Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Alexis Griveau - EP:015
41 Plays4 days ago
Welcome to Episode 015 of Mind the Model: The Modern Marketer's Guide to AI! Your hosts, Nathan Guerra and guest co-host Cat McGinn (filling in for Emmalee Crellin), sit down with Alexis Griveau, the CEO and co-founder of Chime Labs. Before transitioning to founder life Alexis spent nearly a decade working at Google. Alexis shares insights from his journey building an AI-native startup in Sydney.
This episode covers:
- The Power of Context: why providing comprehensive context and connecting data pipelines is essential for turning AI into a "superpower" colleague.
- AI for the Trades: A deep dive into how Chime Labs uses voice AI to solve the problem of missed calls for plumbers and electricians, providing a receptionist that can qualify customers and book appointments directly into calendars.
- The "Human in the Loop" Approach: Why Chime Labs maintains human review processes to manage edge cases and ensure AI handles the nuances of a business correctly.
- The Future of SAAS: A discussion on the "SAAS apocalypse" and how AI-native solutions may eventually take market share from traditional companies that struggle to rearchitect for an AI-first world.
📚 Resources & Links
- Chime Labs: https://www.chimelabs.ai/
- Alexis Griveau on LinkedIn: https://www.linkedin.com/in/alexigraveleau/
- Lovable: https://lovable.dev/
🎧 If you liked this episode, follow, rate, and share Mind the Model with your fellow AI-curious friends — it really helps us grow!
💌 Got thoughts or ideas for future episodes? Drop us a message at mindthemodelpod@gmail.com.
Recommended
Transcript
Introduction and Co-hosts
00:00:11
Speaker
Hello and welcome to Mind the Model. I'm Nathan Guerra and as unusually, I'm not joined by my normal co-host Emily Crawlin, but today i have Kat McGinn who has jumped in because Emily has a little bit of a cold.
00:00:24
Speaker
So ah welcome Kat to Mind the Model. You're joining us today, but you're not our guest. So I'm going to do the normal sort of Mind the Model kind warmup, which is to say like, Kat, what have you been doing with AI this week or two? I spent the weekend i mean just falling down a rabbit hole of vibe coding new product that I'm super excited about. And I definitely, by the end of the weekend, started to believe I could see through time. That's how excited I am about it.
00:00:53
Speaker
And and so so can you is it a like top secret product? or and And what have you been vibe coding it with? um So Claude and um yeah using Claude code and then some some other platforms as well, stitching them all together. And yeah, I'll watch this space. I'm looking forward to sharing it with the world. Kat, maybe we should start with a little
Kat McGinn's AI Journey
00:01:14
Speaker
bit of like, who are you? Who are you? Like, why are you joining us today? how do we how do we know each other? Like, give it give me a little spiel that we can use for the ah to introduce yourself to everybody. So I've been curating um
00:01:25
Speaker
the I think, as as it stands, Australia's only um AI conference and community for the creative industries, Humane. So running that since 2022. And i work for, I recently set up Liminal and Co, which is a founder-led advisory helping leaders with their AI transformation.
00:01:47
Speaker
ah AI expert, journalistic expert, like you're you're the perfect person to come and jump in and help us out. is that Is that what you're saying? I think we absolutely are not calling ourselves experts in this time and space. um A keen, um I've been described as AI's most anguished champion, and i'm I'm okay with that description. Thanks again, Kat, for joining us. um I'm just going to jump straight into it because ah it's absolute chaos here today in so many ways, and I do want to get to our guest.
Introducing Alexei Graveau
00:02:16
Speaker
So today we're joined by a gentleman I've known for, i don't know, probably
00:02:21
Speaker
almost like 12 years, I'm guessing. Alexei and I go way back. He and I, I think he was at Google when I started, in at least in Sydney. So um we'll get into that in a second. But Alexei Graveau, he's currently the CEO and a co-founder of Chime Labs, um a Sydney startup that was recently just funded. They closed their first pre-seed round of nine hundred ah k um I think that might've been USD as well, but we'll find out.
00:02:49
Speaker
um He spent nearly a decade at Google ah before making the leap to Founder Life. a Alexi, welcome to Mind the Model. Thanks. Thanks for having me. i'm excited to be here. We'd like to kick things off with a quick start question or two. Kat, do you want to want
Favorite AI Models and Their Uses
00:03:03
Speaker
to give it a go? Sure. Okay, so to kick this off, what's your favorite AI model right now? And what was the last thing you used ai for? Yeah, cool. and Look, we use quite a ah few different ones. I think our go-to is for the ah for our actual core product, like anything that goes customer facing, we tend to default to Gemini just because it's you know it's got a ah much more scalable architecture and it's um it's all kind of privacy minded. And you drink the Google ah google Kool-Aid. Come on. Of course. It's hard to escape when you spend 10 years there. And exactly like that, right? like We know it in and out. So it's it's nice and easy for us.
00:03:39
Speaker
And then um all of our coding is done at the moment using Claude. So everyone's got access to Claude, whether they access it through, you know, Windsurf or directly through the Claude app. um We're all using Claude to do a lot of development work, and that's really accelerated sort of the the pace of of development internally. And then one that that I'm particularly fond of that I use all the time now, mostly for like internal tools or pilots is lovable. I absolutely love lovable. It's like it's so easy to spin something up as visual that works and that can connect to lots of different sources.
00:04:14
Speaker
So for a lot of our internal tools, we'll actually just spin up um a lovable like project and and use that internally for like dashboarding or reporting or insights.
Context in AI Interactions
00:04:25
Speaker
It's just so easy to use.
00:04:26
Speaker
I'd love to dig into that a little bit more. um But before we do that, do you have a particular prompting methodology or approach? No. What I'd say is what I've discovered over the last few months is context is everything.
00:04:38
Speaker
The more context you can feed into your ai ah the more it becomes a um like a colleague you know or or an employee.
00:04:50
Speaker
What I found is that oftentimes when I get frustrated with AI, it's usually my fault because I haven't fed it enough context. so to give you an example recently we were reviewing our own ad spend and what we were doing with with ads for chime labs and i just started connecting every single data source that we had access to right so google ads meta google analytics uh our google tag manager and when it's got access to absolutely everything then it just becomes like a genius at solving your your ad problems it can really understand like
00:05:22
Speaker
Who are you targeting? you know What's the purpose of every single tag in your in your account? How should I review this? um our you know as everyone we our Our tag manager setup was a complete mess.
00:05:34
Speaker
um I gave it access to everything and it went in and actually fixed every single tag we had and just pushed everything live. So I think context is is everything. The more you can you can connect data pipelines into your AI models, the the better the the end result will be. In that case, you just gave it access to a bunch of APIs, presumably, or, you know, MCP servers and said, go with it.
00:05:56
Speaker
um But for context, are you then building notebooks in Gemini? Is that how you're doing it? are you using notebook LM and then giving Gemini access to those notebook LM? Or like, what's what's your context tip? I mean, it's really just plugging in the APIs. It's probably the biggest, you know, headache in the beginning.
00:06:12
Speaker
um securely connecting everything into into whatever you're using. So in this instance, it would have been Claude, right? so we gave Claude access to to everything. Obviously, you want to be careful with the permissioning, but once it's got access to everything, um it can it can pull data from, you know, disparate sources. So it could be Google Ads, but also HubSpot, right? so like What is the what's happening at the top of funnel? And then how are people converting at the end within, you know, your HubSpot account? So having visibility across the entire journey makes it like it turns it into a superpower.
00:06:44
Speaker
I'd love to hear about when an ai situation went wrong, when it didn't work. What did you learn from it? We've never had any like catastrophic failures, but we've tried.
Challenges in AI Memory Systems
00:06:55
Speaker
you know like There's times where I get really frustrated with ah with Claude um when I've asked it to do something multiple times in different sessions and it doesn't remember um it doesn't remember my methodologies or how to access data.
00:07:11
Speaker
um That can be really frustrating. Generally, it's just a case of like being diligent with your prompting. um So explaining to it, you know, write this to memory, make sure that you write yourself an instruction file on how to access the MCP servers for this and this a piece of software.
00:07:27
Speaker
um you know i think that's something that eventually all of these agentic models are going to solve like that that's really a ah product first problem um they should be you know that should be a level of intelligence that the models have but right now you still have to do a bit of you know force prompting explaining to the model how it should act in in future sessions where are you going for inspiration or information Yeah, I listen to a lot of podcasts on on YouTube, like every like every day. I'm just but whilst I'm working, I've got my headphones and I'm just listening to to podcasts.
00:07:59
Speaker
You know, there's there's some incredible content out there. If you go to the like a Y Combinator podcast or Sequoia or like any of the startup focused podcast or AI focused podcasts.
00:08:10
Speaker
um Every day you get ah inspiration. And then your you know your feed just starts becoming only that after a while, um where like you constantly have new ideas popping up on your Instagram feed.
00:08:24
Speaker
And then constantly I'm just bookmarking things. I'm like, oh, I need to come back to this one and think about how I integrate this into my into my workflow. So it can become overwhelming. And I'm sure there's like some kind of agentic way to download all that stuff and then you know feed it into your... into your work pipeline, but I haven't gotten there yet. Okay, when you solve that, let me know, please. Will do. I have a follow-up. So how do you manage that sense of kind of AI overwhelm? You know, how do you stay grounded? There's a new model every five minutes, but how do you sort of sort the signal from the noise?
00:08:55
Speaker
Yeah, it's um we had this challenge at the beginning of this year, and I was getting really stressed because I felt like the way that we were developing features internally wasn't aligned with the progress that the models had made.
00:09:08
Speaker
And so we spent a lot of time internally um experimenting with, you know, Cloud Code and, you know, Windsurf and lots of different tools until we got a process that we felt was right.
00:09:20
Speaker
And the moment where i actually started relaxing is I listened to a podcast by Andre Carpathie, and he was saying that his code development switched from where he was manually creating 80% of the code and AI was doing 20% of the code um and then flip that on its head to AI was developing 80% of the code.
00:09:42
Speaker
And that only happened for him in like December of 2025. And I was having this freak out in like January of 2026. So i was like, okay, if the best like AI people in the world have only just made the switch, then we're we're okay, you know? Super interesting tips and tricks coming out of him. And I mean, you don't even need to pay attention to him because anything he talks about gets publicized a thousand times across the internet.
00:10:04
Speaker
um All right, moving on to the you and ChimeLabs.
Founding Chime Labs and Industry Insights
00:10:09
Speaker
I'd love to kind of get the rundown on, um I guess, a little more background on you for our listeners. I obviously know everything about you um and ChimeLabs as well.
00:10:20
Speaker
Look, I started my career in in the ad space in London. I was working for a big global agency in London. I spent a few years there and then I moved to Australia, ah did a few more years in the ad space and then moved over to to Google. and And as you said, I think we joined around the same time in 2014. Yeah.
00:10:37
Speaker
yeah and So I spent 10 years at Google working primarily um in kind of technical sales role. So it's kind of a bridge between ah between customers and our kind of technical ads products.
00:10:50
Speaker
And that's where I met my co-founder, Matthew. He was a solutions architect at Google. And we worked on a few projects together. We probably spent like two or three years working on projects together.
00:11:02
Speaker
And at that time, I remember thinking, this is like one of the rare people that actually I could just i could start a company with. Because like starting a company and and finding a co-founder is one of those critical, critical decisions. If you don't pick the right person, it can be a real problem. It's like there's some weeks where I speak more to Matthew than I do to my wife. You know, like that's how like intense startup life can be.
00:11:23
Speaker
um And so we we met at Google. um you know he He left Google in 2024, I did in 2025. And we very quickly you know had one discussion and decided, let's start something together. We were both really excited about the AI space.
00:11:39
Speaker
And um and initially we started as a consultancy. So we spent about eight months just helping you know everything from small businesses to Fortune 500 businesses deploying AI solutions. And that gave us the opportunity to really talk to a lot of business leaders and understand where they were willing to invest resources into AI.
00:12:01
Speaker
um And that was really, really beneficial for us because ah you know we could really get the lay of the land in terms of um where the opportunities were. Can ask about the consulting piece? Was that um was that a pivot?
00:12:16
Speaker
um Because you guys had a company and you were doing stuff. um Was there a pivot involved in that or was it ah a conscious decision to kind of um listen and learn? we and When we started the consultancy, we always knew we wanted to pivot towards a SaaS product.
00:12:30
Speaker
And so it was a conscious decision to um to to do some like almost like market research while we were doing the the consultancy. And then once we had a really firm grasp of what we could build into into a product, um then we made the pivot. And that decision, it took us a bit of time because we really wanted something where we thought we could add a lot of value, where we really understood the the problem statement um and you know where like we were kind of pushing the frontier of what was possible.
00:13:01
Speaker
And so we came up with the the concept of Chime Labs ah in 2025. And the the main insight for us is that although Matthew has spent most of his life as you know a developer and a solutions architect, he actually on the side spent 20 years helping his dad ah build up his construction company.
00:13:22
Speaker
And so he spent 20 years basically solving all of the back end admin with the tools that were available at the time. And so we already had like kind of those those insights like what works, what doesn't work, where are the friction points in that in that process and that workflow.
00:13:36
Speaker
And that led us to to Chime Labs. Right. So we started speaking to a lot of people in the um in the trade space, you know, a lot of plumbers, electricians. I think everyone, whether you're especially in Australia, probably has like a few friends who are tradies.
00:13:52
Speaker
And so we leveraged that, had a lot of good conversations. And then we were like, OK, there's a there's a huge opportunity here. Let's talk a bit about the insight around what the big challenges were for tradies and um how you develop the product to meet those needs.
AI Voice Receptionist for Tradies
00:14:05
Speaker
The first one that we identified kind of the wedge in for us ah was this issue of missed calls, right? We like every time anybody ever needs to to get like a plumber electrician their house, everyone does the same thing. They go onto Google Maps.
00:14:20
Speaker
And then they'll get a list of tradies in their local areas. And then we'll just start calling them. And they'll book in the first guy who who actually answers the call and and confirms an appointment, right? And we knew there was a real pain point there for customers, but also for the tradies themselves. If they're missing calls, and usually it's because they're super busy, right? They wake up very early. They're usually on site by 7 a.m. They're working till, you know, three or four on site. And then they go home and they have to do a bunch of admin when they get home.
00:14:49
Speaker
um So for us, we wanted to solve that. And up until probably, you know, late 2024, early 2025, the technology wasn't didn't exist to solve that problem.
00:15:01
Speaker
And we'd already run a few ah projects where we used voice AI to solve some similar problems. who were like, okay, well, there's a real inflection point here between the technology that we've used already in the past and a problem that we can really solve.
00:15:17
Speaker
And so that's how we decided to build the first product, which is essentially a voice receptionist that can answer calls on behalf of tradies while they're whilst they're busy.
00:15:29
Speaker
We train it up very quickly on their business so it knows, you know, it knows almost everything about their business and it just answers the call like a regular human, right? One of the main things that we need to get over when we're talking to our customers about the technology is they're like, well, we don't want Siri answering the phone for us and we're like, no, no, no, this is the technology is way ahead of that, right?
00:15:50
Speaker
and so we And so the receptionist essentially, it can take calls, it can address questions, qualify their customers. And then if the customer wants to book in a job, then it checks their calendar, finds availability, and then books the job straight into their calendar, right? So there's a direct connection between the products and then a job being booked for the tradie. In terms of training data, what how much data do you actually need to be able to to onload onboard somebody? Yeah, so we can get their kind of baseline agent up and running really within you know five minutes. And the way we do that is we'll scrape their website and extract all the information about their business, which can be locations, opening hours, services they provide.
00:16:35
Speaker
That all gets passed into the agent. um And then usually we'll have, you know, a 20 minute onboarding call with the customer because every business has slight nuances. Maybe they are willing to do stuff on the weekend. They take emergency calls.
00:16:50
Speaker
um So there is a bit of human customization of the agent that goes into it. ah And then over the course of the next like week or two, if there's any edge cases where the AI got things wrong, because that can still happen, we have kind of ah a two layered approach. One where AI will review every single conversation and flag anything that's going wrong.
00:17:14
Speaker
um We can auto update the the agent if ah if it's something simple or we pass it on to a human reviewer to go in and and proactively make changes as well. And how would tradies you know How leaned in are they to the possibility of having an AI receptionist? Do they find that? Are they worried about the impact on kind of the trust of the relationship with their customers?
00:17:36
Speaker
you know They have the same social media feeds as we do. And so they get they're also getting overwhelmed by like AI information and knowledge. And as business owners, they're thinking, they're actively thinking, like, how do I integrate AI into my business? The problem they have is ah finding the time to create the spoke solutions or to really dive deep and do a full transformation can be quite difficult.
00:18:01
Speaker
And so having something that's prepackaged for them and they can just deploy, you know, within within a 20 minute onboarding call is really valuable. And in terms of. uh the nervousness there can be some nervousness you know what we find is it's it's you know it's not surprising but the younger tradies who are starting out they are much more willing to test it out you know older tradies who have been running their business for 30 years and it's going well they're like well do i really need this product um some of them go for it some of them are like you know what we're happy with what we have today we don't need to make any changes
00:18:34
Speaker
I mean, and if you can load up something in five minutes, I guess the proof is probably in the pudding for a lot of these things, right? Being able to show them, Hey, here's a product that already is ready for you to go.
00:18:46
Speaker
Very little you need to do. i mean, is that how you're approaching sales? Yeah, that's right. Um, one of the big things that we did at the beginning of this year when we started commercializing the product was how do we reduce the time to value?
00:18:59
Speaker
So how do we get them to start speaking to their agent as quickly as possible? And so now our onboarding process has completely changed where they get engaged in the process of creating their own agent. So they pick the voice.
00:19:14
Speaker
You know, they talk about their their company and their services, and then immediately they can start having a conversation. And that's a real unlock for them. They're like, okay, now I understand. Now I see the value and I'm comfortable putting this in front of my in front of my customers.
00:19:28
Speaker
In fact, like one of our first pilot customers who started using it, he um he was an appliance repair guy on the on the South Coast. And he was just a one-man band.
00:19:40
Speaker
And he um he told us that in the first month of using it, he got 34 additional jobs that he wouldn't have gotten because he didn't pick up the phone frequently enough. And that was $10,000 in additional revenue that he generated just from you know implementing a Chime Labs receptionist.
00:19:57
Speaker
And he would also tell us anecdotally, he would be like, you know, I was nervous at first, but actually my customers, whenever I come to their house now, a lot of them are like, oh, that AI receptions was pretty cool. Tell me a little bit about it.
00:20:09
Speaker
So there's, ah you know, there's a bit of a talking point there as well. Let's talk about when things
Ensuring Trust in AI Systems
00:20:13
Speaker
go wrong. So you mentioned needing to kind of have some review. What are the processes? What are the guardrails that you've put in place? Voice AI is...
00:20:21
Speaker
is particularly sensitive because when you're talking to a chat bot, if it produces some wrong text in a long paragraph of text, you kind of just skip over it and you don't really think about it.
00:20:34
Speaker
Voice is a lot more personal. um When you have something going wrong in a conversation, immediately it feels off. If the latency is a bit too high, if the AI does something a bit weird, then people get put off immediately. and so We spent a lot of time trying to work out the edge cases and the kinks.
00:20:55
Speaker
like I'll give you an example. You never want your voice AI to try and laugh. Having a voice AI trying to laugh, is just it's just terrifying. Oh, that's something I hadn't really considered. so that was one of the first guardrails we put in. Never laugh. ah That's exactly what it sounds like.
00:21:16
Speaker
And on a call, people are like, what's going on here? The Terminator is like taking over. But um it's it's an ongoing effort. you know um i won't say it's 100% perfect. We're constantly reviewing it. And like I said earlier, every business is slightly different. They want a different outcome from the AI. So we constantly have to be on top of that. So What we do is we have a very robust set of AI evaluation tools internally.
00:21:42
Speaker
Those are both ah reactive and proactive. So proactively, it's monitoring every call and flagging anything that goes wrong. for a human reviewer to take on board.
00:21:54
Speaker
And then we also have a set of reactive tools. So if if a prompt is particularly not performing well, we can put it through our suite of tools to kind of review past calls, everything that's gone wrong, best practice.
00:22:08
Speaker
And then in theory, it should output a much better version of the prompt that we can then test on an ongoing basis basis. But at this stage, we still have humans in the loop. And I think that's critical because sometimes um humans will pick up things that the AI won't. And I think it's critical to have a human reviewer really understand the nuances of what the customer wants um in order to be able to bake that into the instructions for the for the agent.
00:22:36
Speaker
And so when you say ah these calls may be recorded, you actually mean they are all being recorded and analyzed and reviewed. That's right.
00:22:46
Speaker
Absolutely, yeah. Most businesses are going after the enterprise market, particularly from a SaaS perspective. ah You went after tradies. Why that market in particular? or why make that?
00:22:58
Speaker
ah Was it a conscious effort to go small rather than large? Our initial assumption was that this product would be best suited for small to mid-sized trades businesses.
00:23:10
Speaker
What we found out pretty early on is that the enterprise tier in the trades market actually needs this just as much as your small to medium size. So we've kind of stretched o ourselves across the across the board.
00:23:23
Speaker
um So to give you an example, we've got one customer who does over 200 million dollars in in revenue a year. And they're effectively you know a big call center. They coordinate jobs across you know hundreds of different tradies.
00:23:36
Speaker
um And their view is you know this can be this can become kind of a frontline and out of hours resource that they tap into, um especially since they have a lot of churn with a lot of students coming in and working for their call center.
00:23:51
Speaker
So they're constantly thinking about how to leverage AI more within their call center to help alleviate some of those pains. So we've kind of gone across the board. Obviously, the deployment for an enterprise customer is very different to what we're doing for a smaller customer.
00:24:04
Speaker
We try and have a relatively you know white glove approach for everyone. But when you're working with an enterprise customer, there's a lot more bespoke integrations that need to to happen. And so we we yeah we resource that accordingly. Yeah, and it's worth your time from a money perspective as well, right?
00:24:21
Speaker
That's right. So let's move on to, I guess, a big theme on on this show is around the idea of making sure that AI is enhancing rather than replacing human potential.
AI's Role in Productivity
00:24:33
Speaker
Is there an argument that you are, you know, taking jobs away from local admin staff or office managers? And how do you navigate that? Like one of the ways that I like to think about this is I've never met a business owner that's gone out of business because they didn't have enough to do on their to do list.
00:24:51
Speaker
Do you know what I mean? And so the way we think about it is what our AI does is it will unlock resources in certain areas, right? So generally, a receptionist or front of house staff, they don't just answer the phone. They also do 30 other things in the background. They're coordinating team members, or they're also looking at the taxes. They have a million jobs that they need to do. And everyone has additional things that they'd love to do in order to grow their own business, but they generally don't have the time to get around to doing them.
00:25:26
Speaker
So the way we like to think about it is we're doing two things right with this voice receptionist. We're both freeing up the time of that person on that particular problem, and we're allowing them to capture additional value from their customers because you know there's times where receptionists you know at 7pm, they might not be picking up the phone.
00:25:46
Speaker
So there's calls that might be that might have been missed previously. And then what we're thinking about longer term with the product is how do we look at other challenges that those businesses are facing? you know So it could be things like quoting and invoicing.
00:26:01
Speaker
And how do we automate a lot of those, too? um I always think I'm i'm you know very bullish on on AI. I think it has a lot of potential. I don't believe people who who say it's going to replace everyone's jobs.
00:26:14
Speaker
like I see how we use it internally as like ah as ah as a small example and it enhances everything that we do, but we still need people to do those things. It's just that we can now do a lot more because we have this tool or at our disposal. it Slightly off topic, back in my advertising days, I did a huge quoll research piece with tradies and the big insight was that it was a majority their wives and girlfriends that ended up being stuck with the admin.
00:26:41
Speaker
um unpaid. So perhaps you're, ah you're solving a problem of, of free labor as well. Yeah, that's exactly right. That's, that's what we hear a lot as well.
00:26:52
Speaker
And, you know, like I said earlier, either it frees up their time to do, to do other things, or it frees up their time to go work on another, on another business, on something that they want to do, they want to do more as well.
00:27:03
Speaker
Alexei, we're kind of living in a you know, I'm, I'm not sure if I'm the first one to say it, but maybe like Sasspocalypse, you know, where everybody in the world is going like why have a Sass business? I've just coined right now. It was Sasspocalypse.
00:27:21
Speaker
You know, how do you react to that? What's your answer to that? I'm sure you had to um discuss that with VCs as you were seeking investment. um Look, I have two kind of competing views on this.
00:27:35
Speaker
Um, On the one hand, I think it's going to be very difficult for existing incumbent companies to completely re-architect their product with an AI first mindset.
00:27:51
Speaker
And secondly, to retrain hundreds of people or potentially thousands of people internally um to also also think that way as well. Right. Like if you think about a traditional SaaS company, it has tons of processes and guardrails and red tape that every new kind of development and deployment needs to go through, which slows down innovation quite a bit.
00:28:15
Speaker
Whereas a startup, you can be, you know, launching three new features, you know, in ah in a day potentially. So that's the that's the one side. Now, the other side is Distribution is key, right? Distribution is so important. And a lot of those SaaS companies, you know they've cracked the distribution piece. They have you know thousands of customers and sales is is challenging. you know It's really hard. So that's you know one of the things that they've got in their camp is um is is is the contracts they already have with their customers.
00:28:48
Speaker
So I like my personal view in terms of the way it's going to play out is i don't think it's going to be an overnight. Everyone switches to AI native tools. I think it's going to be a progressive thing. I think um I think a lot of SaaS companies are going to slowly decline as ai native solutions take more market share. And I think the ones and I think some of the ones in the market are going to be able to make the transition. And I think they'll do really well.
00:29:14
Speaker
And so have you ah priced Chime Labs in such a way that you think you're going to be a more competitive product for tradies than it would be for them to create their own SaaS version using ai tools in three years time?
Custom AI Solutions vs. SaaS Products
00:29:28
Speaker
You know, there's a perception if you're not like actively building on AI tools, is' a perception that you can as a sort of, you know, non-technical person, just like one shot an entire tool into into life. Yeah.
00:29:44
Speaker
you know As much as I wish that were the reality, there's still a lot of constraints on why that doesn't currently work and probably won't be solved anytime anytime soon um So I think there's a lot of value in you know understanding the customer base, understanding their challenges, using the the insights that you gain from having a lot of customers and distilling those so that every customer's performance improves um over time and just having the you know the resources to um to be able to to maintain a product over time because
00:30:22
Speaker
The other cost that comes with building your own product is that you actually have to maintain it, right? Like things are constantly breaking. There's edge cases that you need to solve for. um And spinning something like a prototype up can be really, really quick, but then maintaining it over time can be really, really expensive.
00:30:40
Speaker
I think the other thing that people don't discuss enough when they're talking about this topic is particularly in highly regulated spaces. So if you're thinking lawyers and doctors,
00:30:52
Speaker
um AI can probably never fully take over a lot of those services. And the reason for that is just insurance, right? Like Claude and ChatGBT, they're not insured to give you financial advice. They don't want to give you financial or health advice.
00:31:09
Speaker
You need a professional who's certified and insured to actually do that, right? So there's one value that will probably never get like knocked off. Do you see, i I noticed this phenomenon where the the less someone is experienced with AI, the less they understand about it, the higher their sense of what's possible and how easy it is to deploy. Is that something that you come across?
00:31:30
Speaker
Oh, a hundred percent. Absolutely. And then they'll start playing around with some tools. They'll get really excited. and then as soon as they, you know, start digging a bit deeper, they'll get really worried. Disillusionment off. Yeah. That's right. Yeah. Yeah. They get very disillusioned. So there's definitely that, that, that curve. Now that I'm working for a very large multinational business, I have infinite tokens, which means that I can just mess around as much as I want. And I've been building.
00:31:56
Speaker
But again, to your point, it's not a one shot and done kind of thing. It is a constant evolution. And I'm still mostly building slop. um you know A lot of it is is or it's a very specific tool for a very specific problem. like It's going to take me a long time to get to you know being able to build something that I can deploy as an application.
00:32:16
Speaker
Yeah, I think the ongoing maintenance piece is one of the critical ones. But I do see an argument for customized software. You know, if you have a very pointed problem,
00:32:28
Speaker
um like This is kind of my my thinking at the moment, right? like So I wanted to create, to stay on the advertising topic, I wanted to create a dashboard for our own advertising.
00:32:40
Speaker
So one of the considerations that i had to make was, do I pay for SaaS solution that already has pre-baked ah dashboards in place or do I build something myself? You know, so there's a balance there. I think sometimes if you have a very kind of narrow problem that you need to solve and you think you have the capabilities and you have access to the data and you can solve it yourself, sometimes it's better to just build yourself and that'll probably increase over time.
00:33:06
Speaker
um Other times you just want a reliable solution that's well maintained, that's secure and you're willing to kind of pay for for for those services.
00:33:17
Speaker
Well, just ah I mean, I guess the follow-up question really is that there's a this real kind of trend around the idea of everyone's going to build their micro SaaS product to meet their own very specific unique needs. but There's a real argument that what you need is a product that's robust and can scale.
00:33:36
Speaker
And I guess, where do you come down on the idea that you're building a very targeted and bespoke platform, but also you want that to reach a big customer base?
Global Expansion of Chime Labs
00:33:46
Speaker
Yeah. The challenge that we're solving for isn't constrained to just Australia and New Zealand. It's basically a global problem.
00:33:54
Speaker
So our ambition is to go overseas relatively quickly. um So, you know, the world definitely isn't lacking trades businesses. There's, you know, and they're, if anything, they're more and more important to the economy every year.
00:34:11
Speaker
um So I think there's ah there's ah an enormous market out there that are all facing the same problems that we can help solve for. It's really nice to hear that Australia is the um the innovation lab, though, for that problem. I think we are a country that has extraordinary capabilities in this space. ah you Do you get much support from local innovation labs or or government funding?
00:34:36
Speaker
Yeah, so we've tapped into a little bit of government funding. There's, you know, the the the startup scene in Sydney, I think, is thriving. It's growing more and more every year.
00:34:47
Speaker
Before I left Google, I was already involved with startups through a project that I was working on at Google. So I could see kind of the evolution over the last, you know, four or five years of how um how the VC network, the the startup networks are all evolving. And it's definitely heading in in the right direction. And I think Sydney's a very, or Australia just in general is a very lucky, a lucky country. It's a lucky place in that it's, um you know, financially quite well off.
00:35:19
Speaker
There's a lot of talented people who want to live here and proactively come here. um And in our particular case, it's also trade heavy economy, right? Like the GDP of Australia is very heavily focused on on housing.
00:35:36
Speaker
And as a result, the trade economy is alive and thriving. There is an argument, and I'm starting to question this myself around what does the future look like for our kids and and you know the next generation to hit the workforce. what did the yeah Would your advice be to build a an AI platform or to become a plumber?
00:35:59
Speaker
Either one of those is good advice, I think, actually. Yeah, I think they're both actually really good pieces of advice. Like the way I think about AI more generally, because i think a lot of people are worried about this is I think it'll just become a tool that everyone uses. Right. In the same way that everyone knows how to use Microsoft Word, everyone's going to know how to use AI tools.
00:36:18
Speaker
um A lot of a lot of people, i think early on, we're talking about, oh, what are young people going to do that are coming out of college? They're going to be out of a job because all those entry level jobs are going to be automated by AI. I don't think that's true at all.
00:36:32
Speaker
What I've seen is the ones who are most willing to learn how to use AI are young people. Right. And when I make hiring decisions now, I make hiring decisions based on, you know, the potential that we see in that person and their ability to use AI. That's going to be a critical hiring decision in the future. So I think.
00:36:52
Speaker
um I think just learning those skills is definitely going to become critical or just become a tradesperson. That's going to be super valuable as well in the future. so ah After the apocalypse, we'll still need plumbers, right? That's right. Exactly. well maybe And maybe that's the perfect segue into like, you know, why Chime Labs is the perfect product for everyone, right? Because ah we're going to be a ah world full of ah plumbers and electricians.
00:37:16
Speaker
yeah and Alexei, what's one thing that we haven't asked you that we probably should have?
New Product for Google Ads Campaigns
00:37:23
Speaker
maybe a little bit around the the evolution of the product. um You know, voice, like we yeah obviously we started with with voice. It was kind of one of the the key things that we we saw was a challenge in the market, but we're going well beyond that now. So we just released a product in the advertising space.
00:37:39
Speaker
um One of the things that we hear a lot from tradespeople is they also want to generate more leads for their business. And right now, either they need to learn ah Google ads, or they need to pay for, you know, specialists in the space.
00:37:56
Speaker
So we built a solution that allows them kind of ah the best of both worlds, where essentially in three clicks, they can get A Google Ads campaign, a Google Maps presence and a website spun up all using information that we know about their business and all run and optimized using AI as well as like Matthew's and my like internal kind of best practice knowledge about how to launch these types of campaigns.
00:38:26
Speaker
um So that's a really cost effective way for them to get up and running and to start generating leads immediately. Right. So they'll they'll end up with, um you know, an ads campaign and a landing page that ties into a Chime Labs receptionist that just automatically starts booking new jobs into their into their calendar. And what's the kind of price point are we talking? are we talking, you know?
00:38:45
Speaker
Hundreds of millions of dollars for one of these things? or They can start as low as $500 per month, you know and that includes the the kind of ad spend as well as the the Chime Labs fee. um And then they can scale up as as much as they want.
00:39:00
Speaker
It's interesting because wasn't the kind of proposition to small and medium-sized businesses that Google made was that it was self-serve and really easy to DIY it, but you're seeing that not connecting?
00:39:13
Speaker
um Having worked at Google for a long time, I can tell you they're still very far away from that. Right. It's still. it still ah It still requires um you know a bit ah ah bit of time and knowledge kind of commitment to get something up and running DIY.
00:39:32
Speaker
well And to do it well. ah Exactly. you can You can waste money really quickly on Google yeah if you're not careful. And all advertising channels. all that To be fair, it's just a google just a Google problem. I love two former Googlers talking about how much money you can waste on Google. It's...
00:39:50
Speaker
Well, we both still have a fair amount of stock, so we we both still love Google in many ways. This message absolutely not brought to you by any partners. Alexi, thank you very much for your time today. Kat, anything else?
00:40:06
Speaker
um I think it's always good to ask the PDOOM question. Where do you sit on the on the optimist or pessimist scale about the impact of AI over the next few years?
AI's Future Impact on Business
00:40:16
Speaker
Yeah, I'm quite an optimist. You know, I don't think it's going to put everyone out of a job.
00:40:21
Speaker
um I think it'll end up being a tool that we all leverage to just do a lot more, you know. um And you see this when you're vibe coding something.
00:40:32
Speaker
It's become so easy now to develop um to develop a pointed solution to solve a problem. As soon as you do that, you think about 10 more things that you could do around that.
00:40:44
Speaker
Right. And I think As soon as more and more people start realizing that, playing with these tools and leveraging it, it's going to unlock a tremendous amount of growth. So I'm actually really, really positive about it.
00:40:55
Speaker
One more question for you. Finish this sentence. In three years time, ai will be blank for small businesses. AI will be...
00:41:07
Speaker
you know, indispensable for small businesses. I think it'll, I think it'll, it will be something that if they tap into, they'll get so much leverage from it, um, that it will allow them to do things that they previously thought was impossible for their business. Right. It'll allow them to grow much quicker and much more effectively. And probably the ones who are going to benefit it from it most are the ones who adopt it earlier. Right. Because eventually it'll become commoditized.
00:41:36
Speaker
But I think we're in a moment where small business owners have an opportunity to tap into these tools and think creatively about how to use them um to differentiate themselves.
00:41:48
Speaker
So like a true optimist. Alexei, thank you very much for joining us on Mind the Model. Thanks a lot, guys. I had ah had a good time. Thank you.
00:41:59
Speaker
Kat, what did you think of that? think it's such an interesting change of pace. You know, we hear so much about these big enterprise platforms and products. It's great to hear ah an AI product that's been really developed with its audience in mind and that kind of deep insight into the use case for the trading market. And I'm just a huge fan of hearing about innovation coming out of Australia. We are a country that punches above its weight in innovation terms, and I think we should celebrate that.
00:42:27
Speaker
Yeah. I mean, look, i like I said, I go way back with Alexi. um Always been a very smart guy, as has his partner, Matt. um But I'm just excited that they're building this here in Australia. It's another great example of Australians' startups just really creating interesting things that I think have real value for normal human beings, which I'm not sure that you can say about everything AI related these days. By normal human beings, do you mean the non-magical people who don't work in the industry? Yeah, people who are not like us who spend way too much time on their feeds. I mean, you know, Alexi said that he said, you know, plumbers are getting a lot of AI stuff. And and it's interesting, you know, when I was home last, people were, they're aware of AI, but they just don't know how to implement it into their daily lives. And so a product like this that I think takes a lot of that pain away, to Alexi's point, you know, people probably aren't going to get real value trying to vibe code this themselves. Yeah. And I think it's really important that we remember that what we're actually looking at when people are thinking that they understand AI is that they're basing that on an an interaction with chat GPT, where they've asked it for a recipe, you know, and there's such a huge gap between what that capability delivers for, you know, the average bear versus what's possible with more complex tools and platforms. And, you know, and I guess this probably why people are now valuing Claude so much more than ChatGPT. ChatGPT has gone after that, you know, ah individual human being market in so many ways, whereas Claude having gone for the corporate market. It just feels like there's a different mentality there.
00:44:02
Speaker
Yeah, but i I mean, I think it's still in terms of just sheer weight of numbers, the the the majority is still sitting with GPT as their experience of AI. But I've got a ah ah wee hunch, which I'd love to run by you. I reckon in 12 months time, we're going to see the rise and rise of Gemini.
00:44:19
Speaker
Last year was all about open AI. This year is all about Anthropic. I reckon it's Google's turn for a go at the top just because of the the huge user data that they sit on. Yeah, look, I would never, ever count Google out when it comes to technology. um And especially from an AI perspective, they were one of the very, very first to do it. um I have to throw my hat in the ring and say, you know i look at for other partners as well. you know i think don't discount the Chinese models. um And i know that Microsoft's going to be building big, interesting stuff as well. And I have to say that for legal reasons.
00:44:57
Speaker
But um like, yeah, there the this game is not over by any means. What a fascinating time to be alive, to watch the the the highs and the lows. Well, again, Kat, thank you so much for jumping in on last minute notice. um I hope the experience was fun and exciting for you as much as it was for me. Absolute pleasure. Thanks so much for inviting me.
00:45:20
Speaker
um And for all of our listeners, you know, I'm a big fan of behavioral economics. So I always like to say that, you know, 99.9% of all of our listeners ah recommend the show to a friend. So should you as well, if you haven't already. um And be sure to share, subscribe, like, and send any questions you might have about the show to mindthemodelpod at gmail.com.
00:45:41
Speaker
And remember, the intelligence might be artificial, but the wins are real. See you next time. Bye.
00:45:54
Speaker
This episode was brought to you by Testchimp. Testchimp is a web developer's best friend. Testchimp uses deep runtime context to automate your testing workflows, generate scenarios, write scripts, and fix bugs all in one place.
00:46:10
Speaker
Stop wasting time and start shipping quality code faster. Bugs suck, but finding bugs doesn't have to with Testchimp.