00:00:04AI-Driven No-Code Website Builder
00:13:15Building Startups With AI and Ethics
00:21:14Ethical Considerations in AI Development
OVER DEZE AFLEVERING
Can you imagine creating a fully functional business website in less than 20 seconds? Listen to our enlightening conversation with Pritika Mehta, the mastermind behind Butternut AI, a revolutionary platform that's changing the game for SMBs, freelancers, and startups. Pritika shares the incredible journey from a single viral tweet to a rapid growth trajectory, all achieved without spending a dime on marketing. Discover how large language models and Pritika's vision to automate software building have come together to create this groundbreaking tool. You'll be fascinated by the "aha" moments users experience when their websites are generated almost instantaneously.
We also tackle the intricate world of AI development and ethics. Pritika opens up about the challenges of handling hallucinations, fine-tuning models with large datasets, and ensuring originality in website creation. The conversation takes a thoughtful turn as we discuss the broader ethical implications of AI technology, including the risks of deepfakes and the importance of regulatory measures. With a forward-looking perspective, we explore the transformative potential of AI in areas like medicine and no-code development, leaving you with a comprehensive view of the joys, challenges, and future possibilities of AI-driven innovation.
Accelerometer Podcast
Accelerometer Youtube
Anarchy
Anarchy Discord
Anarchy LLM-VM
Anarchy Twitter
Anarchy LinkedIn
Matthew Mirman LinkedIn
Engels
Verenigde Staten
LAAT NOTITIES ZIEN 🔗
TRANSCRIPTIE 🔗
Speaker 1:
0:04
Hello and welcome to another episode of the Accelerometer. I'm Dr Matthew Mermin, ceo and founder of Anarchy. Today, we're going to be discussing AI-driven no-code with Pritika Mehta, founder of Butternut AI. Pritika, can you tell us a little bit more about Butternut?
Speaker 2:
0:22
Sure Matthew, and thank you for having me here. So I am Pritika. I'm the co-founder of Butternut. Sure Matthew, and thank you for having me here. So I am Prithika. I'm the co-founder of Butternut AI, and Butternut AI is a generative app platform where people can build their business websites in under 20 seconds with just prompts. No coding, no drag and drop, just prompts.
Speaker 1:
0:40
That's awesome. How exactly does that work?
Speaker 2:
0:42
So you go to butanaai. So you just need to put in the name of your business and a few relevant keywords to describe your business and just go for it and the website will be ready in under 10 to 15 seconds.
Speaker 1:
0:58
How did you get started doing this?
Speaker 2:
1:01
Well. So when these LLMs got matured, you know I could see the world has drastically changed. People will see the impact in five years, but you know how people in SF they move so fast and things originate from here, so you could see how the world has changed. So I was looking at different problems I could solve and the bigger vision was how can we automate the software building industry? How can we, you know, do it without having so many coders there? But the lowest hanging fruit was whenever any business is built, they'll first go and build their website or their app. So can we automate that? Can we get that built in just a few minutes? And that's how we? That was the hypothesis. And then we launched it and I just put up a tweet, you know, and I said, hey, I built an MVP and I just put up a tweet because I wanted like initial 30, 40 people to try it and tell me if I am in the right direction, you know, and give me some feedback. But I put up that tweet at 8.30 pm at night and I woke up in the morning and it blew up. It went viral. People were trying out Patanat. They were building like crazy different websites on it and that's how we were launched. That's really exciting.
Speaker 2:
2:16
When was that? That was in April this year. That's recent. That's very recent. Yes.
Speaker 1:
2:21
Yeah, were you prepared for that much activity?
Speaker 2:
2:24
Not really, actually. You know, we were on AWS free tier and our servers broke and we were like okay and yeah, then we had to upgrade. So we were not prepared for that much traffic.
Speaker 1:
2:37
That's kind of like the dream of the champagne problems, isn't it? I know your servers break because it's too much traffic.
Speaker 2:
2:43
I know. And then, yeah, we did Twitter launch and then within 48 hours, we also launched it on Product Hunt and since then we haven't spent a penny on marketing and right now we have 160,000 users, and all through word of mouth and product-led growth.
Speaker 1:
2:59
Yeah, that's really cool. How did you find this team to work with?
Speaker 2:
3:05
So my team I've known them for the last 10 years and, yeah, so we have known each other since college. So, yeah, they should go back.
Speaker 1:
3:17
Yeah, is this the first project?
Speaker 2:
3:18
that you guys have done together. Not really. We have done a couple of projects and hackathons together built different projects together?
Speaker 1:
3:28
What sort of projects were you working on before?
Speaker 2:
3:29
this, so the other project that I was working on was so, if you know, this app like Chef.
Speaker 1:
3:40
Yeah.
Speaker 2:
3:40
Yeah, so this was another project that I built in college with my co-founders. It was like where people can send their home-cooked food and sell it in the market for people who want to eat their home-cooked food, and that was like we called it Wanmo. So that was one project we did together.
Speaker 1:
4:00
Why did you decide to switch to this?
Speaker 2:
4:03
That was just a project. So I would say this is a company, yeah, and this is like my third company. So before this, I was doing Sock Soho, which was a D2C brand, a premium designer sock brand for men, and we built chatbots where we drew e-commerce over WhatsApp and the market was India. So that went well. Some of the biggest Bollywood celebrities and top CEOs in India they wear our socks. In fact, bono also was spotted wearing socks at home. So, yeah, I think I have always been very passionate about building something that could go into millions of homes and, being an engineer, I have always enjoyed building things from scratch, so that has always been there. So now, after these, as I mentioned, llms got matured, it was like, okay, what are the different problems we can solve now? So that's what led to Butternut.
Speaker 1:
5:01
Yeah, they're just hacking along with it, just and just like I gotta build a website builder yes, it was not like a website.
Speaker 2:
5:08
I I mean the idea was, yeah, how can we help people, uh, get better business, and website was one of the ways to get that yeah, that's really cool.
Speaker 1:
5:20
Uh, is it mostly geared towards people early in their business careers or, like bigger businesses? Who would you say your target market is?
Speaker 2:
5:30
Currently we are targeting people who so there are a lot of YC companies who build on ButtonArt AI. Then I would say SMBs and freelancers, who already have a business but they need more control over their online presence, so it's like people who already have a business. 60% of the people are people who have a business already they could be a lawyer, real estate agents or startups and 40% are who are just doing it as a side project or just started out on their startup journey.
Speaker 1:
6:07
So when somebody comes and uses Butternut and they already have a website, what's the aha moment for them?
Speaker 2:
6:14
For them, the aha moment is, first thing they got a website in 20 seconds, whereas the other website they made it could have taken you know, it maybe took them like three months from a dev shop and paid like $5,000 to get that. So the aha moment is okay in 10 seconds you could get a website done that is customized for your business and for your demography. And then the cost, Would you say that's a lot about the control. Yes, you have the entire control. You don't need to know coding, you don't need to drag and drop no-transcript and I feel like that enables that for a lot of people.
Speaker 1:
7:14
It's creating an entirely new group of people in code Right.
Speaker 2:
7:20
True, that's very true. In fact, I talk to a lot of our customers, and a lot of our customers are also people who bought a domain name, who always wanted to start up something, but they didn't know where to start, and when they found out about ButtonArt, it was so easy for them. So it's like, yeah, enabling people to realize their dream and get them started.
Speaker 1:
7:46
Do you have any of these stories? Like companies that started, or now, like hundreds of million dollar companies that started with Butternut, yet in the last, like four months, it's been a crazy four months.
Speaker 2:
7:57
We will have those stories in next one year for sure.
Speaker 1:
8:01
That's really exciting. What's the craziest website somebody's built with butternut?
Speaker 2:
8:06
The craziest website. Okay, I would say chocolatefactorystore.
Speaker 1:
8:15
I really love that I like that name.
Speaker 2:
8:18
Yes, so it's like a chocolate. There's one owner, he runs his chocolate shop and he does it offline and that's very successful. He wanted to get online. He didn't know how to, and then he found out Butternut and he built this amazing, beautiful website on Butternut and I think that is something quite interesting, that's cool.
Speaker 1:
8:40
Have you tried his chocolate? Yes, I get it for free. Oh, my God, that's cool. Have you tried his chocolate? Yes, I get it for free, oh my God, that's wonderful.
Speaker 2:
8:49
Where's the chocolate from the chocolate, the beans? They get it sourced from across the country, the world, yeah, but they're based in.
Speaker 1:
8:55
California. Okay, so they're local. Yeah, they're local. Where did the name?
Speaker 2:
9:00
Butternut come from. Okay, butternut. So there were, you know, different things I was looking for. So, yeah, I picked the name. And so the thing was it should be easy for people to recall, you know, and when somebody sees it on the billboard on 101 Highway or 280, it should be like oh you know, I can build my website in 20 seconds and the name is Butternut, so recall value should be high. So that was my aim. So, and how would it look once it goes on billboards? That was my second consideration.
Speaker 1:
9:33
And third, domain name was available, so so you were looking at a list of domain names and you're like that one it's available.
Speaker 2:
9:44
Yeah, I was picking like, I was searching for different. I was thinking of different ways I could name the company and what it could be, what it could have like, what could be the recall value, what could be something that people can enjoy talking about, and I didn't want it to sound too techie. I want something, you know, that people use it in their everyday life.
Speaker 1:
10:00
Yeah, oh, that's cool For a brief, while my company was called Custard.
Speaker 2:
10:06
Custard. Wow, that's an easy name.
Speaker 1:
10:09
Yeah, because I was looking through the list of names and I was like, well, that's pronounceable, it's spellable, it's kind of easy to remember, it's tasty. Is the domain name a Wendepin? It might be now. I don't think I renewed it last year. Oh, you didn't. Okay, I'm going to get that. Rock the custard. We can do a deal, we can. That's a sweet name. There was a point in time where literally every time I had any idea that I had, I would go and find the domain name. Oh okay, nobody's ever bought one of the domain names for me. It's not a good business trading domain names.
Speaker 2:
10:43
Ah, yeah, yeah, oh, really yeah. So you have never sold any, so maybe you never bought the right domain name.
Speaker 1:
10:49
No, well, I bought the domain names for me, for yourself. Yeah, I'm happy with my domain name. That's amazing. How many domain names do you have right now? Oh, like 40, 50 domain names. Okay, yeah, it's an absurd number of domains.
Speaker 2:
11:02
So you can start your own GoDaddy, maybe sometime soon.
Speaker 1:
11:11
Yeah, a very tiny GoDaddy. There's a list actually of available ai domain names and there was some guy out there that like some of these domain names, they were listed for like $500,000. I emailed the guy. I was like I'm not paying $500,000 for the domain name, but I'll pay a hundred bucks. They countered me with a thousand bucks and I was like this is a good deal, wow, okay.
Speaker 2:
11:26
That's a good negotiation tactic yeah.
Speaker 1:
11:29
Well, I don't know. Yeah, I guess, starting with like 500,000 anchored, you know, maybe a thousand dollars for the domain name, you never know. That's fine. What was the domain name? Uh, that might have been anarchy. Anarchy might have been the one that I got, like at that huge discount at thousand dollars. That, okay, cool, that's amazing. Yeah, I got some really cool ones. I also got a leet speak. Leet speak yeah, you remember leet speak from like the 90s, where you used to like type with numbers to make yourself sound like a hacker. Oh, yes, yes, yeah, so I have Leetspeakcom.
Speaker 2:
12:04
Dot com oh, bring that train back.
Speaker 1:
12:07
Yeah, I think I need to. I just need to decide what to put there. So, yeah, anyway, what's been the most difficult part of your startup journey so far? Of any of your three startup journeys?
Speaker 2:
12:19
For me, okay, I think when you are building, you know any company, any business or a startup for the very first time, you don't, you're not aware that, even though you hear it all the time, things are going to break. You know there are going to be hard times, but you don't realize it until you have them. So it's like okay, when you're doing it for the first time and things break, you're like this is the end of the world. Everything is, this is done. You know we're not going to survive and that's it. But I think as you grow, as you keep on doing things and as you explore more, you understand these are just part and parcels of the startup journey and you just need to, you know, solve that part and parcel of the startup journey and you just need to, you know, solve that. That's just a challenge that you need to solve. That's not the end of the world. So I think initially, yeah, some low moments were like okay, this is the end of it.
Speaker 1:
13:07
But you just kept on going yeah, and you learn. It's just a part of it.
Speaker 2:
13:15
Have the startups been getting easier to build. I would say each startup is very different because the market is very different, the customers are very different and how you build that startup is very different. So I won't say it's easier, but I would say yes, I am more patient now.
Speaker 1:
13:32
What have you needed to be patient for?
Speaker 2:
13:34
Patient for building something. I would say that, as I mentioned, you know that millions of people could use that, could actually make an impact in someone's life, that could help people in some way or the other. So I think that's my motive of building a startup, like, first of all, do I enjoy building it? That's the first part, and then, secondly, is it solving someone's problem? That's the motto of the startup.
Speaker 1:
14:02
Has this been the startup that you've enjoyed building the most?
Speaker 2:
14:06
I think I've enjoyed all of them. I would say I would not All my babies equally.
Speaker 2:
14:11
Yes, I love all my babies equally. I would agree with that. But it's like I won't say I would go and build a startup that I don't enjoy, because you can do a project that you don't like because that might be done in two months or three months, but when you're doing a startup, that's a long time commitment and you cannot go ahead doing something for years and years if you don't enjoy. So I have enjoyed everything, all my startups, all my projects equally.
Speaker 1:
14:37
So where do you think that Butternut is going? So, after you get full penetration with the current version, the current product, what would your next products be?
Speaker 2:
14:49
Okay, so currently the product is. I would say we launched an MVP. Now we are building the features. We are building the next version where people can build complex websites on Button Nut AI. Once we take over the website building space, we are going to get into the e-commerce space that is exciting where a lot of transactions happen and yeah. So websites and apps these are the two spaces we'll be taking over in a few years.
Speaker 1:
15:19
What is something challenging about using AI at the moment?
Speaker 2:
15:23
About using AI, I would say yes, hallucinations is one part, and so there are so many new things happening all the time. So trying out new things and deciding, okay, what to build on, what is the right thing to build your startup on, is not challenging. It's exciting as well, but a little confusing as well. How do you keep up? I just live in SF.
Speaker 1:
15:51
It keeps up with you.
Speaker 2:
15:52
It keeps up with me. Yes.
Speaker 1:
15:54
You have inbound information in SF, exactly, yeah, I found that even living in SF, it was hard to parse all of the things going on. Like every hackathon I went to, every event I went to, I just learned about like 10 new tools. And then you'd be like when do you have time to look all these tools up?
Speaker 2:
16:11
It is true Like there is so much information, but then you know, you sometimes find gold as well.
Speaker 1:
16:19
Yeah, what sort of gold have you found recently?
Speaker 2:
16:23
Recently. Okay, so I would say the biggest gold everybody has found is OpenAI. So yeah, there are so many other tools here and there, but I would say OpenAI is the biggest gold we all have found.
Speaker 1:
16:36
Do you use OpenAI? Yes, do you like using OpenAI?
Speaker 2:
16:41
Yes, I've tried different models and I would say OpenAI is quite superior.
Speaker 1:
16:45
The quality of the model is what matters for you guys more than the latency or the controllability of the model.
Speaker 2:
16:53
Yes, yes, for us the quality matters a lot.
Speaker 1:
16:57
Do you guys ever run into issues with uptime?
Speaker 2:
17:02
Not a lot. So what we have done is we have built our own model as well by scraping 500,000 websites Wow yeah, so that we can build websites from scratch. You know, generate all the HTML CSS from scratch. So yes, that you know, with this uptime and all there are always issues, but then that's part of it.
Speaker 1:
17:22
Yeah, so your new model? Is it a fine-tuned or an existing model?
Speaker 2:
17:27
It is like yeah.
Speaker 1:
17:30
That's really cool. How was the process of fine-tuning?
Speaker 2:
17:35
So the process, it was Interesting. Again, I think data cleaning was a big challenge for us. Picking the right data and, you know, building using the right data was a good challenge. Other than that, I would say it wasn't. We are two AI engineers on the board, so it wasn't that difficult and we understood, you know, different models, we understood different parameters we could work on. So it wasn't that hard.
Speaker 1:
18:06
So scrape 500,000 websites and clean that. I guess you didn't have like a person sitting there checking every website, did you? No, no, so we built our own algorithms for it. Oh, that's cool. What would you suggest for people who have to clean data like that?
Speaker 2:
18:23
No, definitely they should be using some libraries to do that. And when you are doing data, you are definitely not picking random data. You have certain features or you have certain. You know you are categorizing data on certain parameters. So first of all, you need to be clear in your head what are you looking at and what do you exactly want to see in the data? So once I think you know what you want, it's much easier to build an algorithm or to use any package to scrape anything on the web.
Speaker 1:
18:53
You got to have a picture of the data in your mind.
Speaker 2:
18:56
Yes, what do you want out of it? Like, data will tell you things, but exactly you know what are different things. You want to look at it. When you look at that data.
Speaker 1:
19:05
So what's something that you're excited for in the world of AI? Can I say AGI?
Speaker 2:
19:12
You can say AGI, agi, okay, no, I think I am excited for each and every industry, especially medicine. Like I'm building in the no-cold world, so that is super exciting. Like how, in future, we won't need a lot of coders. We would need people who could lay down, you know the exact requirements, the exact flow, and then we have agents that can build that stuff. So very excited for that. But other than that, I think there are so many problems to be solved when it comes to medicine, biotech, so I'm really excited to see you know how AI can help fasten that up.
Speaker 1:
19:50
So would you classify yourself as EAC, eac, effective Accelerationist? It's the reactionary movement to people saying that we should stop all AI development and it's people saying, instead should stop all AI development. And it's people saying, instead of stopping AI development, let's go faster. Let's make good AI so fast that it'll get rid of all the bad AI.
Speaker 2:
20:12
Yeah, so I think the same thing happened. You know, when internet came, people were really worried oh, what's going to happen to our data? Oh, privacy, this and that. But I think when anything new comes, it comes with a lot of good things and quite a few bad things as well. So I think it's our job as well to look at the bad things and how can we control those things. We really don't want things to go out of our control. So I would really like to accelerate the speed at which AI grows, but at the same time, I would also like to do it ethically, because, being an AI engineer, I could see what problems it can create in the world. So, yes, I would say we need to do, you know, balance it. We really need to move fast with a good AI staff, but we need to control the bad thing at the same time. It's our responsibility, like people who are building an AI space, it's our responsibility to do that.
Speaker 1:
21:06
What's something that you guys do at Butternut to do AI more safely?
Speaker 2:
21:14
So a lot of people ask us if you know they can copy someone's website. We don't want to do that and we don't want to allow that in future as well. Like copying someone's app, you can take maybe some inspiration or something, but copying someone's work is something we would not like to do in future. So, yeah, so I think we don't have a lot of ethical risk when it comes to building websites or taking businesses online, but this is one area where we would be quite aware of, like we'll be cautious, you know, in this area how do you, how do you prevent people from creating copies of other websites?
Speaker 1:
21:57
is there a lot of websites out there?
Speaker 2:
21:59
yeah, so currently, since our websites are built from scratch, you know there are different elements to each website, so it's it's I would say on button that you can. You know, right now everything is like different, but in future, when we are going to introduce this feature where you can say here's my existing website and make a new website out of it, so there could be a way where people put their competitors' websites. So those are some issues, areas of issues where we'll be controlling different things.
Speaker 1:
22:29
What do you think that more people should be doing with respect to AI safety?
Speaker 2:
22:34
When it comes to AI safety, I think it depends on the industry you're working in. Especially if you're working in like video industry, using AI or audio industry, you know deepfakes, like how views is generated or content. I think there's a lot to be done Text generation, image generation, video generation when any layman could have control over generating any video of any celebrity or any politician and use or use any of their wise. So it's like how do you prevent that? How do you know this? Is, you know, maybe a real use or a fake use? So I think those industries really need to also figure out, like if they are giving a tool in everyone's hand to build these things, they also also need to figure out okay, I hope that people are not using it for the wrong purpose. So those checks are really important.
Speaker 1:
23:29
Is there anything that worries you about the future of AI?
Speaker 2:
23:32
I think what only worries me is that I hope the government doesn't stop us from working in AI. I see that some people are scared with AI advancing at this pace, but this is what the future is, and so I wouldn't say anything really worries me. I'm excited about the future.
Speaker 1:
23:57
Well, that's awesome. I would say that's acceleration. That is Well. Thank you very much.
0:04
Hello and welcome to another episode of the Accelerometer. I'm Dr Matthew Mermin, ceo and founder of Anarchy. Today, we're going to be discussing AI-driven no-code with Pritika Mehta, founder of Butternut AI. Pritika, can you tell us a little bit more about Butternut?
Speaker 2:
0:22
Sure Matthew, and thank you for having me here. So I am Pritika. I'm the co-founder of Butternut. Sure Matthew, and thank you for having me here. So I am Prithika. I'm the co-founder of Butternut AI, and Butternut AI is a generative app platform where people can build their business websites in under 20 seconds with just prompts. No coding, no drag and drop, just prompts.
Speaker 1:
0:40
That's awesome. How exactly does that work?
Speaker 2:
0:42
So you go to butanaai. So you just need to put in the name of your business and a few relevant keywords to describe your business and just go for it and the website will be ready in under 10 to 15 seconds.
Speaker 1:
0:58
How did you get started doing this?
Speaker 2:
1:01
Well. So when these LLMs got matured, you know I could see the world has drastically changed. People will see the impact in five years, but you know how people in SF they move so fast and things originate from here, so you could see how the world has changed. So I was looking at different problems I could solve and the bigger vision was how can we automate the software building industry? How can we, you know, do it without having so many coders there? But the lowest hanging fruit was whenever any business is built, they'll first go and build their website or their app. So can we automate that? Can we get that built in just a few minutes? And that's how we? That was the hypothesis. And then we launched it and I just put up a tweet, you know, and I said, hey, I built an MVP and I just put up a tweet because I wanted like initial 30, 40 people to try it and tell me if I am in the right direction, you know, and give me some feedback. But I put up that tweet at 8.30 pm at night and I woke up in the morning and it blew up. It went viral. People were trying out Patanat. They were building like crazy different websites on it and that's how we were launched. That's really exciting.
Speaker 2:
2:16
When was that? That was in April this year. That's recent. That's very recent. Yes.
Speaker 1:
2:21
Yeah, were you prepared for that much activity?
Speaker 2:
2:24
Not really, actually. You know, we were on AWS free tier and our servers broke and we were like okay and yeah, then we had to upgrade. So we were not prepared for that much traffic.
Speaker 1:
2:37
That's kind of like the dream of the champagne problems, isn't it? I know your servers break because it's too much traffic.
Speaker 2:
2:43
I know. And then, yeah, we did Twitter launch and then within 48 hours, we also launched it on Product Hunt and since then we haven't spent a penny on marketing and right now we have 160,000 users, and all through word of mouth and product-led growth.
Speaker 1:
2:59
Yeah, that's really cool. How did you find this team to work with?
Speaker 2:
3:05
So my team I've known them for the last 10 years and, yeah, so we have known each other since college. So, yeah, they should go back.
Speaker 1:
3:17
Yeah, is this the first project?
Speaker 2:
3:18
that you guys have done together. Not really. We have done a couple of projects and hackathons together built different projects together?
Speaker 1:
3:28
What sort of projects were you working on before?
Speaker 2:
3:29
this, so the other project that I was working on was so, if you know, this app like Chef.
Speaker 1:
3:40
Yeah.
Speaker 2:
3:40
Yeah, so this was another project that I built in college with my co-founders. It was like where people can send their home-cooked food and sell it in the market for people who want to eat their home-cooked food, and that was like we called it Wanmo. So that was one project we did together.
Speaker 1:
4:00
Why did you decide to switch to this?
Speaker 2:
4:03
That was just a project. So I would say this is a company, yeah, and this is like my third company. So before this, I was doing Sock Soho, which was a D2C brand, a premium designer sock brand for men, and we built chatbots where we drew e-commerce over WhatsApp and the market was India. So that went well. Some of the biggest Bollywood celebrities and top CEOs in India they wear our socks. In fact, bono also was spotted wearing socks at home. So, yeah, I think I have always been very passionate about building something that could go into millions of homes and, being an engineer, I have always enjoyed building things from scratch, so that has always been there. So now, after these, as I mentioned, llms got matured, it was like, okay, what are the different problems we can solve now? So that's what led to Butternut.
Speaker 1:
5:01
Yeah, they're just hacking along with it, just and just like I gotta build a website builder yes, it was not like a website.
Speaker 2:
5:08
I I mean the idea was, yeah, how can we help people, uh, get better business, and website was one of the ways to get that yeah, that's really cool.
Speaker 1:
5:20
Uh, is it mostly geared towards people early in their business careers or, like bigger businesses? Who would you say your target market is?
Speaker 2:
5:30
Currently we are targeting people who so there are a lot of YC companies who build on ButtonArt AI. Then I would say SMBs and freelancers, who already have a business but they need more control over their online presence, so it's like people who already have a business. 60% of the people are people who have a business already they could be a lawyer, real estate agents or startups and 40% are who are just doing it as a side project or just started out on their startup journey.
Speaker 1:
6:07
So when somebody comes and uses Butternut and they already have a website, what's the aha moment for them?
Speaker 2:
6:14
For them, the aha moment is, first thing they got a website in 20 seconds, whereas the other website they made it could have taken you know, it maybe took them like three months from a dev shop and paid like $5,000 to get that. So the aha moment is okay in 10 seconds you could get a website done that is customized for your business and for your demography. And then the cost, Would you say that's a lot about the control. Yes, you have the entire control. You don't need to know coding, you don't need to drag and drop no-transcript and I feel like that enables that for a lot of people.
Speaker 1:
7:14
It's creating an entirely new group of people in code Right.
Speaker 2:
7:20
True, that's very true. In fact, I talk to a lot of our customers, and a lot of our customers are also people who bought a domain name, who always wanted to start up something, but they didn't know where to start, and when they found out about ButtonArt, it was so easy for them. So it's like, yeah, enabling people to realize their dream and get them started.
Speaker 1:
7:46
Do you have any of these stories? Like companies that started, or now, like hundreds of million dollar companies that started with Butternut, yet in the last, like four months, it's been a crazy four months.
Speaker 2:
7:57
We will have those stories in next one year for sure.
Speaker 1:
8:01
That's really exciting. What's the craziest website somebody's built with butternut?
Speaker 2:
8:06
The craziest website. Okay, I would say chocolatefactorystore.
Speaker 1:
8:15
I really love that I like that name.
Speaker 2:
8:18
Yes, so it's like a chocolate. There's one owner, he runs his chocolate shop and he does it offline and that's very successful. He wanted to get online. He didn't know how to, and then he found out Butternut and he built this amazing, beautiful website on Butternut and I think that is something quite interesting, that's cool.
Speaker 1:
8:40
Have you tried his chocolate? Yes, I get it for free. Oh, my God, that's cool. Have you tried his chocolate? Yes, I get it for free, oh my God, that's wonderful.
Speaker 2:
8:49
Where's the chocolate from the chocolate, the beans? They get it sourced from across the country, the world, yeah, but they're based in.
Speaker 1:
8:55
California. Okay, so they're local. Yeah, they're local. Where did the name?
Speaker 2:
9:00
Butternut come from. Okay, butternut. So there were, you know, different things I was looking for. So, yeah, I picked the name. And so the thing was it should be easy for people to recall, you know, and when somebody sees it on the billboard on 101 Highway or 280, it should be like oh you know, I can build my website in 20 seconds and the name is Butternut, so recall value should be high. So that was my aim. So, and how would it look once it goes on billboards? That was my second consideration.
Speaker 1:
9:33
And third, domain name was available, so so you were looking at a list of domain names and you're like that one it's available.
Speaker 2:
9:44
Yeah, I was picking like, I was searching for different. I was thinking of different ways I could name the company and what it could be, what it could have like, what could be the recall value, what could be something that people can enjoy talking about, and I didn't want it to sound too techie. I want something, you know, that people use it in their everyday life.
Speaker 1:
10:00
Yeah, oh, that's cool For a brief, while my company was called Custard.
Speaker 2:
10:06
Custard. Wow, that's an easy name.
Speaker 1:
10:09
Yeah, because I was looking through the list of names and I was like, well, that's pronounceable, it's spellable, it's kind of easy to remember, it's tasty. Is the domain name a Wendepin? It might be now. I don't think I renewed it last year. Oh, you didn't. Okay, I'm going to get that. Rock the custard. We can do a deal, we can. That's a sweet name. There was a point in time where literally every time I had any idea that I had, I would go and find the domain name. Oh okay, nobody's ever bought one of the domain names for me. It's not a good business trading domain names.
Speaker 2:
10:43
Ah, yeah, yeah, oh, really yeah. So you have never sold any, so maybe you never bought the right domain name.
Speaker 1:
10:49
No, well, I bought the domain names for me, for yourself. Yeah, I'm happy with my domain name. That's amazing. How many domain names do you have right now? Oh, like 40, 50 domain names. Okay, yeah, it's an absurd number of domains.
Speaker 2:
11:02
So you can start your own GoDaddy, maybe sometime soon.
Speaker 1:
11:11
Yeah, a very tiny GoDaddy. There's a list actually of available ai domain names and there was some guy out there that like some of these domain names, they were listed for like $500,000. I emailed the guy. I was like I'm not paying $500,000 for the domain name, but I'll pay a hundred bucks. They countered me with a thousand bucks and I was like this is a good deal, wow, okay.
Speaker 2:
11:26
That's a good negotiation tactic yeah.
Speaker 1:
11:29
Well, I don't know. Yeah, I guess, starting with like 500,000 anchored, you know, maybe a thousand dollars for the domain name, you never know. That's fine. What was the domain name? Uh, that might have been anarchy. Anarchy might have been the one that I got, like at that huge discount at thousand dollars. That, okay, cool, that's amazing. Yeah, I got some really cool ones. I also got a leet speak. Leet speak yeah, you remember leet speak from like the 90s, where you used to like type with numbers to make yourself sound like a hacker. Oh, yes, yes, yeah, so I have Leetspeakcom.
Speaker 2:
12:04
Dot com oh, bring that train back.
Speaker 1:
12:07
Yeah, I think I need to. I just need to decide what to put there. So, yeah, anyway, what's been the most difficult part of your startup journey so far? Of any of your three startup journeys?
Speaker 2:
12:19
For me, okay, I think when you are building, you know any company, any business or a startup for the very first time, you don't, you're not aware that, even though you hear it all the time, things are going to break. You know there are going to be hard times, but you don't realize it until you have them. So it's like okay, when you're doing it for the first time and things break, you're like this is the end of the world. Everything is, this is done. You know we're not going to survive and that's it. But I think as you grow, as you keep on doing things and as you explore more, you understand these are just part and parcels of the startup journey and you just need to, you know, solve that part and parcel of the startup journey and you just need to, you know, solve that. That's just a challenge that you need to solve. That's not the end of the world. So I think initially, yeah, some low moments were like okay, this is the end of it.
Speaker 1:
13:07
But you just kept on going yeah, and you learn. It's just a part of it.
Speaker 2:
13:15
Have the startups been getting easier to build. I would say each startup is very different because the market is very different, the customers are very different and how you build that startup is very different. So I won't say it's easier, but I would say yes, I am more patient now.
Speaker 1:
13:32
What have you needed to be patient for?
Speaker 2:
13:34
Patient for building something. I would say that, as I mentioned, you know that millions of people could use that, could actually make an impact in someone's life, that could help people in some way or the other. So I think that's my motive of building a startup, like, first of all, do I enjoy building it? That's the first part, and then, secondly, is it solving someone's problem? That's the motto of the startup.
Speaker 1:
14:02
Has this been the startup that you've enjoyed building the most?
Speaker 2:
14:06
I think I've enjoyed all of them. I would say I would not All my babies equally.
Speaker 2:
14:11
Yes, I love all my babies equally. I would agree with that. But it's like I won't say I would go and build a startup that I don't enjoy, because you can do a project that you don't like because that might be done in two months or three months, but when you're doing a startup, that's a long time commitment and you cannot go ahead doing something for years and years if you don't enjoy. So I have enjoyed everything, all my startups, all my projects equally.
Speaker 1:
14:37
So where do you think that Butternut is going? So, after you get full penetration with the current version, the current product, what would your next products be?
Speaker 2:
14:49
Okay, so currently the product is. I would say we launched an MVP. Now we are building the features. We are building the next version where people can build complex websites on Button Nut AI. Once we take over the website building space, we are going to get into the e-commerce space that is exciting where a lot of transactions happen and yeah. So websites and apps these are the two spaces we'll be taking over in a few years.
Speaker 1:
15:19
What is something challenging about using AI at the moment?
Speaker 2:
15:23
About using AI, I would say yes, hallucinations is one part, and so there are so many new things happening all the time. So trying out new things and deciding, okay, what to build on, what is the right thing to build your startup on, is not challenging. It's exciting as well, but a little confusing as well. How do you keep up? I just live in SF.
Speaker 1:
15:51
It keeps up with you.
Speaker 2:
15:52
It keeps up with me. Yes.
Speaker 1:
15:54
You have inbound information in SF, exactly, yeah, I found that even living in SF, it was hard to parse all of the things going on. Like every hackathon I went to, every event I went to, I just learned about like 10 new tools. And then you'd be like when do you have time to look all these tools up?
Speaker 2:
16:11
It is true Like there is so much information, but then you know, you sometimes find gold as well.
Speaker 1:
16:19
Yeah, what sort of gold have you found recently?
Speaker 2:
16:23
Recently. Okay, so I would say the biggest gold everybody has found is OpenAI. So yeah, there are so many other tools here and there, but I would say OpenAI is the biggest gold we all have found.
Speaker 1:
16:36
Do you use OpenAI? Yes, do you like using OpenAI?
Speaker 2:
16:41
Yes, I've tried different models and I would say OpenAI is quite superior.
Speaker 1:
16:45
The quality of the model is what matters for you guys more than the latency or the controllability of the model.
Speaker 2:
16:53
Yes, yes, for us the quality matters a lot.
Speaker 1:
16:57
Do you guys ever run into issues with uptime?
Speaker 2:
17:02
Not a lot. So what we have done is we have built our own model as well by scraping 500,000 websites Wow yeah, so that we can build websites from scratch. You know, generate all the HTML CSS from scratch. So yes, that you know, with this uptime and all there are always issues, but then that's part of it.
Speaker 1:
17:22
Yeah, so your new model? Is it a fine-tuned or an existing model?
Speaker 2:
17:27
It is like yeah.
Speaker 1:
17:30
That's really cool. How was the process of fine-tuning?
Speaker 2:
17:35
So the process, it was Interesting. Again, I think data cleaning was a big challenge for us. Picking the right data and, you know, building using the right data was a good challenge. Other than that, I would say it wasn't. We are two AI engineers on the board, so it wasn't that difficult and we understood, you know, different models, we understood different parameters we could work on. So it wasn't that hard.
Speaker 1:
18:06
So scrape 500,000 websites and clean that. I guess you didn't have like a person sitting there checking every website, did you? No, no, so we built our own algorithms for it. Oh, that's cool. What would you suggest for people who have to clean data like that?
Speaker 2:
18:23
No, definitely they should be using some libraries to do that. And when you are doing data, you are definitely not picking random data. You have certain features or you have certain. You know you are categorizing data on certain parameters. So first of all, you need to be clear in your head what are you looking at and what do you exactly want to see in the data? So once I think you know what you want, it's much easier to build an algorithm or to use any package to scrape anything on the web.
Speaker 1:
18:53
You got to have a picture of the data in your mind.
Speaker 2:
18:56
Yes, what do you want out of it? Like, data will tell you things, but exactly you know what are different things. You want to look at it. When you look at that data.
Speaker 1:
19:05
So what's something that you're excited for in the world of AI? Can I say AGI?
Speaker 2:
19:12
You can say AGI, agi, okay, no, I think I am excited for each and every industry, especially medicine. Like I'm building in the no-cold world, so that is super exciting. Like how, in future, we won't need a lot of coders. We would need people who could lay down, you know the exact requirements, the exact flow, and then we have agents that can build that stuff. So very excited for that. But other than that, I think there are so many problems to be solved when it comes to medicine, biotech, so I'm really excited to see you know how AI can help fasten that up.
Speaker 1:
19:50
So would you classify yourself as EAC, eac, effective Accelerationist? It's the reactionary movement to people saying that we should stop all AI development and it's people saying, instead should stop all AI development. And it's people saying, instead of stopping AI development, let's go faster. Let's make good AI so fast that it'll get rid of all the bad AI.
Speaker 2:
20:12
Yeah, so I think the same thing happened. You know, when internet came, people were really worried oh, what's going to happen to our data? Oh, privacy, this and that. But I think when anything new comes, it comes with a lot of good things and quite a few bad things as well. So I think it's our job as well to look at the bad things and how can we control those things. We really don't want things to go out of our control. So I would really like to accelerate the speed at which AI grows, but at the same time, I would also like to do it ethically, because, being an AI engineer, I could see what problems it can create in the world. So, yes, I would say we need to do, you know, balance it. We really need to move fast with a good AI staff, but we need to control the bad thing at the same time. It's our responsibility, like people who are building an AI space, it's our responsibility to do that.
Speaker 1:
21:06
What's something that you guys do at Butternut to do AI more safely?
Speaker 2:
21:14
So a lot of people ask us if you know they can copy someone's website. We don't want to do that and we don't want to allow that in future as well. Like copying someone's app, you can take maybe some inspiration or something, but copying someone's work is something we would not like to do in future. So, yeah, so I think we don't have a lot of ethical risk when it comes to building websites or taking businesses online, but this is one area where we would be quite aware of, like we'll be cautious, you know, in this area how do you, how do you prevent people from creating copies of other websites?
Speaker 1:
21:57
is there a lot of websites out there?
Speaker 2:
21:59
yeah, so currently, since our websites are built from scratch, you know there are different elements to each website, so it's it's I would say on button that you can. You know, right now everything is like different, but in future, when we are going to introduce this feature where you can say here's my existing website and make a new website out of it, so there could be a way where people put their competitors' websites. So those are some issues, areas of issues where we'll be controlling different things.
Speaker 1:
22:29
What do you think that more people should be doing with respect to AI safety?
Speaker 2:
22:34
When it comes to AI safety, I think it depends on the industry you're working in. Especially if you're working in like video industry, using AI or audio industry, you know deepfakes, like how views is generated or content. I think there's a lot to be done Text generation, image generation, video generation when any layman could have control over generating any video of any celebrity or any politician and use or use any of their wise. So it's like how do you prevent that? How do you know this? Is, you know, maybe a real use or a fake use? So I think those industries really need to also figure out, like if they are giving a tool in everyone's hand to build these things, they also also need to figure out okay, I hope that people are not using it for the wrong purpose. So those checks are really important.
Speaker 1:
23:29
Is there anything that worries you about the future of AI?
Speaker 2:
23:32
I think what only worries me is that I hope the government doesn't stop us from working in AI. I see that some people are scared with AI advancing at this pace, but this is what the future is, and so I wouldn't say anything really worries me. I'm excited about the future.
Speaker 1:
23:57
Well, that's awesome. I would say that's acceleration. That is Well. Thank you very much.
OUDE AFLEVERINGEN ZOEKEN
Zoek naar eerdere afleveringen van Dr. Mirman's Accelerometer.
Disclaimer: De podcast en de illustraties die op deze pagina zijn ingesloten, zijn afkomstig van Matthew Mirman, eigendom van de eigenaar en niet gelieerd aan of goedgekeurd door Listen Notes, Inc.
BEWERKEN
Bedankt voor uw hulp bij het up-to-date te houden van de podcast-database .
