Triangle Tweener Talks

00:00:04 - Tina Tang
A black and white image of Amelia Earhart. The model would add color to that in, you know, matter of seconds. But then there were all of these kind of human edits that I wanted that I wanted to do that wasn't just like add color to this, but oh, let's change the color of her accessories and things like that. So I did that in Photoshop and that's when I kind of experienced the combination of machine learning and human creativity combined.

00:00:33 - Scot Wingo
Interesting.

00:00:33 - Tina Tang
And I thought that was very inspiring. And it made me think about how powerful that technology was, but also how inaccessible it was because someone couldn't just go and do that at the time. Welcome to Triangle Tweener Talks, a weekly podcast by Builders for Builders where we explore the startup journey from the idea to the exit and all the lessons in between. With an exclusive of focus on founders from the Triangle region of North Carolina, Tweener Talks is produced by Earfluence. Now here is your host, serial founder and General Partner of the Triangle Tweener Fund, Scott Wingo.

00:01:15 - Scot Wingo
Welcome to this episode of the Triangle Tweener Talks podcast, a discussion with Bristles founder Tina Tang. First, special thanks to our friends at Earfluence who produced this podcast. If you're thinking about hosting a podcast or want to be a guest on a podcast, check out earfluence@earfluence.com this episode is sponsored by Robinson Bryce, a full service business law firm with a passion for supporting the Triangle entrepreneurial ecosystem. Learn more about Robinson Bradshaw startups and venture capital practice@robinsonbradshaw.com and smashing boxes, a Durham based lean design centric digital transformation company@smashingboxes.com in this episode we feature a first time founder who is working her way through product market fit. Tina is the daughter of two US immigrants who ended up in the Triangle. Thanks to her husband coming to attend Duke, Tina got her Master's from UVA in Systems Engineering where she focused on learning about the exploding field of AI and machine learning. At Bristol's, Tina is able to merge her love of technology and math with her creative side. She's also enjoying one of my favorite product market fit strategies. Going small to go big. Hey Tina, thanks for coming in for the podcast.

00:02:32 - Tina Tang
Hey Scott, good to see you.

00:02:33 - Scot Wingo
You. Yeah. I'm excited to hear your backstory and we were discussing before we went live that we've recorded a lot of serial entrepreneurs and you're a first time entrepreneur which is exciting and you're kind of in the thick of what we would call the product to market Fit Journey. So Excited to dig into how that's going and all the things you've learned there and all that good stuff.

00:02:57 - Tina Tang
Yeah.

00:02:58 - Scot Wingo
Before we jump into that, let's start at the beginning. Where are you from?

00:03:01 - Tina Tang
So I, I grew up in Maryland and Virginia, so Northern Virginia, Alexandria, and went to school there and didn't get to North Carolina till 2016. But yeah, that's. That's where I'm from.

00:03:19 - Scot Wingo
Yeah. And then were your parents part of the. The D.C. beltway Bandit kind of thing? A lot of people that live there are tied into the government in some way.

00:03:27 - Tina Tang
No. So I'll tell a little bit of backstory about where my parents are from. So my dad is a refugee from the Vietnam War. So he immigrated to the US in the late 70s. And my mom is from a farming village in rural China. So they grew up farming silkworms. So their story is they both are from kind of Canton speaking places. So even though they're from different countries, they spoke the same language, Cantonese. And when they got to the States, for my dad, it was a super harrowing journey from Vietnam.

00:04:14 - Scot Wingo
He had less helicopters out, kind of. Yeah, that was crazy seeing a lot of the footage from that era.

00:04:19 - Tina Tang
Yeah, yeah. He immigrated to the US So he escaped to Vietnam from. On a boat. So at the time, he was in his late teens working at a pen factory and saving up money. It was really expensive to get a ticket on one of these boats to escape Vietnam. So my dad saved up months, like close to a year of salary. And at the time there were a lot of people trying to get out of Vietnam, so there were a lot of scammers. So you could buy a ticket and there could not be a boat the next day. And that happened to him the first time.

00:04:56 - Scot Wingo
Holy cow.

00:04:57 - Tina Tang
So he saved up months of salary and his bought a ticket to the boat and there was no boat the next morning. So that was devastating. He had to kind of save up all over again, but really wanted to get to the US So he saved up again. And the second time there was a boat. So he made it to the US and got to eventually. So got on a boat, went to Malaysia, worked at a. They kind of have these camps there that refugees can work at, and it's a safe place for them to stay until they can find a way to the US and so eventually he got from the US to or from Malaysia to the US and got to New York and hitchhiked his way to Maryland to meet up with his brother.

00:05:46 - Scot Wingo
So his brother was there.

00:05:47 - Tina Tang
Yeah.

00:05:48 - Scot Wingo
Is that like the classic Coming through Staten island or Ellis island, kind of a. Yeah, yeah. That's interesting.

00:05:54 - Tina Tang
And so he, at that time, you know, didn't have any job prospects, didn't know the language. And so he ended up working at a Chinese restaurant where they spoke Cantonese. And that's where he met my mom, who was also working there.

00:06:10 - Scot Wingo
Oh, wow, cool. And then what, what brought your mom here also?

00:06:14 - Tina Tang
Just trying to find a way to live an easier life. So she's from a very poor village in China where they didn't even have running water.

00:06:23 - Scot Wingo
Is it in the north or south?

00:06:24 - Tina Tang
In the South. It's called Toisan. They both have similar stories, wanted to come to the US to find a way to live easier lives. And so when they got here, they were both kind of in survival mode, just trying to find a way to get a job. And they had dreams of, you know, the American dream of saving up, letting their kids have good educations. They both weren't, they didn't have access to educations where they were from. My dad, because of the Vietnam War, they shut down the schools when he was in fifth grade. And then my mom also didn't have access to a good education. And so probably when she was very.

00:07:07 - Scot Wingo
Young, got put onto the silkworm farm. Yeah, I imagine. Yeah.

00:07:11 - Tina Tang
So when they both got to the States, they both had dreams of starting a family here and giving me and my brother more opportunity. And they had a dream of, you know, buying a house. So it was very humble. Very humble dream.

00:07:28 - Scot Wingo
That's the American dream.

00:07:30 - Tina Tang
Yeah. And they, they were very good at making the most of what they had, so they didn't have a lot, but they were really good at saving. And so eventually did buy a house. And me and my brother were able to have great educations here and lived out what they wanted for us.

00:07:50 - Scot Wingo
Yeah. My founder and my first two businesses was a first generation immigrant born in the US Kind of thing, but his parents were from Lithuania and every one of his siblings, they're just like, the work ethic is just amazing. Like the, you know, it's just they, they, they totally got that from the parents and understand how the American dream works. And, you know, all the work hard, make a life for yourself. The. So you're in Maryland and Virginia. Where did you go to undergrad?

00:08:22 - Tina Tang
So I ended up going to UVA for undergrad. It's kind of like the UNC of Virginia.

00:08:28 - Scot Wingo
Yeah. You're a cavalier.

00:08:30 - Tina Tang
Yes.

00:08:30 - Scot Wingo
Or Oahu.

00:08:32 - Tina Tang
Both.

00:08:32 - Scot Wingo
Yeah. Yeah. And then what'd you study?

00:08:36 - Tina Tang
So at the time, I had no idea what I wanted to do. And I was kind of interested in everything. I liked art, I liked writing, I liked math, I liked science. And I was, I think, particularly good at math and science. So I thought I'd go into engineering. Ended up majoring in civil engineering. And out of college I worked for a civil engineering consulting firm for about a year. Kind of decided that wasn't for me and switched to IT consulting At Accenture?

00:09:08 - Scot Wingo
Yeah. And then in civil engineering, did you like the infrastructure part or more like the airport? So you can kind of do like, roads, bridges, airports. There's like a lot of different transportation, like rail. Was there a particular thing in there that interested you?

00:09:25 - Tina Tang
I like the data analytics side of the field. So I did research in stormwater management and transportation engineering. Kind of like capacity analysis and figuring out where you need animal crossings on roads. I did a bunch of kind of random things, but I really enjoyed the data analytics side of it. And when I was working in civil engineering consulting, the area I was in was called Intelligent Transportation Systems. So it's kind of like all the sensors along the highway and how they alert, like ramp meters and things like that, if you're familiar. So like if there's a automated ramp meter system, kind of when the bar goes up and lets cars onto the. Yeah.

00:10:17 - Scot Wingo
Okay, so you did that. Then you decided you had a passion for it. Did you learn how to program in your engineering program a little bit. Like, did they have it where you had to take a little bit of a basic programming type?

00:10:27 - Tina Tang
We did. We had to take basic programming courses in Java and C. So I had, you know, basic knowledge. I wouldn't. I didn't consider myself good at programming. And so you had the interest. Yeah, yeah. But good problem solving skills, I would say.

00:10:43 - Scot Wingo
Yeah. So the new Accenture and then what kind of program, what kind of projects did you work on there?

00:10:48 - Tina Tang
So I switched into IT consulting at Accenture and started working out of their DC office. And initially my roles were kind of on the functional side. Wrote a lot of SQL queries to analyze reports, lived in Excel and version one, which is like Jira, but a different version of jira.

00:11:08 - Scot Wingo
Like a ticket tracking system.

00:11:10 - Tina Tang
Yeah. So the system I was working in was a invoicing system for FedEx. So, yeah, it's pretty interesting.

00:11:19 - Scot Wingo
Yeah. Did you get to go to Memphis and meet FedEx a lot?

00:11:23 - Tina Tang
No. So we worked out of the USPS office in D.C. where I was on the client site.

00:11:30 - Scot Wingo
Okay, and then what else did you do? I know ultimately you're going to go back to UVA for a master's. What happened? Kind of.

00:11:36 - Tina Tang
What else happened inside of there? So I met my now husband at Accenture. We actually interviewed on the same day, which was funny. Looking back, I'll tell a little bit about Anthony's backstory, because it kind of like, yeah, it's going to come into play here. So Anthony did his undergrad in biomedical engineering at Johns Hopkins, and he was working with very early neural nets. After his undergrad, he was kind of set on going to start his master's at Columbia in New York, focused on theoretical neuroscience. But he had just had this private school education and undergrad at Johns Hopkins, and his parents were really concerned about finances. The master's at Columbia would have been expensive, coupled with living in New York City was very expensive. So they were kind of trying to encourage him to think about maybe see what working's like for a little bit. And so Anthony had a friend who was working at Accenture, and his friend ended up referring him. So that's how he ended up going to Accenture to try to try the working thing. So we both ended up in IT consulting. And after three years of Anthony working at Postal as well, so we were both on the same client site, he wanted to get back to grad school. He was kind of ready. A lot of interesting things were happening in deep learning at the time. Like image classification was getting better and better, and convolutional networks were getting better, and generative adversarial networks were also coming out. So a lot of exciting things were happening. And he had been kind of keeping up with all this and wanted to go back to grad school to pursue his PhD.

00:13:27 - Scot Wingo
What year is this?

00:13:28 - Tina Tang
So this is 2014, 2015, and then 2015, Anthony applies to grad schools and ends up getting into a biomedical engineering program at Duke to get his PhD. And it was a really good program focused on computational neuroscience. He was really excited about it at the time. We had been dating for a few years, and we decided we'd move to Durham together. So that's kind of how we made it to the Triangle.

00:14:03 - Scot Wingo
Yep.

00:14:04 - Tina Tang
And so I switched from Accenture DC to Accenture Raleigh, and that's when I started working in the Triangle. So Anthony's doing his PhD thing, and I'm starting, like, at Accenture. Every project that you're on kind of feels like a new job. Yeah. So now I'm working as Tier three lead of this project in rtp. Did that for six months. And then there was this new project starting in Raleigh, and They needed a product owner. So I became one of the product owners on that project. So there were two product owners. We had a project manager supervisor and an Agile coach. So it's four of us and they kind of dropped us into this building in Dorothea Dix campus, if you know it.

00:14:53 - Scot Wingo
I know it well.

00:14:55 - Tina Tang
Do you know the Broughton building? Oh, you do? So, okay, so they're not very nice.

00:15:00 - Scot Wingo
That used to be a mental hospital. So it's like, it's not fancy, I'll just say that.

00:15:06 - Tina Tang
So it hadn't been occupied for years and years and they kind of dropped us off there and they were like, okay, this is where you're going to work. And it felt very much like a startup situation because it was just tiny team of four and they were like, okay, we're completely new to Agile. This is the first time this client is trying the Agile thing. So let's figure out what all that is. The product owners, you go to work with the clients and figure out what we need to build. We were replacing a legacy system for receiving benefits in North Carolina. They were like, okay, go figure that out. And we got to this building and it was kind of scary.

00:15:51 - Scot Wingo
You're kind of wondering, this been checked for asbestos.

00:15:54 - Tina Tang
Yeah, yeah. We had heard rumors it used to be a mental hospital or a mental institution. And so when we would work like late at night, it would get kind of creepy and when we turned on the water, it was brown. Rpm was like, we're going to fix that, it's okay.

00:16:11 - Scot Wingo
Yeah.

00:16:12 - Tina Tang
And we're like, okay, bring your own.

00:16:14 - Scot Wingo
Water to work for sure.

00:16:15 - Tina Tang
And that project was kind of where I picked up. I didn't call it customer discovery at the time, but customer discovery skills. Because my product owner.

00:16:25 - Scot Wingo
And you have to do your little cards, your stories. Right. So a story, you know, that's the nice thing about Scrum. It forces you to go talk to a customer to be able to create your stories.

00:16:34 - Tina Tang
Exactly. So my role was very, was kind of all client facing. And for the first few months I would just meet with the client and talk to her about the old system and kind of what the problems were, what the pain points were. And that helped us design the new system that we would replace it with.

00:16:57 - Scot Wingo
This super old like mainframe type, green screen type deal.

00:17:00 - Tina Tang
Yeah, like press F4 type deal.

00:17:03 - Scot Wingo
Sometimes that's the hardest because it's hard. The customer can't conceive like what that's going to look like in a modern context.

00:17:10 - Tina Tang
That was one of the biggest challenges because the way they talk about their problems is like, well, then you press F4 and this happens. And then you press F2 and this happens. And when you're designing the new system, you can't just replace those clicks. It has to be a complete overhaul.

00:17:26 - Scot Wingo
For 50 years, people hitting F2 and F4.

00:17:29 - Tina Tang
So that was really interesting. And I got to see kind of the team grow. So after we built out the requirements, then we hired the team. So the team quickly grew from 4 to 80 over two years. And at the end of it, we were able to implement this software across all 100 counties in North Carolina. And that was. It was just very interesting because the counties are so diverse. Buncombe is very different from Wake. And so having to design a system that works for all the counties was very interesting. Like the. There are some folks in some counties that have, like, one person that would manage the system, and in others, it was like 30. So you had to. It was just very interesting working in this complex system.

00:18:19 - Scot Wingo
And then one thing that you mentioned. I want to make sure listeners and viewers understand what the heck is computational neuroscience.

00:18:29 - Tina Tang
So Anthony's degree is in biomedical engineering. And whenever I say that, I get a lot of confused looks, usually.

00:18:35 - Scot Wingo
Let's start there. I know what that is. But just in case other people don't.

00:18:37 - Tina Tang
Yeah. So what he does is he builds computational models of how the brain makes predictions and kind of anticipates reward and things like that.

00:18:52 - Scot Wingo
Yeah. Cool. And that's computational neuroscience. Yep. So taking. Studying the brain and then trying to replicate it in code and seeing if we can learn from the brain and, you know, how it works, which really don't. Turns out, whenever you talk to someone in this field, we don't know much about how the brain works, which I find, you know, just pretty wild. Yeah. A lot of stuff we do in the brain is almost like, you know, basically back when medicine was putting leeches on people, a lot of the brain stuff is, like, pretty. Yeah. They don't know a ton about it. Maybe Anthony would disagree, but I think. Yeah. Yeah. I don't know what percent we know, but they. Most of them will admit we know very little about how a lot of it works.

00:19:29 - Tina Tang
Still very exploratory. It's a very interesting field. If you want to learn more, you should definitely talk to Anthony.

00:19:37 - Scot Wingo
Okay. Yes. All right.

00:19:38 - Tina Tang
He's the expert.

00:19:39 - Scot Wingo
If we want to do a deep dive into computational neuroscience.

00:19:42 - Tina Tang
Yeah.

00:19:42 - Scot Wingo
We need another host for that one, because I'm not sure I could hang with him.

00:19:46 - Tina Tang
You could.

00:19:47 - Scot Wingo
Yeah. So you're chugging away you're building, you know, you're, you're helping, you're learning Scrum and you're. It's kind of like your first Agile. You know how to do Agile software development. Your product owner, did you get a lot of product owners and Scrum Masters have friction. Was their friction with your Scrum Master?

00:20:04 - Tina Tang
Yeah, a little bit.

00:20:05 - Scot Wingo
It's kind of designed. I don't. That's what I don't like about Agile. When, when we did it at channelvisor, we got, we kind of made it one function. Like, I couldn't see the benefit of having two. Like in the, you know, this, the Scrum Master, people I've met, they're all like, no, I'm going to like. But I am the voice of the customer and I'm going to like, stop everything. And it's true. Just like, way too aggressive about it.

00:20:25 - Tina Tang
And it's just like, yeah, they're trained to be aggressive. It's like part of their role. I think that's why they love to.

00:20:32 - Scot Wingo
Fail the Sprint too. There's like all this drama around. The Sprint failed. Oh, God. This is like, way too dramatic.

00:20:38 - Tina Tang
Yeah. Or. Yeah, there's always. I found we disagreed mostly on prioritization things, but it ended up working out. Like, we work together pretty well, but definitely some disagreements, but he was very nice.

00:20:52 - Scot Wingo
Like, teaches you how to. You got to work with some friction and that happens a lot in startup land for sure. So you're living in Raleigh and working at Accenture. Then what happened?

00:21:04 - Tina Tang
So at the end of this project, we've deployed the soft, released the software across all the states. It went very well. I got a promotion and I bought a house. And that was kind of what. So in my mind, I kind of achieved what my parents wanted me to achieve, you know, get it. They always said, get a good job where you just, you can just sit in front of a computer and you don't have to do like physical labor and buy a house. And so how.

00:21:39 - Scot Wingo
When was this? What year?

00:21:41 - Tina Tang
This was 2018.

00:21:42 - Scot Wingo
2018. Yep. How old were you generally? So you're like 25 and you've bought a house.

00:21:47 - Tina Tang
28. 28. It was kind of like, okay, now what?

00:21:50 - Scot Wingo
It's kind of anticlimactic. Boom. You're 28 and you've hit your goal.

00:21:54 - Tina Tang
Yeah, the promotion was kind of this fleeting moment of fulfillment, and then I just felt like going, like, continuing on that career path wasn't really for me. I think I had bigger dreams. Anthony and I would always talk about kind of what the future would look like in tech. And we thought the future was in machine learning. And where I was, I felt like Anthony was learning so much and I was kind of growing at a much slower rate because at that time.

00:22:33 - Scot Wingo
This is interesting. We'll kind of weave the evolution of AI into your story. So at that point, so neural networks were like this backwater and you could do like little things with them, but they were kind of useless. And then they started to get some momentum and then DeepMind came on the scene and they came out with. I remember when they had the robot, the. I forget the name of it, but it won at Go. The game, the Chinese game Go. And that's when. Well, first of all, you had like the one, the IBM one went on Jeopardy, but that one was like. I don't know that one wasn't as like earth shattering. But the thing that was really interesting about the. When DeepMind had the. Ultimately Google bought DeepMind, but this is around that time when it won at Go they. And it beat the GO Masters. It was interesting, but they, they couldn't understand how it got there. That's like. That caused like a lot of ripples out there. That's when Elon Musk was on the board and he was like, whoa, this is kind of scary. We need to kind of know what this thing is thinking about. Even like GO Masters couldn't understand. So when you play go, you tend to like start kind of clustering one of the areas of the board and it would start putting things kind of like way out in the middle of nowhere and ultimately those things would win. But no one knew how it knew to do that. Yeah, I don't think they even know today. So it was really interesting, like as an observer outside the AI world watching some of this stuff. But then somewhere, once DeepMind came on and I think what they did is they start. Started back propagating the networks and they could learn faster. Then that's when ultimately this will get to GPTs through the T part of GPTs, which is the transformation, which is a faster way of back prop. So, yeah, so AI is like cooking right now. It's like on the curve. And you're helping people with F2 and F4 and he's learning all about. He's on the. Yeah, I can totally feel that.

00:24:25 - Tina Tang
And so I just felt like I was behind and I decided to quit my job and go back to grad school as well. And I reached out to my undergrad advisor back at UVA and she told me about this systems Engineering master's program at UVA that was very flexible, would allow me to get into machine learning and kind of explore my curiosities because it was very flexible in terms of what courses you would take. Kind of as long as you got your research done. And it was fully paid, gave me a stipend. So that was really nice because at the time, getting a master's paid for was. Was not common. Like, Anthony didn't have that offer at. For the. Back when he wanted to go to Columbia. So I was like, oh, this is. This is great. And ended up going to UVA for the systems engineering masters.

00:25:21 - Scot Wingo
And did you have to move or could you do that kind of somewhat remoter? You had to move back to.

00:25:26 - Tina Tang
So I did relocate to Charlottesville for the master's and it was very familiar because I did my own. Yeah, the faculty was very supportive. The community was very supportive. And before that, I kind of had these anxieties about starting over. Like, I had been on a career path and now felt like I was starting over. And there was some anxiety with that. But as soon as I got there, it was such a fulfilling experience because I'm out of this kind of working at a breakneck pace and I could focus on new things and think. And I had this kind of academic playground to explore, machine learning. Now I took computer vision courses with Dr. Vicente Ordanez, who became a mentor for me. And I was very interested in kind of what you can do with images and images combined with text, multimodal. And by the end of my master's, I had worked on kind of a lot of side projects where I would apply what I was learning. And in one of those projects, I, the Smithsonian had released a bunch of images as a research data set. They were black and white portraits. And I had trained a general adversarial network to add color to those images. So like a black and white image of Amelia Earhart, the model would add color to that in a matter of seconds. Where previously artists had to spend 300 to 3,000 hours colorizing a single image, and they kind of worked pixel by pixel. So it was really cool that this model could do that. But then there were all of these kind of human edits that I wanted to do that wasn't just like, add color to this, but, oh, let's change the color of her accessories and things like that. So I did that in Photoshop and that's when I kind of experienced the combination of machine learning and human creativity combined.

00:27:47 - Scot Wingo
Interesting.

00:27:48 - Tina Tang
And I thought that was very inspiring. And it made me think about how powerful that technology was, but also how inaccessible it was because someone couldn't just go and do that at the time.

00:28:02 - Scot Wingo
Photoshop befuddles me. I can't. Yeah, I can't grasp it.

00:28:05 - Tina Tang
Yeah, Photoshop was really hard to use, but also like training a model is not easy for the everyday person. So what if. And I kind of started thinking about what if that type of technology was accessible to the everyday person. We all take images, like great images with the powerful camera, powerful cameras on our phone, but we don't have that many photo editing tools that were easy and powerful. So you have the desktop ones like Photoshop that, you know, were really hard to use. And then there was this trade off you made when you used photo editing programs on your phone where you kind of have to accept, you have one tap filters, you have these like kind of simple edits, but you couldn't do more. It was like desktop is where you do is the really intricate edits phone. The phone is where you do like playful one tap things that didn't give you a lot of creative control. And so that was the point where we started thinking about, well, what if we built something that enabled the everyday person to take the images on their phone and create whatever they wanted as if they were Photoshop experts but, you know, using software on their phone.

00:29:18 - Scot Wingo
Yeah, I have a bunch of questions. What's a general adversarial network?

00:29:23 - Tina Tang
So it was a generative model. It's kind of like the early generative AI. They weren't super smart at the time, but you could train it to do things like colorize an image, kind of specific tasks like translating one set of images that kind of has a set of characteristics to match another set of images that had a set of characteristics.

00:29:48 - Scot Wingo
So you would train it on a human colorized image and then that made it smarter at doing a machine colorized image kind of thing.

00:29:54 - Tina Tang
Just so they didn't have to be. One of the new things was they didn't have to be paired images. So you could have a data set of, you know, black and white images and then a completely separate set of just, you know, images of in color. And they didn't have to be, you know, the same image, black and white, the same image, colored. That was a cool thing back then. Yeah.

00:30:18 - Scot Wingo
So you started off saying when you went to undergrad you kind of wanted to do a lot of things. And it feels like this is this perfect, really interesting combination of your creative side and your math and science side.

00:30:31 - Tina Tang
Yeah, I was Very excited about that, that this was a potential path where I would get to explore both sides of what I like to do in kind of down this other career path. So after my master's, my computer vision professor, he always appreciated kind of how I like to apply AI in clever ways. And he referred me to work at Adobe as part of their AI team. And so I was interviewing with them to be a product manager and they were pretty excited about, about that role, but kind of the. But I would have to move to San Jose. You know, being close to my family, like right now I'm still driving distance to my family. That was important to me. Anthony was here as well. And so I was kind of asking them, like, what if could I. Is there a way that I could kind of travel for the job but be based in the triangle? And they were like, no, that's a deal breaker.

00:31:40 - Scot Wingo
What year was this?

00:31:41 - Tina Tang
This was 2019. Sorry, 2021.

00:31:46 - Scot Wingo
Oh, wow. So it was post Covid still and they still wouldn't. Wow, that. Wow, that's interesting.

00:31:50 - Tina Tang
Yeah. They were like, no, the AI team is very concentrated. We want everyone to be in person. And at the same time, I was also talking to my computer vision professor about this kind of initial version of Bristles, the, this photo. Photo painting tool that would allow people to use this, use this virtual stylus that would let them combine different model outputs without them having to think about it. So it'd be like you take a photo and you can kind of paint on different effects that gave you more, a lot more control than say a one tap filter would. And you can kind of mix and match a bunch of these effects.

00:32:39 - Scot Wingo
Yeah. And until machine learning and AI, it was very hard for a computer to do image recognition or object recognition. So we took a picture of this wall. It couldn't. We're looking at a wall with a little shelf with a vase on it. It couldn't identify the vase. I'm actually sidebar. I'm familiar with this because in the e commerce world, you have this problem of you have all these product pictures and like, let's say you have a picture of a pair of shoes in your Nike or under armor or something, and then you, if you're Nike, you've taken them with a certain backdrop. But Amazon has a policy that it has to have a white backdrop. So it seems easy, like, okay, I'm going to change the backdrop from, you know, blue to white. But it's very hard because you can't. It's not just blue, it's like you know you can't just like do the little paint fill can thing and drop it on there. So in the E commerce industry there's this whole sub industry of people that will change backgrounds to white, change the foreground to this, do that, change the rotate something. So it's a. Until this technology came along, this is very, very hard to do. So part of what you're, not only are you kind of changing that, but you can identify to your point of changing in your pictures, changing the accessories and the colors. To do that you have to identify the object. This is an earring and I'm going to change it from silver to gold. Kind of a color, kind of a thing which is, it's a non trivial problem. So is that still one of these adversarial networks that does that or is that a different.

00:34:03 - Tina Tang
The initial version of our bristles prototype was using the GANs.

00:34:08 - Scot Wingo
Yeah.

00:34:09 - Tina Tang
Yeah.

00:34:11 - Scot Wingo
Cool. So you could go like work for Adobe or start your own company.

00:34:17 - Tina Tang
Yeah, so that was kind of the two paths. And my computer vision professor also, he actually has a triangle background too because he got his PhD at UNC.

00:34:29 - Scot Wingo
Yes. Somehow I know him.

00:34:31 - Tina Tang
Oh okay. So he always kind of encouraged entrepreneurship too and he was excited about this idea. He's like, I think you're onto something, you should pursue it. Have you considered looking up grants? And so at this point Anthony and I decided, you know, we are going to do this startup thing and Anthony is going to finish his Ph.D. so he'll be, you know, doing it part time and I'm going to go and kind of get it rolling and figure out how to do the startup thing. And so I was looking up like is there, are there grants for businesses in North Carolina? And I stumbled across NC idea. So I ended up applying with the initial bristles idea of this photo painting app.

00:35:26 - Scot Wingo
We're still in 2021 at this point. Yeah.

00:35:29 - Tina Tang
And the application process was kind of eye opening too because it makes you think about certain questions.

00:35:38 - Scot Wingo
Yeah, it's like the longest form I've ever seen.

00:35:41 - Tina Tang
It's a pretty long form but it's worth it.

00:35:43 - Scot Wingo
Yeah. And it's, it's basically, you know, they want to hear your whole business plan and like if you're staring there, looking at that and you haven't thought about it, it's kind of daunting.

00:35:52 - Tina Tang
So I applied and ended up getting. So there's an interview or there's like a first round of feedback that you get and I remember reading the feedback and I don't know, do you know John Austin, I think so, let's say more so. John Austin was a big part of NC idea and a big part of entrepreneurship in the triangle. He started, I'm sure, I know as part of an accelerator that was out of AU back in the day. And his feedback was funny. He said I've never seen an application where I love so many things and also I hate so many things. And what he liked was that we had built this prototype. I had a demo video in the application and he was impressed with what we built. But what he hated was that we hadn't talked to any customers. We had done surveys. And he's like, surveys don't mean anything. And his feedback was really helpful because that encouraged me to go out and you know, test or show people the product. And at the time the, our product was very kind of vague. It didn't have a specific problem with solving. It was kind of just a better, more control. Photo editor on your phone?

00:37:21 - Scot Wingo
Yeah. Were you thinking you would go to Pro audio pro photo type people maybe as a start or like photographers? It's kind of a, like a touch up kind of a, like a premium kind of a customer or you just didn't know at that point?

00:37:34 - Tina Tang
At that point I thought it was going to be for everyone. Okay, so it's like everyone is going to use this because Photoshop is so hard, but if it was easy, everyone has ideas that they want to visualize. It was kind of what I was thinking at the time.

00:37:49 - Scot Wingo
It's not crazy because there are some of the top apps in the app store are photo editors like pixelmator. You probably know them better than I do. There's a lot of people that struggle with this Microsoft Paint and Photoshop. So there's definitely a need in the middle.

00:38:06 - Tina Tang
And so I created this demo video of different things that you could do with the app. And it ranged from taking a photo of a lake and painting a sunset in the sky, painting in clouds to kind of painting the leaves on a tree and they can change to different seasons to a picture of someone's face and you can add makeup earrings to a photo of a door and painting it in a different color to a photo of a window and adding a curtain to the window. And so that was kind of the demo video with the different scenarios and when I showed it to different people and a lot of the time it was friends or kind of folks I've met in the, in the triangle and they, and everyone kind of focused in on the home scenarios where their reaction was just different. With the painting of sunset in the sky, they're like, oh, that's kind of cool. I don't know when or if I would do that, but when I got to the home scenarios, they were like, oh, so I can see what different products would look like in my house. I could test different paint colors in my house. And all of a sudden it was kind of this. They had this sparkle in their eye that was different. They kind of talked about how useful it would be rather than how cool it would be. And so I picked up on that. And I was still talking to friends, so it's kind of still biased in terms of customer discovery. So then I stood outside of Home Goods and just showed it to people, kind of just walking out and asked them what they thought.

00:39:59 - Scot Wingo
That's a big leap because I'm somewhat of an introvert and you are too. So that takes some guts to kind of just. Hi, I'm Tina and I'd like to get your feedback on this app I've made.

00:40:10 - Tina Tang
Oh my gosh. It was kind of terrifying.

00:40:12 - Scot Wingo
Yes, it is.

00:40:14 - Tina Tang
Anthony definitely gave me a pep talk before I went out there.

00:40:17 - Scot Wingo
You've got this. This is where it helps me in North Carolina. So everyone in Boston, you would have. You would have hung up your spurs. Yeah, probably.

00:40:26 - Tina Tang
So I did that at Home Goods and outside of Crate and Barrel at South Point, when I would talk to, you know, strangers, they. They thought it was a really cool concept and also focused on the home decor scenarios. Well, I biased it a little bit, I guess, because I went to the home stores.

00:40:45 - Scot Wingo
But you were narrowing your focus and you wanted to get more validation.

00:40:50 - Tina Tang
Yeah. So we're still in the NCI idea kind of application process for their 10K micro grant, which is their kind of prove out your idea phase grant midway through. So before the presentation pitch part of the application, we pivoted from photo editing to a home improvement visualization tool and had a much clearer business plan in terms of the customer we were targeting. And we ended up winning a grant.

00:41:24 - Scot Wingo
Yay. One of the 10Ks. Yeah. So this program, I think it's from the tobacco settlement money funds, the NC Idea program, if I recall. You know, so there was this litigation against Philip Morris and all the manufacturers in North Carolina got some dollars. And I believe that's the. Where the corpus of the money comes from. And you know, so. So Tom Rue and. And a whole crew of folks there do a great job of managing that. They have the 10k kind of like the earlier super early stage kind of, you know, super Like, I mean, put pen to paper yet kind of thing, or code to compiler and then they have the 50K. So you got one of the 10Ks.

00:42:05 - Tina Tang
Yes. And at the time I remember John Austin saying, you have a lot of work to do learning about customer discovery, but you can build a product. And that's the part we can't help you with, but we can help you with customer discovery. So that was kind of my first bit of, okay, I think we're onto something because folks that know about starting a business believe we can do it. So Grep Beat reached out to all the micro grantees and did little articles on us and that's when I met you. So we met up for coffee at Starbucks and I was telling you about bristles and that was kind of the second moment of, oh, someone who is really good at entrepreneurship sees potential in what we're doing. And I think these little bits of. I think it's really powerful when someone believes in you, when you have, you're trying to find evidence that what you're doing is worth pursuing. And so those little bits of, okay, we're onto something. We're really impactful.

00:43:21 - Scot Wingo
Yeah, being an entrepreneur is like one of the most exciting and lonely things. You're kind of constantly like, you're in the desert looking for the Oasis and you're like, am I on the right track? What's going on here?

00:43:32 - Tina Tang
Yeah, yeah. I'm super grateful to have all of these mentors. So Dr. Ordonna's was very impactful in kind of making me believe that I'm good at this computer vision thing because there was when I, you know, I never saw myself as, you know, good at programming or anything like that. But so it was really helpful to me that he kind of believed in me in that area. And then there was this entrepreneurship side that's completely unproven. And so it was really helpful to be a part of NC Ideas programming and to meet so many mentors like you here. That kind of helped me along the way.

00:44:17 - Scot Wingo
Yeah, yeah. That's what we're all about in the triangle. The, you know, on my side, it's very unusual to find, you know, number one, female entrepreneurs are underrepresented. Just a lot of them don't come through the pipeline. So that was interesting. AI at the time was very novel. Now it's like, you know, not. But like you were super early on that and I was kind of like, this is pretty. And I was starting to follow the GPTs. This is before chat GPT, but I think we were at chat, we were at GPT2 and I was starting to kind of like get a, you know, be pretty curious about that. It was multi shot and it was complicated, but. And then I know Robby who's like deeper in AI and he like kind of. So if I need to go deep I can just ping him. And then you were solving a problem I could, I knew a fair amount about because of the e commerce world. So that's a, you know, that's what was exciting. It's like, wow, this is a, you know, three rare things kind of in one person.

00:45:10 - Tina Tang
Yeah.

00:45:10 - Scot Wingo
It's also unusual to find technical female founders. So there's a rarity in that as well.

00:45:16 - Tina Tang
Kind of the way the timeline of things went. The Triangle Twiner funds investment is actually the first money in for bristles after the.

00:45:28 - Scot Wingo
But we were after the idea, the 10K.

00:45:31 - Tina Tang
So we had your grant. We had won the grant but we hadn't gotten the funds because there's a kind of a process to that as well. But yes, so that was kind of the initial funding we had to get started to prove out our idea. And then the fun began with customer discovery.

00:45:46 - Scot Wingo
Yeah. So then now you're in the part of the journey we call product market fit. Which is, which is hard. And this is the, this is the. It's hard to mentor people in this because it's literally go talk to customers, build something, iterate and you just gotta like keep iterating and iterating. Iterating until you kind of get there. So yeah, it's a, this is where the accelerators are kind of interesting because they, they, they have a whole program for doing that and they kind of force you. That's basically what the Y combinators at the end of the day they have other parts like a mentor program and a demo day. But the programming, the classroom part largely is basically forcing you to have a cadence of talking to customers and iterating fast. So yeah, you've had to do that on your own.

00:46:26 - Tina Tang
Yep.

00:46:27 - Scot Wingo
So that was two years ago and no, I guess longer. So three years ago. So you've been in product market fit journey for three years. Feels like a lifetime.

00:46:36 - Tina Tang
Yeah, yeah. So how it started was so we were trying to tackle this problem in home improvement. We had developed a, you know, a very early version of the product. You can call it an mvp and at the time you could take an image of your space and take another image of a product, kind of cut it out and recolor it and kind of apply it into Your Canvas image, we call it the image of your space, your Canvas image, it wasn't in an official app, it was kind of in a mobile web app format. And we were trying to solve this problem of how do we enable precision editing on a mobile device? Because we thought about, okay, what are the problems with Photoshop being limited to desktop Photoshop and other kind of professional photo editing software. And what we kind of landed on was on desktop, you have a very fine precision point because you're editing with a mouse on a screen. And on mobile your precision point is limited to the width of your fingertip because that's kind of where that's how you interact with a phone screen. But if we kind of enable this virtual mouse experience on the phone, then we let people do more, more precise edits. And so at the time, you could kind of interact with machine learning models via this virtual stylus on our mvp. And it was very clunky. And in terms of as someone with.

00:48:28 - Scot Wingo
Big fingers, I liked it though, because I could like, I could hold the little bottom of the stylus and then like the little tip is a very precise like a. And then it kind of like shows you exactly where you're going, which is very nice. Whereas the other problem is your finger covers up what you're trying to do.

00:48:42 - Tina Tang
Exactly.

00:48:42 - Scot Wingo
So you're kind of like blind to what's happening underneath.

00:48:44 - Tina Tang
Exactly. So we were very excited about this virtual stylus. So we had this mobile app that wasn't an app, it was a mobile web app. And our first challenge was getting our first customers. So we had talked to kind of people in the DIY space. We had created an Instagram account and gotten some engagement based on showing little demo videos of kind of DIY use cases on our Instagram account. And what we noticed was the. Even early on we noticed that the furniture use cases got more engagement than kind of the accent walls and things like that. So that was our first clue that there was something interesting there. Then our next challenge was, and NCID had pushed us on this, was figuring out if what we were building is useful enough that someone would actually pay for it. Deciding to charge was a very nerve wracking thing because we felt like our product was, you know, so early, not ready. And I think a lot of entrepreneurs struggle with that. But now I would advise just start charging and see what happens. It's better to test early. We had this like very early payment system off of Shopify where you could like buy a subscription. Then I would manually go and send you access to our web app. And we got, you know, quite a few people to pay for it that way.

00:50:29 - Scot Wingo
And what price point did you land on for at this point?

00:50:33 - Tina Tang
We, we were going for $8 or sorry, at that time it was 1499 a year. So very, very low pricing.

00:50:42 - Scot Wingo
So. But a buckish, a buck twenty something a month.

00:50:45 - Tina Tang
Yeah, yeah. And that was too slow, but it gave us, but we were amazed that anybody would pay for it. We're like, oh, someone, someone paid us for this through this weird process. And so we were onto the next phase of customer discovery, which is we know the product is not good enough right now, so we needed to know how to improve it. We didn't know exactly what to do without feedback. So I needed to line up interviews somehow with customers to really understand like what, what their problem was and how they interacted with the product to see where problems were. So I had this idea that we could put a link on our Instagram to a zoom invitation to do in 40 minute interview, video interview with me. And in exchange they get the product for free. And at first we were like, is anybody going to do this? It seems like a lot to ask to ask someone to do a video call with a stranger. They get access to this app who they've never heard of, this company they've never heard of. And we got 200 people to sign up for video calls, which was pretty crazy, I thought. And these video interviews were so helpful. You know, with surveys you can, and you know, data analytics, you can find things like people are spending time doing this, but you can't figure out if that time is a good experience or a bad experience. And you don't know why they're, why they want your product. And so with these 200 interviews, this very obvious recurring theme kept coming up, which was people wanted to use what we had built for furniture refinishing design and they were using our app. And what they wanted to do was test different paint colors on furniture, test different hardware on furniture to decide kind of what to do. So with furniture refinishing, you have a piece that's one of one that you want to update and you have all these anxieties because you don't want to mess up. If you mess up, it's hours or days of rework and you want to make sure it sells. So you kind of want to have a way to get feedback either from a client, if you have a client already, or from your furniture refinishing friends to ask which one do you think will sell better?

00:53:23 - Scot Wingo
So these are not really people doing it for their own use. There's some of that, but these are more like what I would call prosumers, where they. Maybe they did one or two themselves. They really realized they liked doing this and wanted to make a bit of a business from it. I know a couple of these types of folks, they'll kind of go around and find interesting pieces and they'll collect them usually in like a garage or something. And then. And then they'll kind of like, you know, they'll paint them and upgrade them. But the, you know, some of them will do it custom, where they'll have a customer, but otherwise they're. They're taking a. They're trying to understand, you know, what, what's a good color for this and what, what direction should I take it? And that's kind of. So that's what, you know, so it's interesting. It's. It is almost like a. It feels like consumer, but it's like one little tick above. It's not really what I would call SMB, but it's like a prosumer kind of a.

00:54:10 - Tina Tang
Kind of a little. Yeah, exactly. Yeah. So I would hear very similar stories interview after interview. And it's exactly what you said. They would work out of their garages. So they couldn't only work on a few pieces at a time because there's just a space constraint. And they would agonize over these decisions about paint colors and hardware and what would look best. So when they saw our social media videos of how the app worked, and especially use cases where we changed the color of a dresser, changed the hardware, change the base, things like that, they got really excited. And so that got them to find that link and sign up for an interview. And then, so in this interview, the first part was kind of learning why did you decide to interview with a stranger? And I would learn about their problem kind of without biasing them in any way. And then the second part of it was, okay, I'm going to demo the app for you, and then I'm going to have you try it on the call with me. And so in the call, I also got live feedback on the user experience, which was so, so helpful. So, for example, we had the virtual stylus that I mentioned, but a virtual stylist is not. It's not something a lot of people are used to. And because it extends from your fingertip, we kind of need a little bit of loading time before the effects start at the tip of the virtual stylus. And then, so there was a lot of confusion on what do you do if you move it around like nothing's happening. And I wouldn't have learned that if I didn't watch them use the product.

00:55:58 - Scot Wingo
Because in your head, oh, the model's loading. I need to just wait a second.

00:56:01 - Tina Tang
Yeah, yeah.

00:56:01 - Scot Wingo
Because you understand how it works. The customer's like, what the heck's happening?

00:56:05 - Tina Tang
Right? So they're like, you know, they're like.

00:56:06 - Scot Wingo
They don't care about the model. Like, you know, and you're like. And you're like, oh, that's cool. It's like doing this thing. And they're like, they don't care.

00:56:11 - Tina Tang
Yeah, they're. They're expecting editing to happen. At the point they're touching the screen.

00:56:17 - Scot Wingo
They want a paintbrush exactly, like an instant.

00:56:20 - Tina Tang
And so that's how we learned, okay, let's put a fingerprint at the base. So then it's obvious that you need to, like, that's. You're holding the base of the stylus and we need to show an animation of the loading so they know something's happening. And that helped us a lot.

00:56:37 - Scot Wingo
So this is good for listeners and viewers. The. So number one, you've talked to 200 customers, so that's good. Right? So you've learned pretty quickly. Got to talk to customers, our prospects. Number two, a lot of technical people, you don't have this. But a lot of, you know, here's a what not to do. A lot of technical people will say these are the wrong people because they don't understand that. They, like, get so obsessed that their solution is right. They won't bend to the feedback of the customer. I see that a lot and I get it. But, you know, that's like not how this works. Like, you know, those 10. Those companies tend not to make it out of product market fit because they can't, you know, and sometimes they'll be even arrogant enough to say they're just like, dumb. They don't understand how awesome this technology is. And you're like, well, that's not it either. You know, awesome technology doesn't matter if no one uses it. You may, you may get gratification at that point. It's a hobby. It's not a. Not a business without customers, it's like, not going to scale. Yeah, so those are some of the things what not to do. But, you know, the other thing you've done is you've. The way you've done discovery is you're asking them questions before they've even seen the app. Right. So you haven't tainted them on anything. And then you're kind of showing them that. For folks new to this, what did you learn about the types of questions to ask and then maybe like what any, any what not to do in those discovery calls?

00:57:56 - Tina Tang
Well, I think it was important to understand the problem before kind of like what you just said before, showing them your solution just to hear the pain points. Because if they're feeling some sort of pain, they will, they will tell you about it. One thing I think I shouldn't have done was demo the product before having them use it. Because in my mind I'm showing them, look how easy it is. Look at me use it. It's so easy. And of course it's easy for me. You know, we know how it works, we developed it. And then that makes it easier for, for them to pick it up because they've seen kind of the tricks I use to like they watched me use it and now they kind of know better than, than someone coming in without that. Without that demo.

00:58:49 - Scot Wingo
Yeah, Another, another lesson is sometimes you have this vision that's here as a founder. So yours was like renovating a house, Right. Which would include some furniture. So but then you've had to kind of go more atomic. Right. So a lot of times you get to product market fit, you start your vision's here and you're not going to lose that vision. But to get there, you have to kind of go more atomic and like solve a very specific problem. And then, you know, we'll, we'll project forward. But if you can solve this furniture problem and some of the stuff you're learning, there's no reason it wouldn't work with rugs. Wall coverings.

00:59:23 - Tina Tang
Exactly.

00:59:23 - Scot Wingo
Sinks.

00:59:24 - Tina Tang
Yeah.

00:59:25 - Scot Wingo
Countertops. So. So in a way, you haven't traded off your vision.

00:59:30 - Tina Tang
Right.

00:59:31 - Scot Wingo
But sometimes to really get product market fit, you have to like start at a very atomic thing and then kind of scale back up. So that's another thing people resist. A lot of times it's like, you know, the, and it's easy. Traditional venture will tell you they spend all this time talking about tam. This is one of my bugaboos. Total address. I'm sure you've probably gotten this feedback, total addressable market, blah, blah. So, you know, if you, if you use that lens, then it would be, you know, quote unquote dumb to go to a smaller. The home, the home renovation. TAM is huge. Helping people fit. You know, everyone would say that's not a venture scale business, but they don't understand that's part of the journey through product market fit is you've got to shrink the TAM to something you can solve and really drill in on, and then you can walk back up. But yeah, you know, so a lot of times it's very confusing for entrepreneurs, and I don't like that. I've had that myself a lot. Where the other thing is, if you're doing something really awesome, the tam, you create tam, a lot of people don't. Don't get that either. So Uber's TAM was like zero until Uber existed.

01:00:32 - Tina Tang
Yeah, we've definitely gotten that feedback and I've just pushed through.

01:00:36 - Scot Wingo
Yeah.

01:00:38 - Tina Tang
Ignored it mostly. But that is how we think because the technology we're building generalizes very easily upward once we kind of prove this thing out within the furniture refinishers. So at this point, we have. I would call it very, very early product market fit. It's not. And I say that because we. We started with trying to solve this, make this photo editing app that's as powerful as Photoshop, but has a lot of utility value and super accessible on your phone and easy to use. And we have solved that specific problem for furniture refinishers. So right now, the customers that love our product, they are the folks that have struggled to use something like Photoshop to visualize furniture refinishing changes and have discovered an easier way. When we first met, you recommended the book Crossing the Chasm. And so I kind of tried to look at it that way, that we've kind of solved this super, super early piece of it, and now it's a little chasm. So it's a bowling pin. It's for. Yeah, it's the furniture refinishers that are photo. That have photo editing experience and previous frustrations using really difficult to use software. And now we're kind of.

01:02:01 - Scot Wingo
I think it's interesting. One of the things you discovered in your discovery calls is they were using PowerPoint because they understood PowerPoint. So they like, tried Photoshop and gave up. And then they're like using PowerPoint to try to do this, which is like. Yeah, it's a. I don't know, a good analogy, but it's like bringing hammer to screw a nail or something is not the. It is not an imaging app, but like, people kind of understand it. Right. It's kind of simple enough and you kind of. It's got like foreground background and transparency and just enough to kind of like hack it, but it's not good.

01:02:30 - Tina Tang
Even bigger than the PowerPoint thing was a more common thing they did, which was to pull up Instagram stories, which I don't know if you've, if you've used it, but you can add a photo and there's a little marker tool. And so they would scribble different colors on top of images. And a lot of the time that was their mockup of different designs when they asked for feedback. So when I saw that, I was immediately thinking we could definitely solve that 10 times better.

01:03:04 - Scot Wingo
Low bar to jump over. It's crayons and you've got a fancy pen.

01:03:08 - Tina Tang
Yeah. And Photoshop exists, but they're doing this. So that was really interesting. And so back to kind of the crossing the chasm chart. I think we're very, very early on the left side, but we've kind of solved the problem for. You've struggled with Photoshop and you're a furniture refinisher. And now we're kind of getting the furniture refinishers that haven't ventured into Photoshop but have heard about us and they're interested, but because they haven't spent hours doing a mock up on Photoshop, now they come into bristles and it doesn't feel easy for them. So we've kind of crossed into a different segment now of folks that we need to improve the product for, like, even more so in our latest customer discovery interviews. Now we're hearing the folks that love our product are still spending like 20 minutes mocking up a design. And this is a huge improvement compared to, you know, hours on Photoshop. But 20 minutes is a long time. And so the next thing we're thinking about is how do we reduce that 20 minutes to 2 minutes? And so we're working on bristles 2.0 now, which spoiler.

01:04:22 - Scot Wingo
Ooh, you heard it here first.

01:04:24 - Tina Tang
And so we're integrating a lot of AI automated automations in order to make some of the things they're spending time on easier. Like right now they have to copy individual pieces of hardware. And so we can reduce time by kind of automating it. It's complex behind the scenes, but we're excited about solving that.

01:04:46 - Scot Wingo
Yeah. And even in Bristles one, you have a generative AI piece. Talk a little bit about that and how you came up with that idea.

01:04:53 - Tina Tang
Yeah, so we, we're starting to get some feedback about. Like I said, the users that had used Photoshop thought it was easy, but a lot of folks still kind of struggled with some of the, with the virtual stylus. They felt like you still needed a lot of dexterity to use it well, and they just didn't have the patience to kind of get to get from start to mock up. And that's what they wanted. So one idea was, well, we'll integrate generative AI so you can, you know, come up with designs using a prompt. We'd have an inspo feed where you can generate ideas based on a photo of your, of your piece if you didn't have an idea already. And we had some users that really liked it and some users say they still don't have enough control because what they really want is to test specific products. So a lot of folks that come into our app, they have specific paint colors that they either have on hand or they want to try. There's kind of hardware they're shopping for and they want to test out those.

01:06:07 - Scot Wingo
Specific products by test. They want to take the piece of furniture, do the, do the upgrades and then show it to a potential buyer. That's the test.

01:06:19 - Tina Tang
Or just see it themselves.

01:06:20 - Scot Wingo
Or they just want to see it.

01:06:21 - Tina Tang
But yeah, both of those.

01:06:22 - Scot Wingo
Sometimes they want to share it downstream and see if like it'd be interesting to do ab. We're going to do some like brainstorming here. It'd be interesting to do a B testing where they could then send to their customer, like, you know, different options. Yeah, different options.

01:06:33 - Tina Tang
Yeah, yeah. Something we're thinking about as well as making it a more dynamic experience. So they're not sharing just a photo, but kind of an interactive document type thing.

01:06:44 - Scot Wingo
Yeah, yeah, A configurator. That's what we used to call it back in the day. Very cool. So, and what's interesting is when you started this GPT one was kind of like coming and then, you know, now we're at, we're heading towards GPT5. So the underlying AI infrastructure is really, you know, supercharging your ability to do things and sometimes starting early when that wave comes, it can kind of lift your lift with all the boats, which is interesting.

01:07:10 - Tina Tang
Yeah, we've definitely experienced that. Well, you have to prioritize with if you're a scrappy startup, which we are. And so we, for a while early in the early days, we were thinking about spending resources training a very specific kind of furniture segmentation model. And we ended up prioritizing, working on UI things to integrate AI into our, into our app. And meanwhile like meta releases segment anything which had we spent time on that segmentation model would have, you know, all that work would have been obsolete when meta release segment anything.

01:07:54 - Scot Wingo
So yeah, the lesson there is like invest in the part that touches the customer the most. That's the most, that's like the most unique. That's your intellectual property.

01:08:04 - Tina Tang
Yeah, yeah.

01:08:04 - Scot Wingo
And then like understanding how to translate that to what's underneath the AI capabilities that are available to you at any given time.

01:08:10 - Tina Tang
And there's a lot of things you can do to stack kind of classic machine learning techniques with everything that's new and coming out.

01:08:20 - Scot Wingo
Yeah. To avoid errors you don't want. Like the LLM type of a thing will like, you know, distort the dresser or whatever. Like make it look like a weird anamorphic thing. Like the. That stuff's still very relatively error prone. So. So imagine you have to like, you know, make sure there's like some, some middle ground there. Yeah. So bristles two O's come. When is that going to come out?

01:08:42 - Tina Tang
Q4.

01:08:43 - Scot Wingo
Ooh, wait, we're in Q4.

01:08:45 - Tina Tang
Yeah. So we're going to release it iteratively as we develop it. But that's what we are super focused on right now. And it's going well. We have preliminary versions of pieces of it working Good. Yeah, yeah.

01:09:00 - Scot Wingo
So just to recap where you are, you raised a pre seed round. We participated in that. And then about how much did you raise in there?

01:09:08 - Tina Tang
So our precede in total was a little under. So 280k including grants.

01:09:14 - Scot Wingo
Yep, yep. And then you have this wonderful thing called revenue and that is the best non dilutive kind of thing that you can have to whatever extent you're com. I would, I would give a range of some kind but you know, it's been fun. You watched, you started at zero and now you've got your monthly revenues going.

01:09:30 - Tina Tang
Yeah.

01:09:31 - Scot Wingo
It's not like you know, retirement money but you know, this is how, when you start atomic, this is, this is kind of how it works.

01:09:37 - Tina Tang
Yeah. So yeah, we scaled from 0 to close to the 200k ARR stage. Yeah. And it's keeping us going.

01:09:48 - Scot Wingo
So. Yeah, yeah. And then. And you're doing that through, you know, very small. So it's, it's these prosumers and on average they're paying you like what, you have more of a monthly type of a sub now, right?

01:10:02 - Tina Tang
Yeah, it's $8 a month currently. Or they could pay for annual upfront for 48 a year.

01:10:09 - Scot Wingo
That's a big discount.

01:10:10 - Tina Tang
Yeah.

01:10:10 - Scot Wingo
You want to increase settlement, like 20% discount is probably good, but yeah, I get you.

01:10:16 - Tina Tang
Yeah, we might do that with the 2.0 release.

01:10:18 - Scot Wingo
When you're a startup, cash today is worth a lot more than over 12 months. What's the mix of Monthlies and annuals.

01:10:28 - Tina Tang
Right now it's like 40, 60, like 30, 70 actually closer to. So mostly monthlies and some annuals.

01:10:40 - Scot Wingo
Yeah.

01:10:41 - Tina Tang
And we, we've done some paywall testing to. There are little tricks you can do to. To try to get more or. Yeah, to try to get more annuals versus monthly.

01:10:53 - Scot Wingo
Yeah, yeah, sometime. Well, pricing is very tricky at this level to kind of figure out the freemium and where to put the break and when to. You end up having to do a lot of AB testing, which can take time. So the other thing that's hard as a small team is you've got this pie chart that's your time and you can spend infinite time on understanding the go to market strategy, understanding the pricing, adding features, talking to customers.

01:11:19 - Tina Tang
So it's a tricky one for sure. I would definitely say we haven't figured out pricing yet. This is just still our early test of it. One thing that's really interesting is when folks get to the point of they have a mock up that they're really excited about, they're excited to start the project and it does. And it's at the time where they would buy materials for these projects. So one thing we're testing down the road is also affiliate sales with those products they would buy. So hardware, paint, tools.

01:11:51 - Scot Wingo
Yeah. This has been your, I don't know what you put in your form that you submitted to NC idea, but when we first met that was like one definitely one path you were, you were looking at. So it's neat to kind of come back around to that. It's different than you thought.

01:12:01 - Tina Tang
But yeah, yeah, yeah.

01:12:03 - Scot Wingo
So let's, let's go two directions. Let's look backwards first. So you know, you've been at this roughly three or four years. How you know, it hasn't been easy. A lot of people, you've gotten a lot of no's. Yeah, I feel a lot of empathy for that because I've probably had, you know, I've probably had 10 yeses and 3, 4, 500 no's. So that's. I always tell founders like that's the setup. You got to get used to a lot of people saying your TAM is dumb. This is a crazy idea. I would never invest in a consumer company. And you're like, it's not a consumer company. Well, yes it is. So you've got a lot of no's, but you've had some yeses too. So what would you tell founders that are thinking of taking the plunge and they're at day one, I think there.

01:12:52 - Tina Tang
Are a lot of ways to fund your startup. There is kind of traditional vc, there's grants, there's accelerator programs. So what we learned is we could do a lot more than we originally thought we could. You know, when we first started out, we thought, you know, we'd have to raise funds to build a dev team and build a marketing team. All of these things that we kind of out of necessity figured out on our own.

01:13:24 - Scot Wingo
Sometimes that can hurt you. Like imagine if you're day 10, you had like $2 million. You may have figured that out. But it is very hard to have the discipline to not just go spend all that money very quickly. You really have to have the discipline to kind of wait until you've got something that hopefully bristles too, really takes off. Then now you can just really scale it. But you gotta kind of, you know, that's what's interesting about the triangle is we, we don't have. A lot of our failure rate is lower because we don't start scaling until we've, we've kind of made it through to this repeatable kind of a place. So. Yeah, yeah. Do you guys. I'm kind of a recently obsessed with this idea of seed strapping. Where we're seeing this is kind of a generation after you guys. So I kind of like draw a line at ChatGPT. So then companies are starting after. There's where you look at their models and they may not, you know, they want to build big companies and only maybe raise like three or four million dollars. And that's it because they're, they're, they're, they're leveraging the co pilots like the GitHub one and the cursor and replit and there's all this stuff. Are you guys, do you kind of have that mentality where you think you could get pretty far here maybe without having to raise a ton or do you still want to go chase the bigger thing? Like this is kind of more like, let's look towards the future. Where do you see things going?

01:14:48 - Tina Tang
Yeah, we, we do have future plans of fundraising, but not right at this moment. I think we, right now we have definitely utilized these tools coming out. Like Anthony now is full time at Bristles and he, he's able to work on development at a higher level now with these tools. So you still need to know kind of how to advise like a Claude or a ChatGPT to help you write code. But it can be now it's at the, like, he can, he can kind of think about what functions need to stack on top of each other. But then kind of the nitty gritty of it. He can have like, it's like an.

01:15:30 - Scot Wingo
Architect and then people doing the spackling. So you know he's building out the frame.

01:15:34 - Tina Tang
Yeah.

01:15:34 - Scot Wingo
Just like filling in the.

01:15:36 - Tina Tang
So now he's like a team of three. It's just him. So in that way we can do a lot more with a smaller team. So we're able to get to 2.0 for example on our own and then from there we're going to kind of reevaluate, see how it goes on the.

01:15:52 - Scot Wingo
Prioritization of pricing or more features. My advice always is there's a Jeff Bezosism is like solve demand. Like solve customer demand first always. And then pricing will come along like it's, it's so I see people, especially people with MBAs not to pick on them but because they've taken so many classes on pricing they want to really nail that before they put a product out. But you know, imagine you had tried that on day one your, your customer base is switched like you know, so, so it makes sense to like go solve the product part product led growth, what we call PLG gear PLG going get your product to really hum and get all this customer engagement and then that will give you. Then you can. Pricing becomes like so much infinitely easier because maybe you discover there's two tiers like there's the super prosumer and they have a different type of pricing because they have some features because they already knew Photoshop and then you have these junior beginner people and maybe they have a different kind of like you don't know what you don't know until you get like the product really humming and just kind of having a simple pricing that generates some revenue creates a little bit of an economic feedback and that's 200k. ARR. You didn't have to go raise. So non dilutive customer capital is always the best.

01:17:08 - Tina Tang
Yeah.

01:17:08 - Scot Wingo
Because it comes with feedback and people using the product.

01:17:12 - Tina Tang
Exactly. And so I think we've proven that we're providing value to some customers and with 2.0 we're excited to move towards this. Well what we're hoping is it'll be become a lot stickier because they can get to that magic design moment a lot faster. Like it's not 20 minutes, it's two. So right now because it takes you know, 10 to 20 minutes, I think a lot of folks drop off before they get to that point.

01:17:42 - Scot Wingo
Can you see it in the data we or is it more customer Conversations.

01:17:48 - Tina Tang
Conversations is where I'm getting it from.

01:17:50 - Scot Wingo
Yeah, data's good, but conversations are best.

01:17:54 - Tina Tang
Yeah. And then I see it in our. Like what? Like, the negative reviews we get are mostly positive, but we get negative reviews on sometimes the tediousness of the workflow. So it kind of. What's the word I'm looking for? It supports what I'm hearing from the interviews.

01:18:15 - Scot Wingo
Yeah. I came up with a phrase for this. So a pessimistic way to say what you just said is, you know, you could say, look, I took you from using Photoshop in hours to 20 minutes. Like, give me a break. What more do you want? But I like that you're taking the more. You know, And I call that there. There's a thing with customers that want convenience, and it's zero friction addiction. Once you reduce the customer's friction, then they want more friction reduction and they become very addicted to that. You can never go back. And then if you keep getting it down, then they become very addicted to that, that low friction. Because what's also happening. This is inside of your solution. What's also happening? Like, let's say they dip into Photoshop or they try PowerPoint or Instagram Reels. They're going to be like, whoa, that really. That's going to amplify their dissatisfaction and the inability of those tools to do what bristles, hopefully 2.0 is doing so. So it's an important sticky part of this. That's really interesting. And so it's an interesting psychological thing that, you know, I never studied that at all. But, like, I actually spent a fair amount of time thinking about this because it's kind of like our first principles. You know, Elon's first principles are like gravity and what happens to aluminum when you shoot it to space and stuff. Our. Since. Since you and I deal with consumers, there are. Understanding their behavior is the first principle that we have to kind of like, interact with and is really fascinating to me, like, how people behave and why they do things.

01:19:39 - Tina Tang
Yeah. It's also something that fascinates me and Anthony. So we both, from our consulting backgrounds, built software for businesses and we're pretty excited to now be at the consumer prosumer level. It's a lot. It's very interesting.

01:19:57 - Scot Wingo
Yeah. It's harder to learn more because businesses are so rational. They're just like ROI driven.

01:20:02 - Tina Tang
Yeah.

01:20:03 - Scot Wingo
And they're very easy to get information out of consumers. You have to like, really, really kind of read the tea leaves and understand. And then sometimes they'll tell you that, like, A negative review. They will, they will give you that feedback.

01:20:12 - Tina Tang
Yeah. Yes, yes. I think we're lucky to have vocal customers.

01:20:17 - Scot Wingo
Yeah. That means they're passionate. Right. Like, that's a part of persuasion is if you can get the customer to just do something. Like, it's hard to do that. Right. So people have an entropy of not doing things. And if you can, if they get so excited by what you're doing either way that they'll use it and give you feedback or they'll leave you a negative comment that that's actually progress. So that kind of brings me probably to my last question. So I meet a lot of people that get into product market fit and they kind of tap out because it's hard. And, you know, every journey takes, you know, you don't know the end point, which is also, you can. I feel like you can see it coming. Like, you know, this bristle too. You kind of like can see the little light in the tunnel. Hopefully it's not a train. Sometimes I've done that and it's the train. Uh, but, you know, you've been at this three, maybe four years. If we kind of stretch it out. What, why haven't you given up? Like, what, what keeps Tina going every day? You could always go back to Accenture and make 2,300k or whatever it is.

01:21:17 - Tina Tang
I.

01:21:17 - Scot Wingo
Life would be so much easier.

01:21:20 - Tina Tang
I find the problem solving aspect of entrepreneurship really fulfilling. And Anthony does as well. And we, so we both, you know, come from working in corporate and we kind of decided it wasn't for us. It just, I think, like, boring is too simplistic, but it wasn't fulfilling. We want to be kind of on the cutting edge of problem solving, but we also want to own solving the problem. And so entrepreneurship is a really good path for that. And it's kind of what we want to do. So no matter kind of what's next for us is other startup stuff. If, you know, bristles will work out, but if it doesn't, we don't plan to go back to corporate.

01:22:08 - Scot Wingo
You're ruined. Yes. This is the hard. This is the tough thing about it. It's very hard to go back into corporate life after having this type of an experience.

01:22:16 - Tina Tang
It's just very fulfilling. Like, I don't want to do anything else.

01:22:21 - Scot Wingo
Say more like, what fulfills you about it?

01:22:24 - Tina Tang
Well, I get a lot of fulfillment from learning about a problem and then seeing kind of the person's reaction when it's solved. So when I would be in these customer discovery interviews, the 200 that I mentioned, I would see the reaction when, you know, when they painted the piece of furniture and, you know, saw it change and it was realistic, and they were like, whoa, I did that in a minute. And that part of it, it's really fulfilling. And I also like the challenge of. I almost like the uncertainty of. And having to figure out how to solve little pieces. I enjoy breaking down big problems into little bits and solving it bit by bit.

01:23:16 - Scot Wingo
Some people get a lot of anxiety about not knowing where the next paycheck has come. And you've had several times where you're, like, right on the hairy edge of that. Like, how. How have you. You don't. You're very good at hiding the stress. You've never come to me and been, like, stressed out. You're always just like, calm is water.

01:23:33 - Tina Tang
Well, I think when I made the decision to go back to grad school, school I had already committed to. I was going to get my master's. So I had saved up some money for kind of that master's program. And so in my mind, because I didn't have to pay that tuition, that was kind of my, like.

01:23:53 - Scot Wingo
So you've got a little bit of safety net, so you're not totally out there. You're not going to be kicked out of your house or something if you can kind of recover. So, awesome. Well, thanks. So we'll have to have you back in a couple of years to see how bristles 2.0 goes.

01:24:05 - Tina Tang
Yeah.

01:24:06 - Scot Wingo
Get a talk a little bit more about pricing, because by then I'm sure you'll have a lot of interesting stories. And, you know, I'm excited to watch you solve this atomic problem and then, like, start kind of walking up the stack. I think you'll have the benefit of having done this. So you'll be able to, like, you know, be like Pac man chewing through product market fit on each of those things. But also you're going to have this underlying tool set on the technology side that's going to make it easier to kind of go do that stuff. So I feel like you're winding the springs tighter and tighter, and then there's going to be a point where that unleashes, and this is kind of like what it feels like. And then you can really start going very quickly. So we'll have to have you back to hear how that goes.

01:24:40 - Tina Tang
Yeah, I'm excited for that. I think one thing we're really excited about is how this generalizes after we prove it out in furniture. So once we get folks like really, really loving our product. And we see, you know, that reduction in churn, the increase in retention. Then we like this solution applies to kitchen cabinets, bathroom vanities, and even outside of home, like product customization at large. So like customizing cars or something like that. So it's a big space.

01:25:19 - Scot Wingo
Yeah. When you're in product market fit, you have nothing but what I would call low class problems. Like churn, all hard ones. And then like when you're through it, then you suddenly have, there's still problems, but they're high class problems. Like, oh, wow, we've got this, you know, this very target rich environment. We can go explore where to start. And it's like, it feels good to have those kinds of problems. So we'll look forward to that day.

01:25:39 - Tina Tang
Yeah.

01:25:40 - Scot Wingo
Well, thanks for coming in and being generous with your time and we'll catch up with you in a couple years.

01:25:46 - Tina Tang
Sounds good. It was really fun. Thanks for all your support and mentorship.

01:25:50 - Scot Wingo
Absolutely. Yeah. It's been fun watching you bust through all the walls. Thanks for coming.

01:25:54 - Tina Tang
Thanks, Scott. For more Tweener content, check out the Triangle Tweener time substack@tweener.substack.com for more tweener content, check out tweenertimes.com thanks for listening and we'll see you again soon on Triangle Tweener talks.