Build vs Buy - Should You Build or Buy Your AI?

Ah, yet another big week in AI. This is The AI Exhale, hosted by Vibe CFO with Jez and Cam. Five minutes, one thing that matters, or maybe two. We ignore the rest so you can too.

Jez: Today we're going to talk about the build versus buy dual pathway to tackling AI. Cam, take it away on some initial thoughts around that, and we'll just go back and forth. Maybe start by explaining what building or what buying is, and then we can go into a few of the pros and cons perhaps.

Cam: I reckon the build versus buy is just becoming such a big topic. Like, you see it everywhere where people go, oh, I've built this now, so that means everyone should look at what I've built. And the problem with that, Jez, is that people have just literally single prompted and built something, right? Because these coding assistants can just build stuff now, and we're not thinking through. None of these people have any coding ability.

They're literally just white collar workers who've decided to punch this into a coding assistant. And then they're saying, this is the direction everyone's going to go. And nobody's talking about the consequences 3 months down the track. Is it useful? Is it broken? Are there security issues with the way you're doing it?

Like, there's so much DevOps and servers and databases and security. There's not a lot of thought going through this. It's just literally vomiting social media posts to get attention on what people are doing with AI. So the build versus buy, don't you reckon, Jez, is just such a big topic at the moment?

Jez: It is a big topic. And as you say, we see it on LinkedIn, in terms of people posting that they have built an MVP, I guess, at the very least. When you say it's a difficult journey from there on, DevOps, it's their job.

Cam: You know what, my thought is we don't know, right? We don't know how good it'll get. You know, there's this expression, the AI is as bad as it will ever be. That's a common phrase that we've heard over the past few years, right? So every day, like in the next month, a new model will come out and it'll be even better. So people think, okay, that means I can code it myself. What's happening here, right, is when you're trying to replace software and you think you can build it yourself, but then you think that you're going to overtake software, that's like, I don't know.

Something about that doesn't sit right. That means basically you're saying to the software people that they're going to sit still while you overtake them. And I'm like, software people aren't going to sit still.

They're going to keep building using the same tools that you've got, and they've got more time than you, because you're running your business. They're running their business, and that means building these tools that you're single prompting.

I reckon the danger is, right, if you think you're good with AI. This would be a trigger for me. There's something I was thinking about. If you are sitting with a red wine in the evening watching the coding assistant build something for you and thinking to yourself, wow, how good am I? I say that's a red flag. Because let's be honest, you're not doing f*** all. You're not doing anything.

You're just sitting there watching it do it, and you actually don't even know what it's doing. And then all of a sudden you've got this thought in your head that you are a data engineer or a developer. That's nonsense. You're just single prompting something that you don't quite even understand.

Jez: I like red wine. I don't like red flags. What about the people who do spend a bit more than the prompt? I mean, there's people who are building stuff, launching this, sharing this, not just on LinkedIn, maybe even at some small bespoke events at the moment. And there is something in there, if I may say, that allows people to think a bit deeper about the use cases and applying better thinking and processes to work.

So again, we're not discouraging it. But maybe there's something there for some people to try and go and build stuff. Are there any sort of tips if you wanted to usher people in that direction?

Cam: Yeah, I mean, that's the nuance, isn't it? There is software that needs to die, and there is software that you can replace. And so this is the grey area that everyone's dealing with here. What could they build and what shouldn't they build? How far should they go? It saps time, it costs heaps in AI tokens. So how far should you go? Every demo I've seen, of people getting up on stage or presenting online, they're like, I've built 40 apps that run my firm.

And I look at the 40 apps and I think, it's like a GST calculator and an income tax calculator and a form that people can fill in to register interest in your firm. I mean, it's all just total sh*t, really. These are crap tools that I don't think have got a lot to do with actually running your firm, where you get proper return on investment.

And that's your five-ish minutes back. Vibe CFO, Australian founded. An AI CFO for small and medium businesses, out of the box, governed and secure. vibecfo.ai. Now you know.