:23] Richard Webbe: the [01:05:23] Dave Pengelley: think [01:05:23] Richard Webbe: parameters [01:05:23] Dave Pengelley: some [01:05:24] Richard Webbe: in [01:05:24] Dave Pengelley: sorry. [01:05:24] Richard Webbe: MCP that make it good, Matt? What are the parameters? [01:05:27] Matt: The parameters are just it's it's like It's it's giving the tool with uh a narrowed perspective of what the tool is. So what [01:05:37] Richard Webbe: Some [01:05:37] Matt: the your [01:05:37] Richard Webbe: content. [01:05:38] Matt: question is actually the answer, if that [01:05:40] Richard Webbe: Gotcha. [01:05:40] Matt: makes sense. So [01:05:41] Richard Webbe: Gotcha. [01:05:41] Matt: if you if you think of a CLI tool, you can explore the entire tree of commands and you'll get access to all of the things. But if you have an MCP that says do this, you'll have a field that gives you specific um criteria what to put in that field. And like [01:05:58] Richard Webbe: Yep. [01:05:58] Matt: that's it. [01:05:58] Richard Webbe: Yep. [01:05:58] Matt: It's kind of like entering [01:05:59] Richard Webbe: I [01:05:59] Matt: a [01:05:59] Richard Webbe: get [01:05:59] Matt: form. [01:05:59] Richard Webbe: it. [01:06:00] Matt: Versus just free form typing. [01:06:03] Richard Webbe: I get it. Thank you. Jeez, sorry, Dave, you're showing something. [01:06:07] Dave Pengelley: Yes, yes. One of the issues some of the MCPs have is that authentic authentication piece. And it looks like they're tightening up the handshaking and stuff. Like I know, like running local agents, I want to connect them to something like post hog. Post hog fails because it wants to redirect back to a local host endpoint that doesn't exist and it is completely just can't do it. And so part of the new protocol and the new handshake things are making it easier, I think, for Connecting your local hosted applications and agents into different MCP servers and stuff as well. So lots of good developments going on with some of that sort of stuff to smooth [01:06:38] Matt: I actually [01:06:38] Dave Pengelley: out those [01:06:39] Matt: built [01:06:39] Dave Pengelley: things. [01:06:39] Matt: one like this this may not make much sense, but I'll do it really quick. I built one for for my core service because it's gonna eventually be a public service. And to expose things to your agent, so all you have to do is give them the MCP endpoint, the agent checks it out and goes, okay, this is how we register for it, and the whole thing is fully understood from start to finish. With [01:07:00] Richard Webbe: So [01:07:00] Matt: that [01:07:00] Richard Webbe: it [01:07:00] Matt: authentication. [01:07:00] Richard Webbe: sets up it sets up the syntax and the API syntax so everyone knows what language they're talking. [01:07:05] Matt: Even the connection in the first place, like what Dave said about authentication. So, you know, instead of having a plaintext key that you've got to shove somewhere, like an API key, it can initiate the OAuth sequence for you, open it up in the browser, and then you [01:07:17] Richard Webbe: Yep. [01:07:17] Matt: go and just log in. [01:07:18] Richard Webbe: Yep. [01:07:18] Dave Pengelley: Yeah, that that's what I got to set up set up with chart reporter. It's always login so when p people want to connect and talk to their trades, it pops up a little browser window, logs into their chart reporter, authenticates, pushes back, and then you're in. It's awesome. [01:07:31] Dave Pengelley: So good. [01:07:32] Richard Webbe: Yeah, [01:07:32] Dave Pengelley: Um [01:07:32] Richard Webbe: yeah, yeah. [01:07:34] Dave Pengelley: big because uh and you shared this one as well. We won't play the the 30-second clip map, but this guy's making the point um that API keys used to be a developer thing, and developers knew to be very careful with them and what they could do and what they couldn't do. But now every man and their dog who's using any kind of AI, the AI is just saying, Yeah, go go to this developer settings panel and give me an API key. And they're not necessarily aware of the risks of what an API key is possible of doing to that platform. But also, you know, just the leakage capacity of these LLMs, if you're just just chucking it straight into your uh your chat, um is a problem. [01:08:10] Richard Webbe: Hey guys, I have to go. [01:08:12] Dave Pengelley: We all have to go, Richard. [01:08:14] Richard Webbe: It's [01:08:14] Dave Pengelley: We all have [01:08:14] Richard Webbe: so [01:08:14] Dave Pengelley: to go. [01:08:14] Richard Webbe: sad. [01:08:15] Dave Pengelley: There's some other news and things we can go through, but we we're out of time. We're way over time. So thank you to all that have uh held on for our extended episode today. [01:08:22] Richard Webbe: Thanks for everyone [01:08:23] Dave Pengelley: We [01:08:23] Richard Webbe: in [01:08:23] Dave Pengelley: appreciate [01:08:23] Richard Webbe: the chat, including [01:08:24] Dave Pengelley: you. [01:08:24] Richard Webbe: Doug, who's still referring to IDE agents and other things. Well done, mate. [01:08:28] Dave Pengelley: Yeah, no, thank you. You uh thank you, Doug. Yeah, IDE agents like cursor and cursor three is coming out. Yep. Getting that sort of getting model agnostic on your uh development environments. That's why I use Hermes because it's model agnostic and I can do a lot with that. Um I have been in core codecs trying to do some stuff and I've been just completely slaying my quota day after day. It's killing me. Anyway, um we're out of time. Thank you, Richard. Well let's [01:08:53] Richard Webbe: Have [01:08:54] Dave Pengelley: drop [01:08:54] Richard Webbe: a [01:08:54] Dave Pengelley: off, [01:08:54] Richard Webbe: great [01:08:54] Dave Pengelley: man. [01:08:54] Richard Webbe: week. [01:08:54] Dave Pengelley: I will say goodbye. [01:08:55] Richard Webbe: Stay warm. [01:08:57] Dave Pengelley: Yeah, thanks, mate. Um If you are new or visiting to the show, if you haven't already hit the like and the subscribe button, I never do the plug stuff, Matt. I never tell people, I never do the reminder. You've got to remind people or they won't do it. And I never ever do it. So I'm doing it. I'm doing it. Um, please like and subscribe. If you're on LinkedIn, uh, I'm streaming this on my own personal LinkedIn, but we do have the new AI operators page. If we can get some support behind that, we'll be able to live stream straight to the AI operators page on LinkedIn as well. All these platforms, other than YouTube, want you to have a certain audience space before you can go live. So we want to get there with LinkedIn. So please follow us there as well. [01:09:40] Dave Pengelley: That's it. Like and subscribe. [01:09:42] Matt: And [01:09:42] Dave Pengelley: Drop [01:09:42] Matt: share. [01:09:42] Dave Pengelley: some comments. [01:09:43] Matt: Honestly, [01:09:43] Dave Pengelley: Share. [01:09:44] Matt: sharing, sharing to someone and saying, like, hey, you probably should check this out. Like sometimes you guys, whoever listens to this and watches this, maybe it doesn't all connect and click with you, but you probably know someone who this is for. And that kind of sharing action, algorithm-wise, is actually Probably the best way you can support [01:10:01] Dave Pengelley: And [01:10:01] Matt: us. [01:10:01] Dave Pengelley: if some of our content confuses you, well done. Well done for sticking through it. [01:10:05] Matt: Yeah, definitely. [01:10:07] Dave Pengelley: Um yeah. So uh yeah, thank you everyone. Thank you. We we love our our watches. Um it's good to see that happening. So yeah, please subscribe and comment. Comments are another good way of showing engagement, letting the algorithm know. So even if it's uh thanks for the great job you're doing, we'll take that. If it's specific questions or things you want to talk us to talk about uh in a future segment, let us know there as well. But um Thanks, Matt. Appreciate it. [01:10:32] Matt: Thank you, [01:10:32] Dave Pengelley: Thank [01:10:32] Matt: man. [01:10:32] Dave Pengelley: you. Thank you, all our viewers everywhere. We appreciate you. Look out for the comic summary version of this coming out on LinkedIn soon. Now that I'm getting that, I'm fine. I'm nearly there, Matt. I'm nearly getting the pipeline streamed through, like to generate all the transcripts and process through and clip them together and make stories and make comics. Um, and then make clips. Still debugging it a little bit. That's why we haven't seen much clips yet, but we're getting there. [01:10:58] Matt: It's very cool. They're looking very good. [01:11:00] Dave Pengelley: Getting there. Getting there. Anyway. All right. Thank you all. We'll see you next week on the AI operators.