How AI Is Transforming Software Engineering (Isaac Martin)

Full transcript

Welcome back to the podcast guys. Today we are joined by Isk Martin, a acting CTO at Planned Bit. Isk, welcome to the podcast. Thank you for having me. >> To start, could you share a quick version of your background and what Plant Bit does? >> Sure. My own personal background is in startups. I've been doing tech startups, mainly SAS companies, for 15 years or so and been in tech for about 25 years. Plant Bid sells commercial landscaping supplies to commercial landscapers. So, we're just a B2B company. If you aren't in the commercial landscaping space, you probably haven't heard of us. >> What changed for you going from running engineering to actually owning the bigger technological vision? >> Sure. So, I haven't really served as an IC for a very long time. I've been in leadership roles for probably 15 years or so. But those leadership roles have been at startups. So in any good startup, a CTO is doing some IC work typically. I mean even if your teams are getting pretty large, I think it's wise for a CTO to be continuing to build things themselves so that they can keep a connection to what is the reality of software engineering today.

You could maybe argued 15 years ago that this was less important because the trajectory of software engineering was pretty flat. But over the last four years, the nature of software engineering has changed radically. It's changed in ways that the field has never seen that kind of change in the past except maybe moving from, you know, punch cards to dealing with digital information instead, you know, instead of analog information or something like that. But that didn't happen in our lifetime. >> And what's the most drastic change that you've seen? >> Sure. Recently, we had large language models develop themselves to the point where they're useful for software engineering. That happened, you could say about 3 years ago or so is when they were minimally useful. Prior to that, you weren't getting a lot of use out of them, but about 3 years ago, they started to become useful.

And then maybe a year ago, you were in a position where they could really accelerate your velocity by actors, you know, 2x, 3x or something like that. as recent as six months ago or even the just the release of Opus 4.6, you again see a meaningful increase in capability where now when I write software probably 90% to 95% of the code that I'm committing keystrokes were done by an AI. I mean I know what is being committed. I read it and make sure that it's doing what it's supposed to do. But I don't type a lot anymore. I'm interacting with an AI a lot using Whisper Flow to do speech to text and then doing lots of planning and making sure the plan is perfect and then having an agent execute on the plan and output something useful. >> I must say Whisperflow is great for just about everything like not only code but even writing emails or just browsing. >> Yeah, they just released an Android app. So if you use Android, you can use Whisper Flow on your Android now.

I love Whisper Flow. They're so good. >> How did you find out about Whisper Flow, >> man? I don't remember. I've been using it for a long time. Probably somebody on the internet said something and I thought, "Oh, yeah, that sounds good." And I tried it and and it's amazing. I I hardly type ever. I I'm almost always just voice driven stuff. >> And what what do you think is the future? We have Open Claw right now being a very hot topic. The further out in the future you go, the harder it is to predict what's going to happen. And if you're talking about 15 years ago and you asked me what's software engineering going to look like in a year, I would have been able to give you a pretty reasonable assessment of what it was going to look like and what good practices are and things like that. Now, I can tell you with confidence that AIdriven software engineering will just gain popularity and market share. I think there are probably plenty of engineers today who don't do AIdriven software engineering. That'll change over time.

We will get to a point where you aren't going to be hired by a company if you don't do that because you're so inefficient that the companies don't want to hire people who don't do that. So whether people like it or not, they're going to be in that position in the next couple of years. How quickly will that happen? It's pretty hard to say. Yeah, I mean the gains are pretty extreme right now, but humans change only propagates so quickly. There's still a limit of the the larger the company, the more difficult it is to change process and the harder it is to get a 100 people to agree that some process needs to change. So large companies will change slower and be less adaptable. Smaller companies will move faster, kind of the way it's always been. But specific predictions for a year from now, it's pretty challenging.

I just think the current trends will continue to snowball. Going back to planned a bit, what's the real pain that you guys are solving for contractors and suppliers? Well, imagine that you're a large commercial landscaping company and so you're doing millions or tens of millions or hundreds of millions of dollars a year in jobs. So maybe you're responsible for all the landscaping in a city because the city has to contract with somebody to do its landscaping and you're the company that does that. and you're, you know, talking about some massive city like Austin or Dallas. Big, big cities have lots of landscaping needs, especially when they're in environments that are hostile to plant life, deserts, for example. So, if you're somebody like that, you need to source a lot of plants and those plants need to get to the job sites, right, intact. And because plants are alive, you can't just put a plant in a warehouse and let it sit there until somebody needs it and then pick it up and bring it to them. The plant has to be shipped in a specific way that makes sure that that plant is going to be okay on the trip and survive the the experience of being transported.

Plants change constantly because they're alive, right? You could have a frost or something kill a bunch of the plants in your stock. It's a much more dynamic environment than for example what Amazon experiences in trying to bring iPhones from one place to another. That's a lot easier. So what we do for our clients is we handle everything. So a client says we want these plants and we want them at this place at this time and we make sure that that happens. And there's a lot that goes into that. So the client is paying for a lot of convenience. they would otherwise have to deal with all this these logistics and finding the plants, sourcing them. There's no one nursery that grows all the different plants, right? You have to go through hundreds of nurseries or maybe thousands depending on the variety of order that you've got going on. >> And when you're actually modernizing a supply chain workflow, what is the hardest part of it?

Is it data or matching? >> The hardest part is definitely not the software. It's like physical reality like when you're creating systems that involve physical reality and physical reality beyond like servers, right? So SAS applications run on servers and those are physical. But if we're talking about any kind of SAS company that deals with physical space, I did a SAS company earlier in my career that was an IoT company and that company had to install lights and locks and thermostats and all sorts of physical things in the world in order for the platform to work because the platform controlled those things. So in any company that is involving physical things, it's the physical things that are the most difficult part of the company because you can't scale physical stuff by just spinning up more servers. It's really easy to say I want 50 more servers today and as long as you can pay for it and as long as your provider can do it, which of course you any hyperscaling provider like Google or Amazon can do that very easy. One engineer can do that in a couple of minutes. But if you want to scale to 50 times the amount of trucks that you have to deliver, you have to go buy 50 times as many trucks as you had before and move them into the right spot and make sure like they're getting maintained.

I would like to touch up more on leadership since that's something that you deliberately built. What are the two three principles you use actively to keep teams executing well as the company scales? Yeah, if you're going to try and collapse leadership into two or three principles, you got to get to a pretty high level of abstraction. So, if you're talking about leadership, you're talking about a group of people. You can build a company as one person, but then it's not a leadership problem. It's, you know, you're solving other problems. Leadership problems exist when you have groups of people. You're trying to orchestrate them together to accomplish some goal. And so if you're trying to do that, then the things that become important are rules of thumb around people, right? So your people in order to be effective, they should probably be happy. Happy humans are more effective than unhappy ones. People like Stalin and Kim Jong-un might disagree.

They follow processes that drive derive value from unhappy people. But I think that you actually get more value per unit of person out of the happy people than the unhappy ones. So you want to really like try and design processes and create teams that result in people that are happy and that like pushing into the outcome that you want. So for example, let's say you're trying to build an engineering team and you're trying to build like some simple SAS application, some just single page application website. There are a lot of engineers that can build a single page application. But you don't just want an engineer who has those skills. You want an engineer who has those skills but also is close enough to your culture that you can operate together successfully and you're going to enjoy each other's company. Like for example, if I tried to build a SAS application with like a Dothraki from game of like their lives are riding horses in the plains and burning cities to the ground and etc etc right like their culture is very different than mine and we probably wouldn't get along very well if we were trying to solve problems together. Even if you found a Dothraki that can write single page applications pretty well, we probably wouldn't jive very well.

So there's some level of cultural alignment that you want to align people with your company's culture so that they will be happy executing, right? They you want them to be happy. So you want to create an environment that makes them happy. So an environment that they'll be happy in giving them problems that they're going to learn a bit from. You don't want to just have somebody who is the most senior person in the world doing things that they've done a million times because they're be bored and boredom does not make people happy. So, you've got to hire just the right skill set that they can do the thing, but they're also learning a bit while they're doing so that they are engaged and they feel like they're growing. Most humans are happier when they feel like they're growing uh and learning. >> Okay. And what are you guys focused on proving this year?

If we fast forward maybe 12 months, what would be a big win for Planned Bit? Well, Plantid has some growth goals. We exist across a specific geographic area and our goal is to ultimately grow to encompass the entire United States and then hopefully beyond it. So, in order to accomplish that, we need to expand at a certain rate and that's what we really focus on is is trying to achieve those growth goals. For people actually listening who maybe want to learn more about Plan Bit or just follow your work, where should they go? >> If they're trying to follow me, then probably LinkedIn is the way to do that. So, I post to LinkedIn every once in a while with something that I think people might find useful. So, if you find me interesting, then follow me on LinkedIn. Learning about Plant Bid, probably just the Plant Bid website.

You know, it has some information about what Plant Bid is. And if you're interested in buying 200 trees or something that we can help with, there's some contact information on the website that you can follow. >> Perfect is I will add links in the description so people can check you out. Okay. Thank you guys for watching and we'll see you in the next one. Thanks for having me.