mobile banner

Jon Bradshaw & Peter Harris

What is the effect of ChatGPT on startups? Though ChatGPT can have a positive impact, it will also shake up many industries. Here is a quick summary of the areas we think ChatGPT will affect/change the soonest.

How is ChatGPT Affecting Startups?

What is the effect of ChatGPT on startups? Though ChatGPT can have a positive impact, it will also shake up many industries.

Here is a quick summary of the areas we think ChatGPT will affect/change the soonest.

  1. Customer Support: Startups can use ChatGPT to provide 24/7 customer support to their users. ChatGPT can handle a large volume of customer queries and provide quick and accurate responses, which can lead to higher customer satisfaction and retention rates.
  2. Product Recommendations: ChatGPT can analyze customer data and provide personalized product recommendations. This can help startups increase their sales and revenue by providing customers with products that they are more likely to purchase.
  3. Market Research: Startups can use ChatGPT to conduct market research by analyzing customer feedback and reviews. This can help them understand their target audience better and make more informed decisions about their products and services.
  4. Content Creation: ChatGPT can assist startups in generating content such as blog posts, social media updates, and email newsletters. This can help startups save time and resources on content creation while still maintaining a high level of quality.
  5. Virtual Assistants: Startups can use ChatGPT to create virtual assistants that can handle various tasks such as scheduling appointments, sending reminders, and answering basic queries. This can help startups streamline their operations and reduce their workload.

What do you think? Follow the links below and let us know.

Hosted By

Episode Transcript

 
Jon: Are you ready, Peter? Always. All right. Welcome to the Venture Capital podcast with Peter Harris and John Bradshaw. Peter has his experience. I have my angel investor slash founder experience, and we try to give you the best answers on venture capital in the news. So for this episode, it's a two part series specifically on how Chat GPT is affecting software development, where I come from.
 
Peter: And startups.
 
Jon: And startups and how it's affecting the venture capital world. So join us if you want to make sure you're subscribing to venture capital firm. You can find us on YouTube, Spotify, all the places. Are you ready? So how is chat CBT affecting START in software development?
 
Jon: So this week was kind of like a midlife crisis, not a midlife crisis, but like when you take a step back. So I've always assumed for those of you that don't know, I have a staffing development company called codebased.com and I've assumed that when I I'm 39 right now. I was I've always assumed that when I am 70, 80, 98 is going to be a really good where the need for 90% of the developers that are currently on the market are no longer needed.
 
Jon: So companies that used to have 100 developers need ten developers, and I've always assumed that would happen when I'm 70 or 80. So like maybe in another 40 years. And Chat GPT was the first thing that I've seen in the market that's really caused me to take a step back and say maybe that time horizon is not 40 years, maybe it's ten, 15, 20, maybe as early as five.
 
Peter: So, John, first, like it'd probably be helpful to explain why you think it's such a threat. And also, like, what is chat speed for those that live under rocks?
 
Jon: Yeah. So, yeah. Now, now you're interrupting me. Like people say I interrupt you on the podcast.
 
Peter: Just let John speak. No, nobody wants to listen to you, Peter.
 
Jon: They're usually saying, Let Peter speak. That other guy needs to shut up. And then my favorite thing is to respond in the comments. I'll be like, I'm the guy on the left. The kind of left needs to shut up. I'm the guy in the left. So chatty. Betty how would you, how would you describe chat GPT? Maybe I'll ask you that question.
 
Peter: You can't turn it back on me.
 
Jon: All right, so then I'll do it. So chatty, but basically if you go to chat dot open a dot com, it's a partnership on the Open AI network where instead of going to something like Google where you could say, answer this question, where is this? What is this? How is this? You can actually engage with it. So I could you and you can ask imagine asking Google a series of question.
 
Jon: That's it versus refining your initial search. So right now, I might say, what are the best coding frameworks for startup today or coding languages? And I can ask chat GPT that and then it could come through and say.
 
Peter: What are the pros and cons of each.
 
Jon: Word?
 
Peter: And it would like understand you contextually and it would like list out like here are the top languages and here are pros and cons of each and how each gets used and all of that stuff and write it into an article that theoretically you could take, make some minor edits and publish. Right?
 
Jon: But then I could go a step further and say, remove anything that's page related. Yeah. And then I could also say another question. Show me something that's really good for the medical space or with that works with telephony software or different applications. And so your search can become very fast refined, and I can interact with it like interact with you.
 
Jon: So it's basically like saying, Hey, Google, these great results are great, but remove this, add this. Here are some additional context.
 
Peter: How we can do it in like natural language as opposed to having to use.
 
Jon: Boolean searches. Yeah. And this exclude this because no one knows. Not no. One. 99% of Googlers don't know how to use a Boolean search.
 
Peter: Right. Right.
 
Jon: But you can also solve programing equations. And if you look at the use case on chat, GPT, say the first the first demonstration is this application, this language processing search engine kind of understood landing problems that are in English and converting it into code. And in fact, the first time I heard about chat CBT was from a developer who's like, John, there's this guy who does like the 12 Days of Christmas.
 
Jon: He releases a new programing challenge every day and it solved it in 30 seconds. And so then probably for the last week, 90, 90% of my TikTok consumption is examples of people showing how they're using chat GPT to solve programing things.
 
Peter: And to speed up development.
 
Jon: And to speed up development. For the most part, my summary is right now it would be compared to like a high schooler. Sure. So it's not complex, but it's like, Hey, is this the next big breakthrough in development?
 
Peter: Yeah. And not just break in, not just development, but like, I mean, if you think about software and coding, it's really just another language, right? And so is this idea of creating content like software as just content. And a lot of ways and so so had GPT. What's interesting about it is that it creates usable content.
 
Peter: Right. With very little effort, very efficiently at scale.
 
Jon: So this is the part Jackie Beat does not scare John Bradshaw. But what I wonder about the future and really it's really the opportunity is imagine someone like Peter who doesn't know, who's never built an application himself. He could go to a designer and say, Hey, go design this application that maybe has like 50 screens. Here's my mobile app, iOS app, my Android app on my web app, and he could work with the designer to say, Hey, when I click this button, this goes here, This is maybe the types of information, but once that's all designed, imagine taking that like figma file or adobe XD.
 
Jon: Those designs by designer and over the weekend it would program it for a thousand bucks, which something historically might have taken $100,000 or $2 million to build. That's what I'm looking out for is when is I good enough to do that? And that's the startup context.
 
Peter: Yeah.
 
Jon: And that's when, you know, it could potentially majorly affect venture capital. It would affect companies like Code Base and the entire world.
 
Peter: So how would a how would it affect code base?
 
Jon: It depends. I think it becomes an opportunity and allows us to start getting higher profit margins because then you're doing a very specific skill set, working with a very specific application.
 
Peter: Yeah.
 
Jon: And I think but.
 
Peter: How easy is it going to be for somebody in that case really has no knowledge at all to come in and actually build the correct app and know that like, hey, this app is actually going to solve the needs of my customer, It's going to be scalable, it's going to be written in the right language, it's going to be all these things, right?
 
Peter: So, so mostly because a lot of cases you don't even know a lot of the answers to that. Right.
 
Jon: Right. But that's where I think air becomes interesting or could start watching. And instead of having a So what I see happening potentially is when or if it becomes smart enough. And I think the real limiting factor is the hard part about A.I. right now is it's just running an infinite number of models, but it's only really effective if you have a good data set.
 
Jon: And in startups, people can explain what they want well enough. So in the perfect world, you're really having someone who can who can know how to guide an AI engine. Yeah. Is what the future potentially could be. But the mostly applications that you're probably seeing right now scaling, they talk about scaling, but scaling is not a real factor.
 
Jon: And if it's a really good system, it could say, Hey, here's going to your B1, B2 V3, but it could reprogram when timing comes, right? You could say, Hey, it's time to upscale the application. You know, maybe there's a licensing fee of like 10,100, I don't know, whatever it is. And so the next version and then your team, your manual acceptance testers could then go through and test it and say, Hey, now let's have this go live.
 
Peter: Yeah. So my guess is that it'll be like there are a lot of like self-service things that already exist, right? Like I could go on and for a couple of hundred bucks a month, I can build out drag and drop, you know, Low-code no code app with.
 
Jon: Like bubbled audio there because there's a bunch I could create an Instagram competitor and not know how to code with bubbled audio.
 
Peter: Right? So these things already to a certain extent, but.
 
Jon: That's not a I.
 
Peter: It's not A.I.. But, but my point is, is that there are already solutions that present the threat that you're you're afraid of. Right. Or that you're right you're describing. But the flip side is, is that none of them are like really that compelling because you still need someone who kind of knows what they're doing to build something like, yeah, I could build like a very simple Instagram, but I'm not going to build Instagram bubble right?
 
Jon: Good. Why not?
 
Peter: But I wouldn't be it wouldn't be as scalable. It wouldn't be worth talking about.
 
Jon: What are you talking about? Most startups bubble. So I have my biases about bubble, right? Maybe confirmation bias. I think at the end of the day, these no code low code options look ugly.
 
Peter:

 
Jon: But again, it's just a matter of time before it looks better. Sure. But the way the world that I see bubble in is and in fact, I like whenever someone comes to code base, I look at them and say, you know, should code base have a no code, low code solution team? Because a lot of these guys don't want to go spend like, we just help someone who spent $120,000 building out an Iowa iOS app.
 
Jon: Sure. And an Android app and back stuff.
 
Peter: Yeah.
 
Jon: And so but for a no code solution, you could work with an architect for like three K test the model. And if it works, when you when you get to say maybe doing 10,000 a month in revenue, then at that point you should start considering moving off, assuming bubble can do what you want it to do.
 
Peter: Right. And I'm just saying, why does why do not all of those things still apply just at maybe bigger scale with a.
 
Jon: What's the point is. So the answer is.
 
Peter: Like it helps you. Maybe it gets you instead of to ten grand, it gets you to 100 grand. But eventually you get to a point where like architecting the thing matters right and right. But so, you know, they're like all of these things that start mattering more. And so ultimately chatbots are like AI in general is in in terms of how it's used with coding becomes like this enabler is it makes you more efficient, it makes you faster, it makes you all these things.
 
Peter: But at the end of the day, if you don't know the fundamentals of coding and customer behavior and a lot of these these things that really matter when it comes to actual product design, it doesn't matter. Like the guy's not going to really understand all of those nuances. And so I think like the big opportunity is you move from like pure, like coding farms or just a bunch of developers, you know, writing code to a more nuanced situation where it's like developers need to be more like product managers or they need to understand and like they need to be like architects.
 
Peter: They need to be product managers, they need to be a bunch of these things. And then they use AI as a way to like, speed up and fill in gaps. Like I think about outside of development, if you think about just writing articles or even podcasts, like, right, like theoretically I could be a threat to what we do.
 
Peter: But the flip side is the flip side is, is that I could also just be like this huge enabler to what we do and allow us to like up our game and do a much better job and do it faster and produce like better content.
 
Jon: Yeah, so it could be from a software perspective, I could just continue and for startups could just be continue needing to do what other innovations have done. So for example, ten years ago, if you wanted to create an e-commerce website, it very likely would have cost you 20000 to $100000 to launch today. You could do it on your.
 
Peter: Own for like 20 bucks a month on spot. On Shopify.
 
Jon: On Shopify, exactly. And then all of those developers who are working on on inferior tasks are now freed up to go work on more complex tasks. Right? So I most likely will just slowly enable developers be 5%, 10% to 10% more effective. And these developers can then say, well, what are more complex or more challenging problems? What I'm concerned about or the question is at what point I feel like I will will gradually get better but have major breakthroughs.
 
Jon: And at what point can I, with a team of developers of like say, ten developers do what I and a team of 100 developers could do. And at that point there's going to be massive change like a company like can try to could could decrease its dev size and try to as a local company with probably a couple thousand employees, they could lose their developers or but.
 
Peter: What about the argument that like there are a zillion dev tools that have been released over the years that have reduced the number of devs that have been necessary and yet all that's been all we've seen is an increase in the number of devs that are demanded right across the board.
 
Jon: Or or.
 
Peter: Because software keeps leading the world. And perhaps the bigger question is how much of the world is left for software to eat?
 
Jon: That is not a good question.
 
Peter: But because of that is like if you view it in one one, one view could be that it's just infinite. There is no end to the like the amount of software that the world could eat and so, or the other way around.
 
Jon: But you're also seeing a lot of more niche niche problems get solved. So you're seeing instead of if, if someone were to launch a CRM today, most likely they would not go after Salesforce. They would say, I'm going to create a CRM specifically for lawn care businesses.
 
Peter: Sure.
 
Jon: And that never would have existed before. Sure. And it's a much better experience in Salesforce because it's identifying the problems of specific users. Yeah, and that's what we're seeing, I think, happening as well.
 
Peter: Now there is a debate like is that venture bankable? Probably not. And so like maybe it just facilitates a lot more lifestyle businesses and maybe we move from like, hey, you know, the traditional lifestyle business was, you know, owning a restaurant or something, right? But now it's building using AI to build a CRM for lawn care businesses.
 
Jon: So for example, you know I on appointment dot com and one of when I look at like threats or concerns the founder of acuity scheduling talked about it was really hard for them to be competitive as a generalist when you have all of these other scheduling tools are popping up like in dental software. So for example, they said they're like our scheduling software is much better than what's out on the market, but a bundle tool with an inferior scheduling tool would beat them to almost every sale.
 
Peter: Yeah, because integration matters. Right. Yeah.
 
Jon: So these are the things that we're watching. So I know it's a it's a fun time to be alive. It's very dynamic and definitely chat as the developer side of me is like, Whoa, what just happened in my world? Yeah. And I think most likely it's going to be, you know, what we're looking at Code Base right now is almost like paired programing.
 
Jon: So three weeks before CBT was announced, there's a thing called GitHub copilot built by Microsoft. And the idea is that it's coding with your developers and it's trying to predict what it needs to do next. And right now my initial thoughts are it's an okay tool. And that gives you sometimes big wins, but it's only and it's beneficial to use, but primarily for a senior dev because I think it's really easy to go down a rabbit hole and just start following the air or the air thinks you want to go, not where you actually need to go.
 
Peter: And if you don't have enough experience to know where you need to go, then yeah, you end up in this maze and in the wrong spot.
 
Jon: And I think that's.
 
Peter: All that comes back to, like where I still think like devs that are good will be able to leverage this to become like.
 
Jon: But, but here's my other.
 
Peter: Superpower, right?
 
Jon: But here's my other really big example of where it could go, let's say, of a legacy application that's 50 years old with multiple languages. No one wants to talk just poorly documented. That might be in the banking system, right? Sure. And maybe this is a really big idea and not a threat, but there could be an API who could go in, look at all the historical logs.
 
Jon: It could watch the application actively assuming assuming no new changes for management, no new front end designs, it could technically flash the entire system over a weekend or a month max and then just have, hey, check this out, look at our coding and that's when it gets really exciting to me. When could chase banks say, Hey, we're going to significant overhaul our entire system?
 
Jon: Yeah, no new changes. All of the same rules apply. And at that point we would have near-perfect data to make very good recommendations.
 
Peter: See, this is where I think getting to specialization matters when it comes to dev. So if you are just like a generous developer, I think you're going to be a tough spot. But like, think about like the example you just laid out. I don't think you're going to be able to go to just any API platform and kind of guide it through that process, right?
 
Peter: But I think you could set up a dev shop where like that's what you do, right? And you, you know how to like what's the right AI, what tools like how's the right way to set it up? How is the right way to check and make sure that it's doing it correctly, fix bugs, etc. etc. and actually deploy it correctly.
 
Peter: Like I think those will be huge opportunities for people to create interesting businesses. I think the day of, you know, more generalist type developer skills and dev shops, etc., etc. will ultimately get crushed because that skill ultimately gets commoditized away by AI. Right? So I don't know, that's just my $0.02.
 
Jon: So I code base, we're just embracing it is what we're doing.
 
Peter: I love it.
 
Jon: What questions do you guys have? What did we cover? Not cover.
 
Peter: But do you think Chat GPT is going to going to put like, you know, is that the beginning of the end?
 
Jon: I think Chat GPT ultimately is a much bigger threat to being in Google. Perhaps, you know.
 
Peter: People prefer Egypt's not connected to the Internet. Yeah, so but it needs to be connected.
 
Jon: Could be, Yeah. So my guess is a startup that could come through and, and start plugging that goes into data like the internet. You're going to have much it's going to be tricky because your searches are going to be much more expensive to deliver because I think, Chat, Djibouti is for every search. It's either for every query or a set of queries.
 
Jon: It's like a penny or two. Yeah. And so for something like Google, they're caching a lot of data. They're they're looking at the data and then they're archiving in the right change and say, when someone answers this question, here's the top hundred responses were chat. GPT has an infinite number of possibilities. It could go down right right.
 
Peter: Yeah I think I know it's definitely a it's definitely a threat to like the beings and Googles of the world for sure.
 
Jon: But it wouldn't surprise me if they have something like this already up their sleeves or they've tested it.
 
Peter: I'm sure they're working on all kinds of stuff with it, right? I mean, they are. We know they are.
 
Jon: Though. There we go. So how did we do, Peter? Did we answer all of the startup related stuff? Because I focus more on the tech side.
 
Peter: Yeah. I mean, so Paul Graham talked about how, like, the corner case of chat CBT is that you just have like two founders and they just kind of design the product and I built up for them, right? And then like what does that mean in terms of like total number of startups? And that really leads into this discussion around venture capital and the impact on VC, Right?
 
Jon: Okay. Which is going to be part two of our series on chat CBT.
 
Peter: So tune back in for part two.
 
Jon: All right. And right now we're publishing twice a week. So this next episode is just a couple a couple of days out. All right. Go to venture capitalist firm. Let us know if you would, a venture capital law firm. We just launched a Slack channel. So you can ask us any of your questions and I'll make sure Peter responds to 100% of them, will respond to the best.
 
Jon: What we'll do.
 
Peter: Is.
 
Jon: We'll find like the top ones and kind of try to like.
 
Peter: Do our best. All right. We'll see you in Slack.
 
Jon: All right. See you guys.