Jarrad Grigg

All right, Brian, good to have you on the show. How you doing?

Brian Foody

Good man, it’s good to be here. It’s my first first video podcast. I’ve done other podcasts, but yeah, just audio. Yeah, let’s see how we go.

Jarrad Grigg

Yeah. There go. Yeah. It was a first for everything. So yeah, I appreciate you you coming on. and for those that don’t know you, you have quite an illustrious history of making a lot of things. I think I counted like seven odd businesses and ideas that you were a founder of directly, and then others that you worked for, and you know it’s it’s ranged from all sorts of types of businesses.

Brian Foody

Mm-hmm.

Jarrad Grigg

And so I guess to start with, I’d I’d love for you to sort of tell viewers your history I guess and and what you’ve been involved in over the years. You were doing sort of AR products, you’ve done crypto products, you’ve done digital storybooks, all sorts of stuff. So run us through how you got to where you are today.

Brian Foody

Yeah. Yeah, okay. yeah, it’s good to be on. Thanks for having me, Mate. loving what you’re doing with the podcast. so yeah, thanks for having me. Appreciate it. so yeah, where would I come from? So I started out sort of classic, just working in a financial bank in Dublin, doing like tech, and then I was yeah, I realized I wasn’t really enjoying that and I wanted to like create things. Like I always got into technology. I did a master’s degree in security and forensic computing.

Jarrad Grigg

Thanks.

Brian Foody

And then I realized I wasn’t going to build things with that. I was just going to audit. So I got I went to I didn’t go into that career. so I went to London eventually and worked in a marketing company and we were building things and I really liked that. And from there I decided any full time job I did, I just wasn’t getting enough time to create things. So I went out, sort of what you’re doing now, I went out consulting, building apps for people, building websites, helped companies like one was in telecare. They’re like they put aged care devices in people’s homes. And so building all the infrastructure for them. so yeah, it was really nice, like helping people get to the point where they could like they listed on the ASX and stuff. then yeah, I guess where you would have seen me was Bamboo was another company that I came across when I was consulting. They had just raised money. and so I came on to help them out and then I ended up becoming sort of the lead engineer slash CTO there and getting that product live. so more transitioning into sort of like a co founder role there and we scaled that up. Which is really amazing. was like building a bank from the ground up. and yeah, it gave me a real appreciation. Like in building a bank, we had to like we had to spend a lot of time building the fundamentals really well. But once those were in place, we could really accelerate. and what was really interesting at Bamboo from where we are now is I had to run customer support because we were running out of funding. So like we had no one to run customer support. So I had to run customer support and I was doing the engineering. But it was actually really quite useful because

Jarrad Grigg

Mm-hmm.

Brian Foody

I created my own CLI before the Claude C L I has come. I had like I I created like it was called the Bam the Bamboo CLI. So when I do a customer support ticket, I would like add some code and I’d add it to the CLI. So like pretty quickly I was able to fix things. Things I wasn’t able to fix were product, I was able to put in the CLI. so the it never became like a cost center. Yeah, so now it’s like full com I remember saying I think the whole business in these CLIs. And then like it was like now we see it now where like people are running businesses through them, which is quite interesting.

Jarrad Grigg

nice. Right. Mm.

Brian Foody

yeah, and from there when we when we exited bamboo, I went to work for a company called Austel, you the shipbuilder. So they were doing yeah, they were doing smart ships. So they wanted to make the vessels smart, so they wanted to extend the sort of life cycle past the selling of a ship. And so they were building MarineLink Smart, which is really like for Perth it was a really amazing product. It was deploying, you were tapping into all the IoT sort of readings from the vessel.

Jarrad Grigg

Yeah, yep.

Brian Foody

And then optimizing it so it reduced the fuel usage. so that was a really interesting project working with the team at UWA. so I did that for about a year and a half. Partly it was just because it was in UWA and it was like a nine-day fortnight. It was a nice transition. but yeah, I mean it was intense. It was like a lot of that was the early days were more like DevOps. I really got into the DevOps stuff and optimizing things. So when they when I joined, they took like four weeks to get a vessel hooked up, and by the end, we could do it in like a day.

Jarrad Grigg

Nice.

Brian Foody

deploy it onto a vessel, like making it all end to end. and again, it was like once the data was coming in, we were able to like grow the product pretty quickly. since then, once my kids were born, I did what you what you’ve done now, going out on your own. So I’ve been consulting again for the last few years, mostly with ASX companies. So kind of bringing that, you know a lot of places where I go, technology is like viewed as like they don’t really like it and everyone’s afraid of it. So trying to bring that

Jarrad Grigg

Mm-hmm.

Brian Foody

mindset where like technology is in Australia we don’t really have that where technology is viewed as something that can like I don’t know be enjoyable and improve your business. It just feels like something like a chore for a lot of businesses. It’s like, we’ve got to do this tech thing and it’s gonna be a pain. so I’ve really enjoyed that. But over the last few years I’ve got the itch to go back to startup. so now we’re building out vertical AI. we’ve focused on voice AI as our sort of main main wedge in, I suppose.

Jarrad Grigg

Mm-hmm. Mm-hmm. Yeah.

Brian Foody

And I haven’t realized how hard voice AI was gonna be. So it’s funny, I was I was sending you all the products that I’ve built. So when Jen when Jen AI came out, I think since GPT two, me and my friend Linus have been building. so I I got a load of products I released. We’ll probably go through them. But like the last year I haven’t released anything because we’ve just been working on this problem of voice AI, which has been like it’s such a deceptive problem. It’s almost like it’s like a smaller case of self driving. You know, where Musk said we’re gonna solve self driving next year.

Jarrad Grigg

Mm.

Brian Foody

like fifteen years ago. Like vo yeah, voice is kinda like that. You get to ninety percent really quickly and then the last ten percent is just like a brutal grind. Like it’s definitely the hardest thing I’ve built. Like so much harder than building an investment app or a bank. It’s just yeah, the non determinism. Yeah, yeah.

Jarrad Grigg

Every year, yeah. Yeah. I’d I’d love I’d love to touch on that. Yeah, i voice response and the latency and the overall experience. I want to talk about that for sure. I saw your the video that you posted on LinkedIn and it was super impressive, so I wanna cover that. Just before we continue, your video feed is frozen on my end. I hope it’s still working on yours, so I might I don’t know if it’s

Brian Foody

Yeah. it is, yeah. let me see.

Jarrad Grigg

if it’s gonna s record like that on your end, so if if so, we might have to start again. I’ve got a I’ve got a nice smile if you could, yeah. I might

Brian Foody

Okay. Do you want me to try rejoining?

Jarrad Grigg

got ya. So I’m I’m worried that that entire intro is just going to be a single frame.

Brian Foody

Yeah Do you want me to just do a quick recap a lot quicker, like a minute? I can do a much shorter intro. That was a bit long-winded, anyway. Yeah.

Jarrad Grigg

Yeah, let’s just do that. So No worries. Alright, so yeah, fill me in on on your career so far. how’d you get to where you are now with that with vertical AR?

Brian Foody

Yeah, cool. I’ll do a very quick recap, suppose. Kind of classic software engineering background and then realized I didn’t want to be a guy sitting there just taking tickets and doing it. I liked front end, I liked back end, I liked creating from the ground up. So I kind went from software engineer to marketing companies building. You know the way marketing companies are, you have to build really quickly because the client has a request. So I really enjoyed that. Then the next natural transition was freelancing, going out to companies.

Jarrad Grigg

Mm-hmm.

Brian Foody

So working with sort of early stage startups to get them going. And from there moved into actually building startups. So working for Bamboo or to kind of lead engineers slash CTO. So we built a bank from the ground up, which is really nice. mean, the more, I know if you find this, but the more you build yourself, the more you realize you can build. Like you just realize there’s so much ceremony around things. Like you’d think, oh, I could never build a bank. And then you basically build one for an investment app. And like, yeah, it just makes you like.

Jarrad Grigg

Very cool. Yeah.

Brian Foody

I don’t know, your confidence grows in what you can achieve.

Jarrad Grigg

Yeah, yeah. It’s an impressive product too. I I I quite liked bamboo when it came out. and I I didn’t realise you were the CTO, so it’s it’s you did well there, I thought, yeah.

Brian Foody

Yep. Yeah. Yeah. And to be fair, the guys who I didn’t I didn’t think of the idea came on, but like it was a really good idea and that like everyone was throwing their money at meme coins rather than viewing it as something where like if you can put a bit of money into it over the long term. I think last year there was 40,000 investors and nearly all of them were in the green. So like it’s nice to have created something that went on like in the background, it’s still running and it’s like, you know, helping people. So yeah.

Jarrad Grigg

Right. Yeah, yeah.

Brian Foody

Yeah, think Bamboo is definitely one of those where we had to a lot of groundwork in, but then it just ran. So then I went, to be honest, after Bamboo, like I did lead engineer role at Austal on an AI learning project on ships, but then I really got into like indie hacking. So like shipping products fast. And so we go through a few that I did, but like, it’s sort of this balance Bamboo was like taking a long time to build something and then it just ran and it ran really well and we never had any bugs. And then you have this other mindset, which is like ship really fast and then like.

Jarrad Grigg

Mm-hmm. Yep.

Brian Foody

A lot the time you’re doing a lot of re-work. So I’m still sort of like between those two camps. Like there’s a sort of a balance between both where you do product discovery. But yeah, I do see a lot of people struggle with, especially in AI where they ship quick, but then they’re, it’s just designed terribly. And then they end up doing it, creating a lot of bugs and stuff. So like there’s a sort of balance there, I think that you have to find. Yeah. So yeah, since then we’ve moved on to vertical AI, which I thought was going to be one of these things that we shipped quickly because it was voice AI.

Jarrad Grigg

Mm-hmm. Right.

Brian Foody

We got to like 90 % very quickly, but getting the last 10 % has been a real grind. It’s like I didn’t realize how hard of a problem we had chewed up. So you’ll see my feature, I was building a lot of products and now the last year we haven’t released anything because we’ve been deep in the weeds of voice AI where yeah, it’s like full self driving. Like Elon said, it’d be a year, it’s 15 years later. I said we’d have something a month ago, it’s like a year later and we’re finally nearly there.

Jarrad Grigg

Mm. Mm. Yeah. Yeah.

Brian Foody

Yeah, it’s definitely one of those problems that is deceptive.

Jarrad Grigg

Yeah, I definitely want to touch on that. thanks for the intro. That’s there’s a whole heap there. and so I think for people watching, you are you’re more of an engineer than a designer, but you strike me as someone that has a product mindset, right? So you’re not someone that just sort of takes a spec and builds it. seem very outcomes based. But so what is the goal we’re trying to achieve and you just kinda get after it. is that is that a fair assessment? Like you you’re definitely much more technical than you are on the design sense.

Brian Foody

Yeah, yeah, no, I’m definitely more of an engineer. Sometimes I feel like for the AI, I’m nearly too much of an engineer. wish I was less, because you know with AI, you’re like, maybe we need to down, because engineering-wise, where sometimes the person who doesn’t think like that can get some better outcomes in the AI era. They’re not limited by, I guess, what they don’t know. So yeah, but definitely, yeah, I’m not good at design systems. I would like someone who can create a design system, I think, is really powerful now, because it’s like,

Jarrad Grigg

Yeah. Yep, yep. Yeah, yeah. Yeah.

Brian Foody

Yeah, that’s something I would invest in, like, because don’t have that skill. But I’m pretty good on UX and product thinking and the outcome. yeah. But I do need, like you’ve seen, I partnered with Linus a lot in the past, who worked with me at Bamboo. So having someone who could, yeah. Yeah, I’ll have to send him this. So yeah, he’s doing well. But yeah, someone like you or Linus who has that design mindset and continuity, yeah, I partner well with. Yeah.

Jarrad Grigg

Yeah. Shout out to Linus. He’s he’s great. He’s doing well. Yep. Yeah. I find it I find yeah, it’s absolutely true, right? Is is if I I’ve definitely built a lot of things, but I’m not an engineer at your level. and that sort of distance from the code base has sort of helped me a little bit. and probably from the other side as well, right? Like you you know how things should be built, but from a design standpoint it’s that’s probably okay. a and you kinda move on, right? So you definitely need the expertise of of the other,

Brian Foody

Mm. Yeah, yeah,

Jarrad Grigg

And it’s also I I think you raised something before in Utro, which is great, which is you know at bamboo you had to really move slow and get it right and then watch it scale, whereas this indie hacking thing is all just about shipping super quick and break and that’s an inch it’s a good tension, right? Like so what what have you found works best for you and and and in this sort of shipping fast mindset, have these tools helped? Like how do you go about you know shipping really quickly?

Brian Foody

Yeah. Yeah. and Yeah, yeah. I mean, I can ship really quickly. guess where I, when I ship, I like to like add features. So I’m not a very good at saying like, this is the simple product I’ve built. So I always like to think kind of, I probably think a little too big sometimes and I could do it having the simpler products. But yeah, when we built bamboo, so does this technology called event sourcing or a mindset of building technology? like say when you look at your bank balance, you’re not looking at, say your balance says $1,000.

Jarrad Grigg

Yeah.

Brian Foody

There’s one way of building technology where that $1,000 is like a field in a database. And so when you withdraw money, you update that from 1,000 to 900. But the problem with that is if it ends up in the wrong state, you’re like, how did it get here? Like, this agent update it? Did this person withdraw it? So how we built Bamboo is a thing called event sourcing. So the balance of 1,000 is just a replay of all the events. So I deposited 500. I withdrew 100. I deposited 600. So everything is always like, you’ve got your event store and then you’ve got like what are called read projections. So whenever anyone would call up with a support ticket and say, my Bitcoin has disappeared. I’d like pull up their account and look at all the events and then I’d be like, I’d be able to see exactly what happened. And so like that was very satisfying. Like it was hard. You have to really map things out upfront because like.

Jarrad Grigg

Mm-hmm. Hm.

Brian Foody

the event comes in and then a process takes us and then the process might submit another event and then you end up seeing your balance on the screen. But I actually think it’s probably like a really good way to build right now because the code is not actually that important, right? It’s actually like the understanding of the system. So I’m coming back to event sourcing again now. When I did the indie hack and I didn’t think at all like that, I was like, okay, I can build this quick. I get in the tools, I start going. But like, I didn’t know I was in the system at the end of it. Like I didn’t have the spec.

Jarrad Grigg

Mm-hmm. Mm, mm-hmm.

Brian Foody

really, so I would find things could break pretty quickly on me. So I’ve gone from that sort of really strict engineering. I still did that strict engineering when I was doing consulting, but on the Indy I didn’t. And I did find I kind of missed that clear understanding of the system. So now I’m going back.

Jarrad Grigg

Yeah. Yeah. Yeah, but I guess the the goal the the goal’s different, right? Like you you’re just trying to get traction. You’re trying to get find something that people like and then you can fix it later.

Brian Foody

Yeah, exactly. But then there’s like, do you ever get to fix it? That’s the other side to it. But yeah, you’re definitely right. There’s a balance. It’s so good having this perfectly engineered product that no one likes. So yeah, there’s the balance there.

Jarrad Grigg

Yeah. Yeah, yeah. I I I’m of the mindset of like you said is y you have to understand the system. So I’ll spend a lot of time making sure that I have clear requirements and what done looks like. I don’t worry so much about the in-between, but I definitely review I I use MOMO diagrams a lot. So I’m always asking my agents, ex show me visually what what what does the system look like? So you can have the parts of the system in your head, but

Brian Foody

Mmm. Yeah, yeah. Yeah, yeah.

Jarrad Grigg

Yeah, I’m but I’m not I’m not looking at the code specifically, right? So that’s my way of working fast.

Brian Foody

Yeah, because you’re more of a designer using these tools to build products. So yeah, it’s quite interesting. We probably come at things differently now. If I was to build a new product, I would probably build the API first. And then I’d generate the UI. I’d say, hey, Claude, here’s the API. Give me five versions of this product. And then I’d go from there. Whereas you’d probably come from the other way at it. You’d be like, hey, Claude, this is the design. Now give me the code.

Jarrad Grigg

Yeah. Yeah. Yep. Yep.

Brian Foody

It’s quite interesting.

Jarrad Grigg

Interestingly, interest I I I did start like that, but the more I lent into these tools and and the way that they they work well is I’ve shifted dramatically to be requirements first. and so I I my harness is kind of like a grill me session where I I come up with an idea and we we go back and forth a lot on how it would work. and it’s not about UI.

Brian Foody

Yeah. Yeah. Yeah, okay. Yeah. Yeah, okay.

Jarrad Grigg

Once once I’ve got okay, I can understand that this there’s value here and this will work, and we we generate the requirements, part of that step is to create it’ll do a terminal UI first in a markdown file. So it’ll just generate a rough look of it. And I’ll say, okay, the the flow feels right there, so I’m not, you know, detailing it. And then when I’m h happy with that, I’ll actually get it to generate the design with a rough design system first, in pencil.

Brian Foody

Okay, yep. Yeah.

Jarrad Grigg

So it it’ll just spin up the the UI there and I’ll review it. Now generally that’s where most ideas for me die because the the design capability of models is I I still think awful. And I get to that point and think, okay, this is gonna take me a lot of work to reflight to refine the experience. and I’ll just park it for now, right? but if it’s worth it, I’ll I’ll iterate on that. And so nothing’s been built yet, and I do I do spend a lot of time on that phase to your point, right? Is is for me it’s

Brian Foody

yeah, nice. Hahaha Hmm. Yeah.

Jarrad Grigg

I get the requirements right, but then I spend a lot of time on designing experience and then when I’m satisfied then I say, okay, now build it. so yeah, it works for me. but definitely the I I think it’s okay. I’m not gonna say it’s great. I I definitely feel it’s the best of the sort of AI native applications and models seem to be able to use it quite well.

Brian Foody

Yeah, interesting process. How do you find pencil?

Jarrad Grigg

a lot better than Figma. the Figma MCP is a a challenge to use. I don’t find it super reliable. But I do have the new Figma agent. I got access to that and it’s not too bad. it’s good at making broad changes. So there’s mechanical changes where it’s like update this variant, you know, change this aspect of the design system. But it’s not creative. None of these tooling toolings yet I don’t feel have really grocked that.

Brian Foody

Yeah, yeah. Yeah, okay.

Jarrad Grigg

how to design a good UX UI.

Brian Foody

So remind me, what’s the workflow with pencil? Is it like getting Claude to generate the designs or is it getting Claude to take the designs or is it both? Okay.

Jarrad Grigg

It can do it can it can do both, yeah. So it just connects to the file and you can work in pencil with an agent on canvas where you can choose your model or it connects to your active session. So you can use codecs, you can use clawed code, whatever. and as long as you’ve got the file open, it will be able to draw on the canvas and read from and everything like that.

Brian Foody

Okay. culture. Okay.

Jarrad Grigg

But we’re the advantage of the tool like that over Figma, which in my view is a massive advantage, is you can feed it context. Because you’re working in your IDE, you would have the code base, right? So you have all your requirements. So I don’t really build anything in unless I’ve got like r really strong PRDs. So when it’s designing something, it’s not just kind of making something up out of thin air. It has it needs to meet the requirements. So it’s got more context. Whereas the Figma agent I found

Brian Foody

Yeah, okay. Yeah, okay. Yeah.

Jarrad Grigg

doesn’t. It still just tries to design things. Like I got it to make a dark mode of of a particular UI just for experiment and the results were just awful. So I think yeah, I think we’re still

Brian Foody

Yeah, You find our design is little like writing, isn’t it? gives you some like, the first thing looks good, but like over time, it doesn’t have that continuity. Like as you read like long AI prose, you’re like, wait a second, this is crap. But it takes you a while to figure it out. Yeah, yeah, yeah.

Jarrad Grigg

Yeah. Yeah, yeah. It says a lot without saying anything. Yeah. And the the UX is all and also if you don’t really know it’s like you with probably the way that you would build a system. If you don’t really know what you’re doing, you kind of accept what it spits out at face value. And everything starts to look the same. I see it everywhere. So I think the benefit of these tools is that anyone can now build.

Brian Foody

Yeah.

Jarrad Grigg

But the benefit that the flip side is that anyone can build. So there is a a a high volume of output, you know, and not a lot of I think, you know, rich quality. So, you know, it’s you see a lot of apps getting shipped to the store, but I don’t know how successful they are who’s shipping real products that people are using. So and y you got a lot of experience in that. So I kinda wanna touch on that a little bit. Yeah. It’s like the way that you

Brian Foody

Yeah. Yeah. Yeah, yeah, yeah. Yeah. Mixed success is part of the indie game. it’s what a B2C product is like 1 % chance. So yeah, whereas a B2B is five to 10%. So yeah, it’s like, yeah, there’s a lot of failure.

Jarrad Grigg

Yep. Yep. Yeah, but I mean you’ve done a lot of it, right? So there’s some key lessons that you’ve learned in in in in what to do and and what to how to approach creating products from scratch and making them work. So what are some of those lessons you’ve learned over over all the products you’ve made and how have you kind of rolled that into vertical AO?

Brian Foody

Mm-hmm. Yeah, it’s a good question. think the main thing I learned was like bedtime story was an interesting one because it’s sort of like the hello world now of generative AI apps. It’s like generating a story, a kid’s story. It’s so simple to do. when we kicked it off, it was just at the start where AI image generation came out. So was GPT-2 and the first version of chat GPT image generation came out. And we just realized that I think probably a lot of people realize at the same time that you could put text and images together and create a story.

Jarrad Grigg

Mm-hmm. Mm-hmm. Mm-hmm. Mm-hmm.

Brian Foody

And so we shipped it into something that looked good, because Linus did the design on it, and it looked good. And he had a bit of a following on Twitter. And so we posted it, and it went on this real viral, like, it went hugely viral. like, yeah, the revenue went up pretty quickly. We threw a paywall up. So yeah, it did really well. But then one thing that we did realize, which is we created a site map. So every story created went back into the site map. And so the Google SEO grew for like AI stories.

Jarrad Grigg

Thanks. Mm-hmm.

Brian Foody

Because everyone who came in, we let them generate one story for free. But if you generated a story for free, it went into the Google sitemap. If you paid, we didn’t put it in. So all of a sudden, we had like 50,000 pages which were saying, AI, bedtime story generator. So think I didn’t even realize at the time that Linus was the guy who realized SEO was so important because he had worked on something. But I think that was my first time realizing how important SEO was. Because after the viral tale, you get this virality. And then you’re forgotten about because X is like a

Jarrad Grigg

Yeah. Well Yeah.

Brian Foody

to 24-hour memory. yeah, thinking about designing that sort of SEO from the start definitely helps. Because even now, like there’s a study we get 500 visitors a week on bedtime stories just from SEO. So yeah, SEO was definitely, I underestimated that. And I think a lot of people who ship products don’t think about that distribution element. You can do content, takes a lot of time to create content, and you might go viral. That’s one way to do it. But SEO is like,

Jarrad Grigg

Yep. Nice. Mm.

Brian Foody

If you can figure out SEO, you get a guaranteed funnel into your product. And especially with AI, you want to have enough volume that you can actually test ideas. So a little bit of A-B testing. If you don’t have that volume, it’s quite hard. And Google Ads and stuff now are just so expensive. So if you’re trying to kick off an indie product, I always recommend people think about an SEO strategy. Because,

Jarrad Grigg

Yeah. Yeah. Mm. Yeah. SEO doesn’t work right away, right? Like it’s it takes a couple of months to really kick in. And so you’re relying you’re relying on that virality of the idea first to get the impressions.

Brian Foody

That’s the thing, yeah. Not necessarily. Like, not necessarily. It might just take time to kick in. So trying to find keywords that are underserved. It’s like the internet’s way of, where is a busy street that doesn’t have a coffee shop? But a lot of us don’t think that way. We don’t design products based on what the market wants. We design them because we want to build them. Especially in India, you’re like, I’ve got this cool idea, and I want to build it.

Jarrad Grigg

Yeah. Mm-hmm, mm-hmm. Yeah. Yep. Yeah, yeah, yeah.

Brian Foody

So that’s the hard part. And that’s where you’ve got to get creative with socials if it’s something that’s not going to have an SEO funnel. But if you are sitting there and thinking, OK, I want to create a product, that’s one way to go about it. Look at like, ah, hate traps, and these places that let you look up which keywords are underserved. That’s a good way. I did it for another one, and I got traffic within like, it was within a few weeks it started to appear on Google. And then I just kept feeding it. And now it has like, I think I get

Jarrad Grigg

Mm, yeah. Yep. Yep. Yep.

Brian Foody

probably 200 visitors on that. So I’m going to try and build that out a bit as it goes. And the good thing now is once you get traffic, can go, hey, you can plug in whatever your AI operating system is, whether it’s Claude or whatever. You can have like pull in the analytics and pull in the conversion funnel and just let AI in the gradient to center it. so.

Jarrad Grigg

Nice. Mm. Yep. There’s a bunch of systems trying to do that now, right? Like close that product loop of looking at analytics, finding issues with it and and self-improving. Like post hoc have got their their system coming out soon. I it’s a it’s a huge, a huge opportunity. Like I I used to do that myself manually, obviously. I’d sit there and watch hotjar hotjar recordings and find the issues and then write the tickets and you know, that’s that’s the way to for product growth. So I I see a lot of that happening.

Brian Foody

Yeah. Yeah, yeah. Yeah. Yeah. Yeah, it’s made that job. just, humans are not great at seeing patterns among data. Data scientists are, but if you’re a product guy, it’s generally, there’s a bit of intuition, but sometimes you just miss simple things. So I went back to story, bedtime story last week, because we’ve been working, so with Vertical AI, we’ve been working with companies, giving them that AI operating system over their business. plugging in data sources and letting them operate that way. And so I went back and looked at my own products that way. And so with Bedtime Story, I plugged in the Mixpanel MCP, and I pulled in the Stripe one, and then I pulled in Versel for the logs. And I was like, look at all this and tell me where there’s gaps. And then it found there was locations where customers were clicking the Subscribe button, and they weren’t getting through. And so there was a small intermittent bug that it found. And already I’m seeing conversion go off because there was a leak there in the funnel.

Jarrad Grigg

Yep. Mm. Right.

Brian Foody

So it’s definitely like, yeah, I think it’s becoming easier to run like multiple smaller products. You have this little assistant come in and help you. Yeah, because I hadn’t looked at.

Jarrad Grigg

Yeah, yeah. And are running those as like a like are you running those as like an automated task, a scheduled task or a routine or something? What you what are you what are you using to do that?

Brian Foody

I don’t do them as scheduled yet. no, even when we go to businesses, what I recommend is spend some time doing it manually first, then turn it into a skill, then run the skill for a while. then when you’re absolutely like you’ve done two weeks of not getting involved, then you can think about making it more scheduled. yeah, even right now, there’s a lot of learnings I pick out even of just being involved in the loop. And I still like to keep my head what’s going on. I think very quickly, things can decay if you’re not in the loop.

Jarrad Grigg

Yeah, yeah. Mm-hmm. Mm-hmm.

Brian Foody

I mean, we’re getting there. I just don’t think the AI is there yet. I’d worry about it slowly degrading. Yeah, because we had a guy who a business who wanted us to put in an agent to do their accounts on routine. And I was like, no, you don’t want us to do that. Just stay in the loop. We’ll reduce it from an hour of your day to five minutes of your day, but spend the five minutes for now because.

Jarrad Grigg

Yeah. Yeah. Yeah, yeah. No, I think you’re right. It’s a good work. You’re not gonna get a good result. Yeah.

Brian Foody

Yeah, accounts are not a place where you want to let an autonomous agent do things just yet. Yeah.

Jarrad Grigg

It’ll be Yeah, agree. It it’s a slow it’s slow moving. the gains are real but you can’t just hand it hand the keys over just yet. so yeah

Brian Foody

No, but it’s so tempting, right? Because it’s like you get it. Do you know that story of the guy who made a guy sit people sit in the Uber cars when they were nearly fully autonomous, but they still had to have a human in there? No. So Uber were experimenting with self-driving, but there was a guy who had to sit in the car. But he was actually like, I think people didn’t realize they were legally responsible for the Uber when they were in there. But there became so little interruptions that the people in the Ubers just weren’t watching anymore.

Jarrad Grigg

No, no. Right.

Brian Foody

And then eventually the Uber ran over someone. I’m sure they killed them, but it was pretty serious. And that guy who was being paid very little, he was legally responsible. So does this thing with autonomy where you have to be very careful? Because you don’t start looking at it, and you start clicking yes. Was that Homer Simpson meme where you’re the bird pecks? But the ATO aren’t going to care that, well, you can’t say, hey, the AI did it. You’re still legally responsible.

Jarrad Grigg

Hm. Yeah. Yeah, yes, the bird. Yeah. Yeah, that’s right. You’re still liable for it. Yeah.

Brian Foody

So we try and design the workflow so that the human has to be involved a bit more. Because yeah, you’re still responsible for the output.

Jarrad Grigg

Yeah. Yeah, yeah, for sure. So ease into it, right? So if you’re going to improve your product in an automated way, do it yourself first, spend the time to understand what it’s doing, see the results every period of time, and then automate it. Don’t just launch it straight away.

Brian Foody

Yeah. But then, yeah, also understand that you will definitely get lazy over time and try and design something that makes you snap out of it and actually review things. Because yeah, we’re just this human nature to be like, yeah, that’s taken care of. Yeah.

Jarrad Grigg

Yeah. Yep. Yeah, yeah. So how much of how much of your code now do you actually write or read?

Brian Foody

I still do quite a lot, probably more than anyone in the team, really. Like for every hour or two where I do some AI coding, I generally go in and like try and hand code. And then I always spot like, I always spot small things where I’m like, no, that’s not what I wanted. Yeah, yeah, I do, I do a fair bit of hand coding still. I really enjoy it. I guess for me, again, it comes back to like, if you were a good coder, it’s even harder to accept this phase. Like outputting code was never really the slow part for me.

Jarrad Grigg

Yeah, yeah. Mm.

Brian Foody

So yeah, in some ways I’m sort of like my own, you’ve become a victim of your own success because you’re like, you don’t want to give up the thing you were good at. But I still think it’s a benefit for me, but over time it probably might become like a liability for me as they get better. I’m not sure.

Jarrad Grigg

Right. Yeah, yeah. At all? But y I it’s a good point actually, ‘cause you you y you know what good looks like. So it’s probably frustrating for you, it’s probably, you know, in the last six months or so where you would give it a task and you would see the output and say, this I wouldn’t write it like this. I’d I’d do it this way. Whereas I’m the same with the design, right? I look at it and think that this is awful. Do not use this. So for you, yeah, it’s it’s probably more painful in the short term because you’re having to correct its mistakes and work with it in that way.

Brian Foody

Yeah. Yeah. Yeah, yeah. Yeah. Yeah.

Jarrad Grigg

So but do you see it getting better? Is it ha has it given you the ability to to sort of think more about architecture or standards quality or but how is it impacting your development workflow?

Brian Foody

Yeah, I I read a really good post the other day and it was like, you should be able to do this with the non frontier model. So you should be designing the system in a way where the coding task is actually really easy. coming back to the bank, like when we did event sourcing, say you submit, so you have a bamboo portfolio and it’s 50 50 between Bitcoin and Ethereum. And you say, I want to invest $100. OK.

Jarrad Grigg

Yeah. Yeah. Mm-hmm.

Brian Foody

That submits an event to invest $100. Then a process listens to that and it says, okay, let me look at your portfolio. Your portfolio is split this way. You’ve asked for $100. I’m now going to split that into separate trade requests. And so it submits a trade event. And then there’s another process which looks at those and goes, okay, this is a $50 request for Ethereum. I’ll look at the Ethereum price and I’ll look at this and I’ll buy this amount of Ethereum deducting the fees.

Jarrad Grigg

Mm-hmm. Mm-hmm.

Brian Foody

then it submits it back, and then there’s another process. So when you design the system like that, each individual piece of code was actually really easy. But if you asked AI to do it all together in one, it’s really hard. And then if there’s a bug, and you go, hey, there was a bug in this, but it’s a big, huge file, then very quickly you’re just spending your time on rework. Whereas the other one I could do with AI models from a year ago. So there is this sort of trade-off where it’s

Jarrad Grigg

Yeah, yeah. Yeah. Yeah. Yeah.

Brian Foody

The better they become, you can also become lazy and just design really terrible systems. And then you get locked into like, when Fable came out, I had it spike up something. And now I realize none of the other models can actually work with it because it was so complex what it created. So can very quickly, I’m already vendor locked into a model that doesn’t exist anymore. So when Anthropic IPO, if you’re not designing your system right, all of a sudden you’re like, you’ve just locked yourself into a lot of cost.

Jarrad Grigg

Yep, yeah. Yeah, yeah. Yeah. Yeah.

Brian Foody

That’s something to be very careful of. I think you should be able to do the coding tasks with a pretty dumb model. That would be a challenge, I think, for a lot of companies. See, can you actually do it? Yeah. It kind of falls down a little with UI. With UI stuff, needs a bit more intelligence. But yeah, on the back end, I think it’s more system design. Yeah.

Jarrad Grigg

Yeah. Gemma f Gemma Four running locally or something, or you know, just something small. Yeah, yeah. I mean you touched on a good point there. Yeah, I think I the UI side of things super interesting for me. I experiment with a of tools, and you’re right, you you mentioned previously about design systems and they interestingly it’s something that I just just don’t like doing in Figma is creating these really rich, large design systems by hand. because it’s just mind num it’s mind numbing work. But when you have a really established one,

Brian Foody

Yeah, okay.

Jarrad Grigg

clear patterns, then yeah, it’s agents love it. They they they can build sort of better better UIs but not not better UX, you know. It’s not gonna pick up the human experience of you can turn those three clicks into one if you use this sort of animation and state change or something, right? So you that’s for me is where I need I like to spend a lot of time. But there are there is now a convergence of tooling on

Brian Foody

Yeah. Yeah. Yeah, yeah, yeah.

Jarrad Grigg

Catching the slopification of UI, right? So a lot of players are looking at ways to, whilst you’re building, catch that drift. I find agents love to, they’re really lazy and they just love to build new components and not use variables. They just do their own thing. And so you have to set these strict rules on them. And there’s a lot of tools now that try to catch that and also try to catch UI.

Brian Foody

Yeah. Yeah, yeah, yeah, they do, don’t

Jarrad Grigg

problems when it’s built. So it’ll point out all the you know, you’re using a purple gradient. those sorts of things. So I think it’s moving in the right direction. and for me personally I’m I am excited about it because it gives designers the ability to iterate more and to build.

Brian Foody

Yeah, okay. Yeah.

Jarrad Grigg

Because like you like with your products, you know, you have to learn really fast what the problems are. And I feel like with a dis a a designer, you know, if they’re building statically in in Figma, they’re not really seeing all those edge cases and all the mistakes and the the you know, it’s nice design a ha to design a happy path, but it’s another thing to actually see a user engage with it and see their failure points. It kind of makes you a better designer to do that. So I’m I’m a I am a big fan of of designers using these tools, but it is a really big shift.

Brian Foody

Yeah. Yeah, yeah.

Jarrad Grigg

to get designers who have been traditionally focused in a design program to open up Codex or Terminal or something like that and start building. But

Brian Foody

Yeah. Yeah, interesting. I remember, you know the guy, what’s the guy who runs Basecamp? You know that one? He’s always, yeah. I think they forced their designers to use HTML because they’re like, the designer should be using the material that’s the same tools that the developer is going to use because a lot of the time, like the developers in Figma putting on these gradients and then it’s like, they don’t realize that that’s like a lot of work for the developer, which may be less now, but.

Jarrad Grigg

DH H? Yeah, yeah. Yeah, yeah. Yep. Yep.

Brian Foody

It’s probably a lot easier to work like that as a designer now, where you’re actually in the same space as the developer. Yeah. Yeah.

Jarrad Grigg

Yep. Absolutely. And you get so much more empathy for the engineering. You know, so it used to be this, you know, somewhat butting of heads. I want this animation like this, you know. And and so if you can if you can build it, even if it’s crude, you can ship it to an engineer to say, This is kind of what was thinking, you know, and they can look at it and review. So I I love it. I think it’s a fantastic

Brian Foody

Yeah, yeah. Yeah. Yeah, yeah.

Jarrad Grigg

Period that we’re sort of wandering into. Okay, so I want to touch on vertical AI because it’s a very cool product.

Brian Foody

Yeah. Yep.

Jarrad Grigg

And a very cool idea. So I have a sort of similar product for small business, unrelated to voice, but voice was one of those areas that I did want to get into. I did want some sort of thing that could call subcontractors or staff members and support them all. And you know, when I saw vertical AI, I thought, fantastic, Brian’s done it. He’s he’s built this system and I n I know from my brief experimentation just how hard it is to not.

Brian Foody

Yeah. Yeah.

Jarrad Grigg

get the latency correct, but also the the cadence of the conversation and making it feel really natural and then taking all of that and turning that into something that can action itself in in a customer systems.

Brian Foody

Yep. Yeah.

Jarrad Grigg

Super complicated. so I was pretty I was really impressed by your latest video and so I I’d love to talk about the product itself, especially how it’s 100% Australian hosted, and how you kind of got to where you are now and if possible see a form of some form of demo of the design. Like I’d love to see how you created the website, for instance, some of the assets and things like that. share share the product with the with with the audience, it’d be great.

Brian Foody

Yeah, Okay. OK. Yeah, I guess some background on it is I’ve been playing with tools a lot for my kids. from a creative, I found generative AI was a creative tool for me first, like something I could build new products with it. And so one of the things I built, we called it Sprout Player. Me and David Burroughs built it. So it was like an interactive audio experience for kids. So we cloned Winnie the Pooh’s voice was out of like patents. You could use Winnie the Pooh’s voice and you could use his thing. So we cloned Winnie the Pooh voice and we made games for our daughters, which was like, the animal and things like this. And it was really good because you could prompt it to say, don’t give too much upfront, progressively give more of a hint. And they really loved it. But what we found was kids are the hardest test audience because they don’t wait patiently. And the AI just wasn’t there yet. So it is an idea we’ll revisit again. But I went to Frio Fest and showed off the product.

Jarrad Grigg

Excellent. Yeah. Mm.

Brian Foody

And then a friend who works, they have a lot of contact center traffic. She was like, could you use this for contact centers? And I said, yeah, there’s no reason you couldn’t. So we did a sort of design partnership with him to implement it there. And then we realized everyone was going to need this. So we went down the product path with it. And so a lot of what we’ve been doing is when we built it for him, it took like four weeks and it was all like custom Python code.

Jarrad Grigg

Mm-hmm.

Brian Foody

We’re like, OK, this is not going to scale like four weeks per person. So we did go down the, so we were talking to enterprises, but a lot of them is just too long of a sale cycle. So we more worked on getting the cost to create it down. So right now, you can create it with an MCP. So you can create an agent very quickly with an MCP. So after this, I can send you log in, and you’ll be able to create your own voice agent. So we can get the kind of cost down to creating a voice agent to about.

Jarrad Grigg

Mm. Mm. Mm-hmm.

Brian Foody

Realistically, it’s probably a couple of hours. The first one can be like 10, 15 minutes. I mean, we can try creating one on this call or on this thing, maybe, maybe. But you do need to train it and give feedback. So you would call it up and then you go into the MCP and say, hey, you should say this instead of this. But yeah, the hardest thing about all this is, number one, you have to be really quick to respond. So even though all the frontier models have accelerated, we’re still but

Jarrad Grigg

Live devote. Love it. Yeah.

Brian Foody

stuck back on like GPT 4.1, Gemini Flash 2.5 or Cloud Haiku. So they’re the only really three models that can give you fast enough speed in Australia and that you can access through Australian inference. So even though things have got really intelligent and voice AI were stuck back like from a couple of years ago, nearly intelligent.

Jarrad Grigg

Mm-hmm. And that’s true to the just because you need Australian residency and they don’t allow those models or or the Yeah, okay, yeah.

Brian Foody

No, it’s due to the fact that you have to respond very quickly. And that’s just not a focus of the frontier. They’re all focused on long-running, agentic tasks. They don’t really care about how quick it takes. They’re being focused on thinking models. So when you’re going to cloud code, for example, you know the way goes off thinking. None of that stuff is available to you. So you have to design certain elements around it. Then you have to have something reviewing it, because it’s not as intelligent. You need more review to make sure it’s staying on track.

Jarrad Grigg

Mm. Got it. Mm.

Brian Foody

And yeah, then doing in Australia is even harder because a lot of the voice stuff wasn’t available to us. So trying to keep it in Australia. So it has been doing AI on hard mode. But that is also like, that’s been the challenge. What we happened to probably one thing we need now is like about 0.3 of a second to come off the LLM inference. So how the voice AI works now is you have someone talks into a phone that goes into like a whisper model gets transcribed text.

Jarrad Grigg

Mm. Yeah.

Brian Foody

the text gets sent to an LLM, which gives output text, and then that goes down into the text to speech. And so that all has to happen within like by a second. That inter-end has to happen in a second. So the whisper part is good, that’s fine. The LLM is generally the bottleneck, and that’s where we’re stuck with the lower level ones. So we just need like, we need, we’ve played around with some open source ones, but they’re still just not intelligent enough. So we do want to host that LLM ourselves in Australia and do the inference ourselves, but.

Jarrad Grigg

Uh-huh. Mm-hmm. Yeah.

Brian Foody

It’s sort of we’re self-funded startups. There’s only so much we can do. And I’m hoping just a model drop just takes care of the problem for us in the short term before we host ourselves.

Jarrad Grigg

Yeah. So you’ve had to almost build like your own eleven labs.

Brian Foody

Yeah, and it’s funny, my friend was in America and he seen a guy at 11 Labs and he compared it directly to full self-driving. They thought it was even 11 Labs thought it was going to be a lot easier. it’s like, I don’t know how much funding they have raised, 250 million or something. So yeah, let me see. Let’s try and create one here. So first I can share my screen here and show you the one on the website.

Jarrad Grigg

Mm. It’s amazing. Yeah.

Brian Foody

share my entire screen. Okay, so can you see my screen?

Jarrad Grigg

Sure can. Thank you.

Brian Foody

move this out of the way. What am I doing? Virtual AI. So the phone picks up in less than a second now. The website, we haven’t pushed that change out to the website. So yeah, here’s some beautiful design that one of our guys has done. So you can try a call. You can either call on the phone or you can talk in the browser. I actually really like the design. It’s very interactive. I don’t know if you can hear this.

Jarrad Grigg

Mm-hmm. I like that actually I like that you actually use the product on the site. I think that’s a great feature. I’ve seen that as more of a trend lately is you know, why not, right? Like don’t hide things. Okay. Yeah, so you can

Brian Foody

Yeah. Yeah. Hi, how you going? think it’s hearing some background noise here from us. it’s going to get. Let’s call the phone number here so I can demo it. And the phone number is actually probably better. You can try this afterwards.

Jarrad Grigg

the right one. I’m not getting the audio through.

Brian Foody

9 3 a.m. This morning, you’ve reached vertical AI. I’m Sarah. What brought you to us today? So yeah, mean, that’s ultimately it’s about getting like, it has been a lot of engineering, like just getting times down is a huge part of it. And so what we’ve added in as well is like quickly scaling out the integrations. So yesterday we had a customer who wanted HubSpot integration. And so we have like, we developed it in a way where we quickly had integration. So as you talk, can, yeah, you can, you can.

Jarrad Grigg

Yeah. Mm-hmm.

Brian Foody

plug up those integrations. I’ll try and get into the test side of it where we can use text. Because we also do text. So let’s try this one. And so this one is just a demo one we did every day super.

Jarrad Grigg

Hmm. So how do you plug this into s into small business? Like does it i because it’s be it moves beyond just the phone call, right? So it actually it acts on their behalf, right?

Brian Foody

Yeah, so it plugs up to their phone system in general. So we are talking to some companies about it. You can also put it in an app. So it can replace something like Intercom. But yeah, generally, it’s pretty easy to hook up to your phone system. And then we have like, there’s things like a tool to call forward. So a lot of people can just do a quick call forward. So we do a transfer to human in here. And you can put in the number to transfer to.

Jarrad Grigg

Mm-hmm. Mm-hmm.

Brian Foody

But what we’re finding, we spent ages doing a lot of fancy UI stuff, and now we’re finding it’s just like an MCP is the way to interact with this, which is quite interesting. All the interface design we did. we have a knowledge base as well.

Jarrad Grigg

Yeah. Yeah. I agree with that. You know, the more I try to build things out with the custom UI, the more I realise I don’t want to use the UI. I I want to in interact with a service through my RDE, whether it’s Codex or Cursor or whatever. seems to be the way things are going. it’s it’s it’s almost a more enjoyable experience because every time you open up a software product, you have to learn the UI. And every team that’s created it has a different

Brian Foody

Yeah, exactly. Yeah. Yeah.

Jarrad Grigg

a different way of of interpreting your need and you have to try and figure it out. And if I can just use an agent that interacts with the endpoints and I it figures it out. so it seems to be the interface interface of the future.

Brian Foody

Yeah, yeah, yeah, yeah, yeah. Yeah, I think for us as a startup, that’s one of the things we’ve realized. I think you have a certain amount, it’s like a video game, you have a certain amount of resources. And if you’re fighting someone else, you need to make strategically bets. So really, one of the bets we’ve made now is that there won’t be much of a user interface. You can see the user interface is not great. It just has things in it. But here, I’ve asked for opening hours. You can see it’s grounded in the knowledge base. So for a lot of people, just simply voice agent with knowledge base does a lot of it, and then a call forward.

Jarrad Grigg

Mm.

Brian Foody

You know, like so much of it is as simple as that. Like that frees up your team from like so much crap. Like they actually get to focus on a lot of that. And then like the simple like things like looking up an order. So when we look at a lot of companies, like 80 to probably like 80 % are just very simple. Like there’s no reason that voice AI can’t handle them. And they can handle them 24 seven very easily. So everyone has these out of hours, just like we’re not here. So.

Jarrad Grigg

Yeah. yeah. Yeah.

Brian Foody

Like a lot of our thing is just just turn this on for out of hours and like help your customers. And there’s no reason you can’t like, your sales team can’t wake up to the CRM populated with five queries in the morning. So yeah, I’ll show you the empty. No, you go on.

Jarrad Grigg

Yeah. Yeah, and I guess yeah, please. I guess th the the I think there’s been a lot of hesitation around that type of service previously, right? So a lot of businesses don’t like the idea of chatbots because the experience is not great, right? and while some of these sorts of services look like chatbots

Brian Foody

Yep.

Jarrad Grigg

They’re so much better. And they are actually capable and can get it done if they have context, right? So are you are you building a like a custom knowledge graph or you know, you’re scraping the business’s knowledge and and setting that up for your agent to access? Or how how are you handling that?

Brian Foody

Yeah, we do have custom scraper that we built. So we have our own thing built on top of something called Parse. So any site, we can quickly develop a knowledge base for them. But I mean, it depends on the customer too. There’s this sort of offering where we can offer people something fast and cheap that they train themselves, or we can spend time working with them to craft it. So we can do both of those. So if we look at yours is overflow, right? Overflow.

Jarrad Grigg

Mm-hmm. Mm-hmm. Okay. O V O V R F L O without without the E. Dot Studio. Take the E out.

Brian Foody

And over. That’s Judeo. So let’s try and create one for you, I guess. I haven’t been…

Jarrad Grigg

Take the OV OVR. Yep. That’s it. There’s my gigantic orange head.

Brian Foody

VR, yeah, there we go. Look at that beautiful one. What would you like to handle in it? Just general inbound inquiries?

Jarrad Grigg

Sure.

Brian Foody

So let’s see how we go here. So yeah, we have a pretty comprehensive MCP. And one of the good things is, especially when we talk to some, I remember talking to CEO and I was like, how do you know that the contact center, that the customers are happy or what’s going on? And he was like, I have no idea. We just do a weekly status. I was like, I think you know where to look into it. And so one thing when you build it from AI native from the start, for me as a CEO, I’d want to go in and say, hey, give me a summary of the contact center yesterday.

Jarrad Grigg

Mm.

Brian Foody

and then pick out one call where the customer was really pissed off. I know you’re getting such a quick temperature check for your business then. I don’t know, for me, I call it customer-driven automation. Because a lot of companies are like, we need to embrace AI. And then they go into a boardroom, and they’re planning things out. And then they’re like, OK, we need to get our data lake. Whereas our sort of idea is, first, get the flow of information from your customers. That should be the driving force.

Jarrad Grigg

Yeah. Yep. Yeah. Yep. Yeah.

Brian Foody

So yeah, once you have a voice AI answering, taking crit, and even then if you’re not answering all the calls, it can still take in the summaries. So once you have the system there, know, it might be answering some, it might be going to the contact center for others, but all of them are feeding into like an MCP that you have where you can read them. And so like, you’re very easily able to like, even if you’re in a boardroom planning it out, like show me an example of where the customer had this problem. Like, is this actually a problem for business?

Jarrad Grigg

Yep. Yeah. Mm. Yeah.

Brian Foody

Yeah, so like letting that drive the automation of your business. Okay, so looking up your LinkedIn now.

Jarrad Grigg

look that. like if you you’re right, like the benefit comes if you can get a lot of information in one place quickly. So as it’s kind of happening, if you can if you can capture everything in one place, these this type of tooling is so powerful for small business.

Brian Foody

Yeah.

Jarrad Grigg

Because you’re not waiting, you don’t you’re not paying this. I find that as a team scale, they pay this sort of coordination tax where this person has to tell that person, has to write that down, and this just takes all that away. And everybody has access to that context immediately. It just it’s a mindset shift in terms of how you work, right? And once people figure it out. It’s just so incredible. So I like that you’ve captured that. But I I also like that, you know, you’ve you’ve gone for a very sellable idea first, which is look, it makes phone calls. You know, you can speak to this thing. But there’s so much more under the hood there that you’ve actually built. it’s really cool.

Brian Foody

Yeah, yeah. Yeah, and that’s where you’re like, this is one idea where I stayed very simple compared to my Indies where I like say I’m going to take over the world with this idea. It’s very simple. So here, how should the agent handle pricing on calls? Quote from prices, don’t quote, stare to call, full detail.

Jarrad Grigg

Yeah. Yeah, go full detail. Let’s go. Let’s do it all for me.

Brian Foody

Call outcome. Do want to capture inquiry as a message, book into a calendar or message and booking?

Jarrad Grigg

Let’s go both.

Brian Foody

So let’s see how we go here. So ideally, this should be able to answer for your business, chat to them, and then book in your calendar or whatever. I think, to be honest, we’re probably, the integration work might need some manual, but we’ll see. The guys are doing a great job on the MCPs, so let’s see how we go. Well, yeah, you can see how quickly, like, we’re not asking people to,

Jarrad Grigg

Great. Yeah. Yeah. That’s okay. Fantastic. Yeah.

Brian Foody

I don’t know, people have this idea that, no, it’s a new system. It’s going to be like a lot of work. Like if you’re actually like AI native, it’s very easy to plug this in and it’s not going to be much in your workflow. And what we’re looking at now for like even from a small business perspective is we realize like the AI operating system is quite important as well. And making it sort of, we don’t want Australian businesses to get locked into US companies. Like for us, I think that’s just like, we do not want that. We don’t want every small business using Anthropic, right?

Jarrad Grigg

Mm. Yes. Yeah.

Brian Foody

We’ve seen with Fable, it’s just able to like, you can’t be a small business or even a normal business competing on the international stage and your peers in America are using a higher intelligence model while you’re using another one. We’re in a good place in Australia where we can sort of play the US China off each other and get what’s best for us. So that’s one area we’re really passionate about is sort of making them, you know, the inference layer not dependent on US tech. So yeah, yeah. And we’re trying to.

Jarrad Grigg

Yeah. Yeah, yeah. I think that’s super important. And and and you’re right, but most people will just launch Claude and and start having a play. That’s their AI strategy, is just to tell people to start using it. But it’s not it’s not a great long term approach at all. I completely agree. I I you know I I get the feeling that, you know, local models are probably like six ish months behind Frontier. and I’m seeing a a lot of progress. in ZO I just released a model Yesterday, which was quite good. But you still need quite quite strong hardware, right? You need the twenty, thirty, forty grand’s worth of hardware to run these these models on your desk. So and it it it’s it’s

Brian Foody

Yeah, and it’s a lot to ask of people to invest in that.

Jarrad Grigg

Yeah. It’s slow going though, right? Like if you start small like with you have you they can see the benefit of it. then the cost becomes negligible when the benefit is there. Like to purchase the hardware and have it in the in the business, you know, it’s it’s it’s kind of a no brainer when when you see it working and in the power that it it can give the business. So yeah.

Brian Foody

Yeah. Yeah. One thing I would say to any small business, if they are going to do it, is there is make sure you turn off this help improve our AI models. So if you’re using Cloud Desktop, it’s just like, especially like, Cloud Desktop is a good place to like sort of have a play, but then you should be using something more open. Like we try and work with them to get them off it, to be honest, so that they can access any model they want. Also a small business, you want to like,

Jarrad Grigg

Yeah, yeah.

Brian Foody

The main problems SMBs have, or small business wallets, they want to create a lot of social content. They want to automate a lot of customer inquiries. Social content is actually like, you’re better off with Gemini and ChatGPT models, whereas the automation part, you’re better off with Anthropic models. So you want to select the best model for the job. having something, turn off that setting. Don’t train American companies. Then we also have.

Jarrad Grigg

Yeah. Mm-hmm. Yeah.

Brian Foody

Yeah, we’re adding in this layer to vertical AI where the companies can have their chat layer.

Jarrad Grigg

And then they can just embed that embed that wherever in their digital right.

Brian Foody

to where? Yeah, exactly. So here we have a Claude auto. So we do have a of a routing layer. So if you ask for the weather, it shouldn’t go to Opus. know, like what’s the weather down south. So for most businesses, they’re not going to be selecting which model they’re using. Whereas you want a router in front of it. And let’s see. Let’s check back to see. By the way, I love your product design. I think it’s a good time to give you a shout out here.

Jarrad Grigg

Mm-hmm. Yeah. Mm, mm. I I I’ve been behind the shadows my whole career, so putting my head on a website is makes me deeply uncomfortable, but I kinda feel like I have to do it, so.

Brian Foody

Yeah. Yeah, Matt, I had to record videos on the weekend and I was so hard. But you have to do it, Yeah. OK.

Jarrad Grigg

Yeah. Yeah, that’s right. But there is a new version coming which is which is AI native too, so I’m I’m moving off of a CMS stack and rebuilding from scratch. Okay, great.

Brian Foody

code. I’ll say it’s blocked on the dev create, it’s in prod. So yeah, we have a dev and a prod environment. See, I wanted to create it in dev and I didn’t have it authorized, so we’ll try and create it there.

Jarrad Grigg

Mm-hmm. So what so in building this product, what are some of the key like I wanna I want designers to be able to look at this and get an understanding of how like what lessons can they learn or how could they apply some of your learnings into their own stack? Like what kind of problems have you faced from going from an idea to where you are now and sort of how how how have you yeah, how have you solved those?

Brian Foody

And yeah. question from a designer perspective. mean, first, we started definitely with a customer problem first. So like this whole solution came from actually like a customer asking us to build something and then realizing that was going to come from a lot of people. So I mean, that’s a lesson that’s hard for me to learn because I always want to create something really fancy. like starting with an actual like someone asks you to solve a problem, then there’s probably like 20 other people who want that problem solved. that’s a good point to be like.

Jarrad Grigg

Mm-hmm. Mm.

Brian Foody

As a designer, you get asked to work on a solution. And then you might go, hey, actually, a lot of people might want this. You have the skills to create the design. Then you go into something like. So I think for designers, coworker is probably a good place. Or Cloud, like I did say, not to get locked in. I don’t want businesses locked in. But for designers, use the best and free thing that’s there right now. So you can go in here and talk to Cloud Code, get it hooking up. Take your designs, build something out. You got like a Stripe MCP.

Jarrad Grigg

Yeah. Yeah.

Brian Foody

One thing I’d show you here is story time. This is bedtime story here. if I do, yeah, of course I run Claude with accept danger, see accept permission. it’s the way. So yeah, I’ve got mixed panel and Stripe hooked up. So like there are two things you can do. So here I can say, how is the conversion funnel looking for the last four days since our changes?

Jarrad Grigg

Yep. Always. Is there any other way? Yeah.

Brian Foody

So I did do a big analysis on this.

Jarrad Grigg

I think it’s just it’s worth it’s worth pausing on that point just there, right? Like the traditional method of getting that information, you go to mix panel, you log in, you might you might not even have a report there, so you have to look through the default, or you have to set up some sort of custom report. the whole thing could take you, you know, five, ten minutes just to just to get through. And here you’re just converting that entire process into natural language and it’s figuring it all out and giving you the data straight back. Like this is whilst you know, a design look a designer will look at that interface and and go, Gross I what is this? But you know, when you you get past that, the experience is actually great because you’re just thinking and you’re turning that thought into text and you’re getting the result you want.

Brian Foody

Yeah. Yeah. Yeah, so an example with a context example would be, so we went out to, when we were doing some discovery, we worked with some smaller business. So Steve is a guy who runs a business called Effortless Superhuman. So he’s a functional medicine guy. So I don’t know if you know the functional medicine space, but there are sort of premises that traditional medicine is quite broken. It’s like specialists everywhere. And like a lot of people bounce around the medical system because

Jarrad Grigg

Mm-hmm. Mm-hmm. Okay.

Brian Foody

So no, example would be, an example he gave a long time ago was a guy had really tight hamstrings and kept going to the physio and not getting it, like nothing was working. And he ended up, had parasites in his stomach and his body was just completely inflamed. And that was causing like the tight hamstrings was like a downstream problem of like something completely different. So functional medicine starts with like, look at your gut, look at your bloods, like look at your body as a whole, and then try and find like, is there a root cause behind all of this stuff? So what he does is, People come into him, he’ll get a gut sample, he’ll get a DNA sample, get a blood sample. And so he’s got to synthesize a lot of information. And so each patient was costing him 60 minutes. And so when the team went in, so quite similar to what you see here, where I’ve plugged up Mixpanel and Stripe, we plugged up a system that’s like secure to pull in patient information. We put in a couple of MCPs and all of a sudden he’s got a central place where he can pull all that in. And we designed it then with a couple of like human

Jarrad Grigg

Mm-hmm. Mm-hmm.

Brian Foody

like considering that we don’t want someone to get just over time kind of skip steps. we worked with him to know what steps he needs to be involved every time. And so he went down from like 60 minutes per patient to 15 minutes. And this is a guy like diagnosing chronic pain. So like you can accelerate these types of people. There’s not many like him. So like it’s a huge win really. Yeah.

Jarrad Grigg

Mm. I mean Yeah. It’s incredible. It’s incredible. I mean it is a as a if you’re a business owner, you look at that and and you want a piece of it. Like that’s that’s what you want. Great job. So where do you w

Brian Foody

Yeah. So where is our? yeah. So here’s back to our checkout funnel. It actually looks quite good. Anyway, it’s got some analytics here. So subscription errors have gone from 22 to 1. So we’ve solved our subscription errors problem. there was a subs, imagine that me as an engineer, there’s a subscription error. So like, just being able to get information in, you can spot so many things. Your thing is still building here. So we hope they have a voice agent by the very end. Anyway, yeah.

Jarrad Grigg

There you Yeah. Yeah. That’s okay. We can keep chatting. So I’m I’m I’m interested in in you know, you’ve been around for a long time building things. So you’ve you’ve done everything by hand, you’ve seen this technology come in, you’re now using it to accelerate the way that you do things. What’s the end game here in your mind? Like where do you where do you see things going as intelligence improves? You know, I I have no doubt that you slammed Fable as hard as I did. I walked away from that thinking I mean, it was another level up and it was it was I thought it was pretty excellent. You know, these things are just gonna get better and better. Where does it leave software engineers and designers in your mind? and where do you see it going?

Brian Foody

Big question, eh? Yeah, I guess I come back to that idea of you still don’t want it to create. There’s a saying in language that’s like, don’t outsource the thinking. You outsource the doing. But I still think as a human, you want to understand your system, You still want to be in control of it somewhat. So I still think humans are needed for that system design. I think there’s going to be a lot more tooling around visualizing what the agent is doing and understanding it. So I’ve been able to digest that. I think we’re going to see a lot more in that space. But I just think when the Anthropic CEO said AI is going to write 90 % of the code, I didn’t fully appreciate that what he was seeing was that there’s just going to be so much more code written. It wasn’t necessarily that it was going to take 100 % of what’s written and do 90 % of that. There’s also the fact that it created a whole new 90%, right? So yeah, I just think there’s going to be so much technology.

Jarrad Grigg

Mm-hmm. Mm.

Brian Foody

I don’t know, I’m not seeing the like, doomer scenario kicking in where there’s nothing to do. I just think there’s gonna be a lot to do. Like a lot of the software, I actually think we’re in the dark ages of software for the last 10 years. Like, I work, like I go out to some people and I look at the software they’re using and like, this is just, how did you work like this for 10 years? Like, this is terrible. Like, they’re like, having to pay someone a few grand to like move a button around. So like, I think there’s just so, even look at like,

Jarrad Grigg

Yeah. Yeah.

Brian Foody

architecture right like from the look at the architecture back in the day and you look at the architecture over the last 20 years it’s terrible right so hopefully we can get back to just like a bit more craft and everything we do that’s kind of my like real optimist scenario and like like there’s a bit more bespokeness to things there’s a bit more love for what we do yeah we get out of the software dark ages where it’s 20 grand to add a new web page or update a button you know things like that

Jarrad Grigg

Mm. Yeah, yeah. Yeah.

Brian Foody

Like you see the value you’re delivering for clients by yourself. We can dream a bit bigger, I think.

Jarrad Grigg

Mm-hmm. Yeah, I I think you’ve touched on a really good point there. It’s that cost increased cost makes things worse. And it’s the same like construction, you know, architecture, you look at that and you know, the average you can argue you look at those stone carved buildings from way back and

Brian Foody

Yeah.

Jarrad Grigg

we have lost something there and it’s that because that cost is so expensive now. You know, I’ve seen te technology where you have robots now that can do stone carvings so that those intricate sort of stone carved buildings can make a bit of a comeback because the cost goes down. and definitely with software, you’re absolutely right. Like the cost of like you say, moving a button it shouldn’t be high. And the p and and and the doomers would say, you don’t need engineers anymore, you don’t need to like

Brian Foody

Yeah. Yeah, yeah.

Jarrad Grigg

have people anyone just gonna be able to generate this stuff so the industry’s over. I I I kind of agree the more I use it, the more I see it going the other way around. Where, well if the cost comes down, people will come up with needs for it. Because I’m I’m speaking to a lot of people now that are not engineers, they’re not designers, they’re not they’re not even in the industry, but they’re building personalized software.

Brian Foody

Yeah. Yeah.

Jarrad Grigg

Like I just had this idea, you know, so that you get this explosion of use cases, which means you get this explosion of need to to help with those use cases. So I I see that I see the the amount of developers will probably go up. The amount of design devo design becomes really important. and like I you’re you’re dead right. It wasn’t 90% of the existing code that was getting written, it was it, it was a new ninety percent. I really like that. and I think he was wrong about all the I think there was a comment.

Brian Foody

Yeah. Yeah. Yeah, yeah, it’s.

Jarrad Grigg

That Daria made about yes software developers not being needed or something, or that I think that’s totally wrong. And and you I’m sort of seeing that in hiring as well. So some of the other companies are now hiring more

Brian Foody

Well, yeah, mean, he wants you locked into Fab. Like I said, a good challenge for any software engineer is to be able to build what you’re doing with a lower level model. Have you designed the system in a way? Or is it like, I now need a 10 million context AI, which is going to take the whole code base because it’s so scattered. And I mean, that’s just money in the bank for Anthropic, right? It’s like, hey, in order to make one change, we just talked about reducing costs.

Jarrad Grigg

Yeah. Yeah. Yeah. Yeah.

Brian Foody

If making a change, agencies might have charged you five grand to do it, but if you have to pay entropic huge token costs, there’s a similar manner where if you design it, you’re going to get lower costs. So yeah.

Jarrad Grigg

Yeah. Yeah. I y I actually s yeah, you’re right. Like I see a lot of businesses jump at the opportunity to deploy this technology and they run these little sort of pods, you know, they try the they try the experiment and they quickly find that yeah, they keep running out of tokens and they have to, you know, get on the max twenty plan across X amount of users and then they realise, hang on a second, this is a massive annual cost that I have to pay.

Brian Foody

Yeah.

Jarrad Grigg

and it’s prob it’s definitely the most expensive subscription that you can probably get over a broad team. but if you compare it to other other costs in the business, like what it’s saving, if you get the value from it, it’s it’s not too bad. But you’re still locked into, as you say, that that provider. So yeah, it’s not as magical as as people make out to be, I guess. and

Brian Foody

Yeah. Yeah, yeah, it’s, mean, there is a magic to it, but yeah, it’s very easy to get vendor locked. And I mean, also a lot of businesses have like, they want to push more code, but they don’t want to pay the, they don’t want to pay the bill of like, if you’re going to push more code, you need to be able to evaluate the quality of it. And like, that’s usually the hard part upfront. Like when I, when we built Bamboo, I could ship to production because I had like, like a hundred percent test coverage. And I knew, like, I knew that I wasn’t going to break the system.

Jarrad Grigg

Yeah. Mm, mm-hmm.

Brian Foody

But if you have a legacy product and now you see AI and you’re like, ooh, I want some of that, but you’re not paying the bill upfront to like, you’re just creating more bottleneck on the test side, okay? And you’re creating more bugs and rework. So it’s kind of like a fat person who wants to more cake, but like, like, no, you got to get in shape first. I do something like, you know, there’s a bit of work to do first there. Like you just can’t keep like, yeah, yeah.

Jarrad Grigg

Yeah. Ha ha ha. Get your test suite in order. Yeah. And that that’s a good that’s a y that’s a really good point with these sorts of systems is that I’m the same, you have to have really robust test suites. as many edge cases as you can think about, so you have confidence that it’s it’s producing something that’s not gonna break.

Brian Foody

Yeah. That’s something we struggle with on vertical AI is like voice is actually really hard to quality get. It still like needs the human in the loop a lot. So like we’ll automate, we look at the text and go like, okay, the text looks good. And we do have a voice AI model that can like evaluate, but it’s not a human level evaluation yet. So the evals will say everything is perfect. And then you call it up and you’re like, whoa, that was horrible. It like interrupted me. You know, it’s such a human experience still. And it’s probably the same in design. Like you can’t fully automate.

Jarrad Grigg

Mm. Yeah, yeah.

Brian Foody

design evaluation or UX, things like that.

Jarrad Grigg

Completely, yeah, a hundred percent. And you know, I well run all the tests on like what I build and it’ll say it’s done and it’s all great and then you use it and y it is a manual process. You have to it takes time. I still try to tell people that it takes time to stick the requirements in here and and feel the system and and use it and think about what is the outcome that the user wants and how do you improve that, right? ‘Cause it agents are awful at it. And to your point about, you know,

Brian Foody

Yeah. Yeah. Yeah. Yeah.

Jarrad Grigg

a a a cheap agent should be able to to to do the coding. It’s the same with with letting them just run free and and design the UX. Like it you have to kind of set the the guardrails for the areas that it’s working on. If you just try and get it to design complete flows, it’s just it’s pretty awful. So it’s an additive tool. Anyway, that’s the way I look at it.

Brian Foody

Yeah. Yeah. And the further you get down without, like if you let the AI cook for ages and then you’re like, you realize it’s gone too far. It’s very hard to come back and pay the cognitive tax, I guess, to fix it. And so it’s very easy to just go, I’m just going to keep hammering AI until it gets it right. And yeah, like that’s where you end up in a bit of a, I guess, a lot of organization.

Jarrad Grigg

It’s i it’s easy to do that, but it it n it’s it’s not a hole that you dig yourself out of. yeah. Yeah.

Brian Foody

No, but it’s such a big cognitive tax to pay, right? You’re like, and I’ve got to understand what it’s been doing. So I’ll come back and see if I’ve got your agent ready. Where are we at? I mean, even anyway, you’ll see we sort of run our business through an AI operating system. So OK, where are we? We need a card. we need something’s going on with my Telnex MCP. So I need to manually do this part.

Jarrad Grigg

Yeah, yeah, yeah. Yeah.

Brian Foody

fix this offline. So again, like you can see, like look at when I go back to non-AI native, like usually I would just say, buy a number and hook it up. Now I’ve got it going here. It’s like, it feels like going back into.

Jarrad Grigg

That’s okay. Yeah. Yeah. I guess another interesting point is if your businesses are choosing software, try and find things that are AR native now, from the ground up, like not just bolting it on, because things are moving so fast and if even if it takes you three to four months to become more If you sta if your software stack is already there, it’s gonna make it much easier to get the benefits.

Brian Foody

Oh, massively. OK, so we’ve got a voice AI brought. How much time do we have left? I might have to do this. OK, we’ve got your phone number. We’re over time.

Jarrad Grigg

I think yeah, that’s okay.

Brian Foody

Okay, I’ve got your number here and you are going to be on them. Yeah, that’s you. but that’s gonna assign that phone number to you. So again, this is a good example of what an AI native business would be. You just said if you go out and buy a software that doesn’t have an MCP or something, being in this space where you’re moving between systems, then all of a sudden you have to go out to a different user interface. It’s okay if you know that user interface, but imagine you leave and then a new person comes in and then they’re like, what’s this manual step I have to do?

Jarrad Grigg

yeah. Yeah, yeah. Or or some people, you know, a good portion of their week is just clicking into this thing, copying that out of there and moving it over there and then generating that report over there. You know, there’s there’s this low level administration which if you’ve got the right tooling, it all just goes away. so yeah. This is great.

Brian Foody

Like I can turn this into a scale afterwards. Yeah. Yeah, completely. Yeah. So here we’ve got your overflow assistant and it’s been created in here. Move this out of the way, sorry. So yeah, did we have actually, we can go in here and test this. we need a card. Let me put you on the free tier. Should be able to put you on the free tier. Okay.

Jarrad Grigg

Ha ha.

Brian Foody

Unfortunately, because the telnet’s MCP is out.

Jarrad Grigg

It’s okay.

Brian Foody

Yeah, I might have to do this offline. Yeah.

Jarrad Grigg

That’s okay. I think y we can we can we can kinda edit out some of this stuff. But the what I’d lo what I would love is as a little video at the end of you know, you kinda calling it and and it talking back. that would be super neat.

Brian Foody

Yeah. Yeah, if you give me a couple of minutes, I’ll have it. Yeah, it’s amazing when you’re in that flow state and then you have to go back to these user interfaces. just, yeah, it feels horrible. Yeah.

Jarrad Grigg

Yeah, it’s clunky, isn’t it? So your so you mentioned before that you know your website was was built by the team. do you have designers working on it and and how are they using this sort of t this stack to produce the work for you?

Brian Foody

Sorry, could you say that again?

Jarrad Grigg

so you mentioned before on your website that you had the team sort of create it for you. How are they using this stack to build out the the the site? You know, how much of their work is handled, you know, in terminal? what is their w what’s their workflow?

Brian Foody

Yep. They are very AI native. they are mostly coding in cloud code a lot of the time. But a lot of our focus now is more on the evaluation side. we got to the point where it’s very easy. We’ve worked on like, it’s sort of like a production process. Like you want to get a car created a lot quickly. Like the voice agents, we wanted to automate as much as we could here in creating it. So yeah, now it’s more on quality gating. So how can we make sure that what we build is like a high quality?

Jarrad Grigg

Mm-hmm.

Brian Foody

How can we evaluate that without a human in the loop as little as possible? So yeah, we’re building out a lot of developer tooling. And now we’re moving more on into that integration of the operating system and the voice agents. So just like you’ve seen here as an MCP, for us it should be as a dentist, can you call these three patients and rearrange an appointment for them? So trying to enable those integrated voice experiences in whatever AI operating system you’re in, whether it’s our one or somebody else’s.

Jarrad Grigg

Mm-hmm. Mm.

Brian Foody

And yeah, but they’re very, they’re probably 90 % using something like cloud code to build and the rest is like manual stuff. It’s pretty heavy. But we are also at like the very early growth stage. So that would probably, I think that might tail off a little bit. I’m not sure yet. We have like, we’ve, because we’re paying a lot of, a lot now to get the eval side of, I’m hoping we can keep accelerating on the code side because we have good confidence that the system is good.

Jarrad Grigg

Yeah, okay. Yeah. Yeah.

Brian Foody

Okay, where are we at? The telnet is hooked up and just make this account free.

Jarrad Grigg

It’s good. It’s I also find it good when when people see the prompts, right? They actually see the way the people working in these. It kind of demystifies it a bit because in their minds I think it’s this some hyper complicated approach. how do you even how does an AI build something? No, no, it’s you just talk to it and if you don’t know, you just ask, you know, and you kinda work your way through it. it’s a a learning accelerator, you know

Brian Foody

Should be. Yeah. Yeah, completely, Yeah. Yeah.

Jarrad Grigg

The amount of things I’ve learned in the last you know six to eight months has been phenomenal. you know, I was using it you know, my own startup

Brian Foody

Yeah.

Jarrad Grigg

Involved in menus and you know we were using the f one of the early versions of Mid Journey to do all the food graphics for the menu designs and also using the early version of GPT to generate menus in JSON. so it would it would you know give us five menus you know that have 10 items and it would generate those and we’d suck that out and pull it into our system and it’s just a lot of fun. and it wasn’t it wasn’t once you start to use it.

Brian Foody

Okay Yeah, okay.

Jarrad Grigg

You know, you can do so much more with it. But I I I started to see it get progressively better. and it wasn’t until it’s December last year really. It was Opus four point five and Claude Code combined over that Christmas break. I’ve just learned so much. more than more in my like in more in the last five, six months of my almost my whole career I’d say. it’s been a lot of a lot of fun.

Brian Foody

Yeah, OK. Yeah, is. I actually haven’t been creating voice agents. So even for me, this is learning through AI. The voice has changed a few things since I was last in it. But yeah, just cooking here and figuring it out. Yeah, mean, it’s pretty, like even for this, is, I guess, something for your viewers too, is we have offered a referral program for this. So anyone who’s good in MCPs can literally grab.

Jarrad Grigg

Yeah.

Brian Foody

make some money for themselves. Where is it here? So yeah, you can either create them yourselves or like refer someone to us and we make them. And then we have like a referral tier. So it’s a nice way to make some additional revenue. Cause a lot of it is just like the distribution elements. we’d rather like, rather than pay Google ads, we’d rather like have someone who knows a company and refers them and then they make money as well. So the way we’re structured, our growth model is like referrals. We’re hoping to grow via referrals. And so the way that pricing model works is you make money.

Jarrad Grigg

Mm. Yep, yep.

Brian Foody

ongoing as long as so our pricing model is 25 cents per minute for the AI agent. So yeah, you would make like up to 35 % of that ongoing. So we’re hoping that pricing model works or that growth model works. So yeah, if you want to, if you know any businesses who need a voice agent, hook them up.

Jarrad Grigg

Mm-hmm. Yes. I might, I might, yeah. I think it’s a great product. look I think I I think we might have to cut it sh short.

Brian Foody

Yeah, let me try and call you here. I think we’re nearly done and then I’ll leave you. eight. Yeah, that’s cool. This is the last check. eight, six to eight, five, seven, six, two. This is a mark. Then we’re done.

Jarrad Grigg

Yeah. Yeah, it’s all right. Ha ha ha.

Brian Foody

Can I help you today? How does your business work? I’d like to get someone in on design. Let me grab some details about what we do, then we can talk through getting you set up. Yeah, sure. I run a company called Fertigal AI, and I’d like to get someone to help me on the design front. needs a bit of a pick up. Ideally, they’d be working like a few days a month. Got it. We work

Jarrad Grigg

Yeah.

Brian Foody

usually running two to four months. The best next step is a free 30 minute discovery call with Gerard, where he’ll scope your project properly and give you a real quote. Does that sound like something you’d like to explore? That sounds like a good fit for a design engagement. Yeah, cool. Could I, could I book in for next Wednesday, please?

Jarrad Grigg

Ha ha ha.

Brian Foody

Absolutely, next Wednesday works great. I’ll need your name, email, and I can get that booked in for you. Anyway, yeah, that’s sort of like a quick cook up. It’s got all the knowledge of your company. Things like that we’d work on then is like, you shouldn’t say OVR or FLO, so we’d like add some things like that. So yeah.

Jarrad Grigg

Fantastic. Yeah. You refine the experience. I mean it was that like fifteen, twenty minutes to go from nothing to a knowledge agent, with voice.

Brian Foody

Yeah, yeah, and that’s with some integrations that were done, which I should have tested before our meeting. yeah, all running in Australia. So yeah, it’s cool. Anyway, thank you for having me on. I better let you get back to work.

Jarrad Grigg

That’s all right. And hosted in Australia. It’s great. Very, very cool. No, that’s alright. Look, it’s been an absolute pleasure. and it’s great to hear about your your past as well. so thanks for your time and yeah, we’ll chat soon. So I appreciate it. Cheers, wait. All right, so

Brian Foody

Chat soon buddy, and I’ll see you around, yeah? Catch you later, bye.