Founder Interview
How Boltzbit Landed Its First $150K Annual Contract with Generative AI Search Technology (Interview with CEO Yichuan Zhang)
- Interview Date
- April 27, 2022
- Interviewee
- Yichuan ZhangCEO and Co-Founder
Company Metrics at Interview Time
Annual Revenue (2022)
$150K
Customers (2022)
1
Team Size (2022)
9
Seed Funding Raised (2021)
$1.6M
Pre-Seed Funding Raised (2020)
$800K
Historical Snapshot
These numbers were reported by Yichuan Zhang during his interview with Nathan Latka in April 2022 and are a historical snapshot, not current figures. See Boltzbit Limited’s current numbers.
Key Takeaways
- 01Boltzbit signed its first customer on a $150K annual contract in 2022
- 02The company was founded in April 2020 when the first line of code was written
- 03Boltzbit has 9 full-time team members including 3 co-founders, 4 engineers, 1 marketing person, and 1 business person
- 04The company raised an $800K pre-seed round in 2020 and a $1.6M seed round in 2021
- 05Yichuan Zhang holds a majority equity stake among the three co-founders
- 06The first use case is generative AI-powered search across mixed document types including text and images
- 07Boltzbit is exploring fintech and digital marketing as its next target verticals
- 08The platform uses a pay-as-you-use cloud SaaS model based on data uploaded and compute hours
- 09Public datasets such as COCO and ImageNet were used to train the initial generative AI models
- 10The company is actively hiring but comfortable with current development capacity until the next customer is onboarded
Company Metrics at Time of Interview
| Metric | Value | Source |
|---|---|---|
| Annual Revenue (2022) | $150K | Founder interview, April 2022 |
| Average Contract Value (2022) | $150K | Founder interview, April 2022 |
| Customers (2022) | 1 | Founder interview, April 2022 |
| Team Size (2022) | 9 | Founder interview, April 2022 |
| Engineers (2022) | 4 | Founder interview, April 2022 |
| Co-Founders | 3 | Founder interview, April 2022 |
| Pre-Seed Funding (2020) | $800K | Founder interview, April 2022 |
| Seed Funding (2021) | $1.6M | Founder interview, April 2022 |
| Year Founded | 2020 | Founder interview, April 2022 |
Growth Breakdown
Revenue
Boltzbit reported $150K in annual revenue in 2022, derived entirely from its first customer on a single annual contract. The company's intended pricing model is pay-as-you-use, based on data uploaded and computational hours consumed on the platform.
Customers
The company had one paying customer at the time of the interview. Yichuan Zhang noted the team was actively exploring fintech and digital marketing verticals to land the next customer, relying on partnerships and co-sell arrangements.
Team
Boltzbit had 9 full-time employees in April 2022, comprising 3 co-founders, 4 engineers, 1 marketing person, and 1 business development person. The team was still hiring but planned to accelerate headcount growth once the next customer was onboarded.
Funding
Boltzbit raised an $800K pre-seed round in 2020 and a $1.6M seed round in 2021, bringing total disclosed funding to $2.4M. Zhang described the valuation as fair given the company's deep tech nature and early stage.
Growth Strategy
Founder Network and Domain Expertise
Boltzbit landed its first customer through personal connections and deep knowledge of a specific niche problem. Zhang emphasized that understanding the problem well was key to winning that initial deal.
Generative AI on Public Datasets
Rather than purchasing proprietary training data, Boltzbit used publicly available datasets such as COCO and ImageNet to train its generative AI models. This approach reduced upfront data costs and leveraged the self-learning properties of generative AI.
Vertical Expansion Strategy
With one customer in search technology, the team is using that early success to identify adjacent verticals. Fintech and digital marketing are the two verticals Zhang named as active areas of exploration.
Partnership and Co-Sell Model
To enter new verticals, Boltzbit is pursuing a combination of direct sales, technology partnerships, and co-sell arrangements. Zhang described this multi-channel approach as the primary go-to-market motion for customer number two and beyond.
Cloud SaaS Pay-As-You-Use Pricing
The company's core business model is consumption-based, charging customers based on how much data they upload and how many computational hours they use. This model is designed to lower the barrier to entry for businesses wanting to adopt AI without deep machine learning expertise.
Best Quotes
“So, it means to really give people who have the data, the power of AI, so people can just develop AI from their data without having knowledge of machine learning or deep learning.”
“So now we have the first customer, it is like 150 ks for the annual contract, right? But our main, our main business model is kind of like pay as you use, it's kind of cloud based SaaS. Model is basically based on how much data you upload on our platform and you train the model and then you pay as the computational hours.”
“So we found it very likely by our connections and it is also something like very specific niche vertical that is we know the problem pretty well, Right? So this is why we still like really try to use this early success to figure out, to discover more markets, more similar customers.”
“Yeah, it is for search engine technology. So basically use Generative AI to automatically parse different types of documents, not only text image can be mixed and then combine those information together, allow user to interactively search.”
“Initial dataset, this is a good thing of generative AI because it can learn to find the data itself. So we use a lot of public data set and, you know, a lot of generative AI trained on very public data set.”
“I think that is kind of fair valuation because we are quite deep tech and we are in the very early stage, right? So I would say that's kind of fair, but of course I would like to have the valuation even higher, but these kind of things also, you know, it's always if you put it too high and later you get yourself into problem also.”
“We are we are still actively hiring, but roughly, we are we are at a stage where are quite happy with our development power. When we get on board with the next customers and we are potentially start to get the next customer sent in, we will start hiring again.”
“Something I wish I knew when I was 20. I think I wish I knew business more. I think back to 20, I'm a nerd. I only study scientific stuff.”
What Happened Next
This interview captured Boltzbit at a very early stage in April 2022, when the company had just signed its first customer and was exploring new verticals. The numbers here reflect what Yichuan Zhang reported at that moment and should be treated as a historical snapshot. Visit the Boltzbit company profile on GetLatka for the most current revenue, customer, and funding data available.
View Boltzbit Limited’s current profile and metricsFull Transcript
Chapters
- 0:00Introduction and Background
- 0:23What Boltzbit Does and Who It Sells To
- 1:04Pricing Model and First Customer
- 1:38Founding Story and Team
- 3:02Equity Split Among Co-Founders
- 3:56Funding History: Pre-Seed and Seed Rounds
- 6:35Valuation Discussion
- 7:10How Boltzbit Landed Its First Customer
- 7:42First Use Case: Generative AI Search
- 8:19Training Data and AI Model Development
- 9:08Next Customer and Vertical Expansion
- 9:56Team Growth and Hiring Plans
- 10:19Famous Five: Books, CEOs, and Tools
- 11:00Personal Life and Lessons Learned
Introduction and Background
Nathan Latka
00:00Doctor Yichuan Zhang has been passionate about AI since the start of his PhD at the University of Edinburgh back in 2010. Since the first year of his PhD, he started to publish several papers on Boltzmann machines and approximate inference in one of the most influential AI conferences, NeurIPS. His research contributes to the core of his current company, Boltzbit, at boltzbit.com. Doctor Zhang, are you ready to take us to the top?
Yichuan Zhang
00:22>> Yes.
What Boltzbit Does and Who It Sells To
Nathan Latka
00:23Okay. So Boltzbit's tagline is make AI accessible to everyone. What does that mean?
Yichuan Zhang
00:29>> So, it means to really give people who have the data, the power of AI, so people can just develop AI from their data without having knowledge of machine learning or deep learning.
Nathan Latka
00:42And who are these people? I mean, who are you selling to?
Yichuan Zhang
00:45>> Selling to is business. Yes, mainly it's business and particularly for a lot of companies who have a lot of customer data and product data and they want to make automated decisions like product recommendations and user search, answer a lot of decision questions. This is
Pricing Model and First Customer
Nathan Latka
01:04And what are are these companies paying on average per month or per year to use your technology?
Yichuan Zhang
01:10>> So now we have the first customer, it is like 150 ks for the annual contract, right? But our main, our main business model is kind of like pay as you use, it's kind of cloud based SaaS. Model is basically based on how much data you upload on our platform and you train the model and then you pay as the computational hours.
Nathan Latka
01:35And so you said one customer today?
Founding Story and Team
Yichuan Zhang
01:38>> Yes. One now we have our first customer. Yeah.
Nathan Latka
01:42Well, congratulations. That's very exciting. When did you write the first line of code for your platform?
Yichuan Zhang
01:48>> That's back two thousand twenty April, I think. Yeah.
Nathan Latka
01:532020. And are you the sole co founder or do you have a team?
Yichuan Zhang
01:59>> Now we have a we have team. Yeah. So we have three co founders and a plus now four developers. Yeah.
Nathan Latka
02:07Okay. So there's seven people.
Yichuan Zhang
02:09>> And one marketing person and one business person.
Nathan Latka
02:12Yeah. There's nine people full time? Yep. Okay. And when you said three co founders, did you guys decide to split equity evenly at the beginning or not?
Yichuan Zhang
02:22>> No, it's not evenly, like kind of I take some majority because I lead on the initial ideas and the others are my friend. Yeah.
Nathan Latka
02:32So you you own the most out of the three co founders?
Yichuan Zhang
02:35>> Yep.
Nathan Latka
02:36Okay, very cool. Did you put a bunch of your own money into the business to build the MVP or no?
Yichuan Zhang
02:44>> I would say kind of yes, because we spend quite some initial time with our salary and work on some prototypes before we raise the first round of money.
Nathan Latka
02:54Yeah. How many months or how many years?
Yichuan Zhang
02:57>> That's roughly like four to five months, something like Yeah. That, I would
Equity Split Among Co-Founders
Nathan Latka
03:02And how much did you end up raising?
Yichuan Zhang
03:06>> Our first raise is 800 k. And then last year, we raised $1,600,000.
Nathan Latka
03:11What year did you raise the 800 k in?
Yichuan Zhang
03:14>> That is 2020. Yeah.
Nathan Latka
03:17Okay. 2020. So the 800,000. And was that that's sort of your traditional pre seed round?
Yichuan Zhang
03:22>> Yes. That's the pre seed round.
Nathan Latka
03:24And you said the run last year was for 1,600,000?
Yichuan Zhang
03:28>> Yes.
Nathan Latka
03:291,600,000. Okay. And how did you was that a did you come up with a valuation for that or was that a convertible note with no cap?
Yichuan Zhang
03:37>> No. That is a VC investment. Is yeah. It is not loan. Yes.
Nathan Latka
03:44Most people, when they're raising their seed round, are selling between, you know, 15 to 20% of their business. Is that what you did?
Yichuan Zhang
03:54>> Yeah. Roughly. Yeah.
Funding History: Pre-Seed and Seed Rounds
Nathan Latka
03:56Okay. So that means you raised it like a 10 or $12,000,000 valuation, something like that?
Yichuan Zhang
04:03>> Yeah.
Nathan Latka
04:05Oh, what's going on there, YouTube? Good to see you guys. Now imagine this. You love watching these interviews with SaaS founders, but imagine if we took all of the valuation data out from over 2,807 interviews I've done manually. Saves you a lot of time. Well, we've done this. We've built it into the beautiful interface inside of Founderpath. Check this out. I'll show you how you can access this in a second, but you log in, you connect
04:28your Stripe account, you see your valuation real time. You can see what it changed over the past eighty eight days and even set goals for valuation this year. Now the secret evaluation is there's many different ways to value a SaaS business. So the reason you're gonna see three or four different valuations inside of your Founderpath dashboard, this is all free by the way, is because depending on who's doing the buying of your SaaS company, you're gonna
04:52get a different valuation. A VC is gonna pay a different valuation. Private equity firm is different. If you're gonna do a minority sale, that's different. And if you sell the whole business, that's a different valuation. You can see all those when I hover over here.
Yichuan Zhang
05:05>> Right? So
Nathan Latka
05:06the teal is what a VC would pay. Yellow is what private equity And red is if you sold the whole thing outright. Now what's cool about this is this is not built off random data. Again, you guys hear these interviews on YouTube. All these datas are built from real time valuation data points founder share with us on the show. So traction 1,200,000 seed round 3.7 raise. They sold 22% of their business. Go in here and filter
05:30by the event. Maybe you only wanna see companies that have sold the whole business. Well, here are a bunch that have been acquired the valuation and the multiple. Maybe you're going out right now and you're raising your seed round. We'll go in here and look at all this recent seed deals that went down, what they raised, what valuation they raised at and what percent that they sold. There's never been a larger dataset of SaaS valuations than
05:55what you can get now inside of Founderpath. And we're thrilled to bring it to you. All right, we're gonna go back to the YouTube video here in a second, but if you wanna check this tool out, if you wanna jump in and sign up, you can check it out for free to get your valuation at this link. This link, founderpath.com/products/valuations. Or if you go to founderpath.com and hover over products, click on get your valuation here, and
06:20go ahead and sign up to give it a whirl. Again, all that valuation data live right inside the platform. I hope to see you there. Alright. Let's jump back into the interview. Okay. Was that fair? Do you think that was a good valuation or bad valuation?
Valuation Discussion
Yichuan Zhang
06:35>> I think that is kind of fair valuation because we are quite deep tech and we are in the very early stage, right? So I would say that's kind of fair, but of course I would like to have the valuation even higher, but these kind of things also, you know, it's always if you put it too high and later you get yourself into problem also. So we try to balance this in the best way.
Nathan Latka
07:00Yeah. So monthly recurring revenue today, it sounds like it's about $12,500 or about $150,000 a year from one customer?
Yichuan Zhang
07:09>> Yeah, roughly like that.
How Boltzbit Landed Its First Customer
Nathan Latka
07:10Yeah. How did you land the first customer? That's a very big first deal to land. How you find the customer?
Yichuan Zhang
07:17>> So we found it very likely by our connections and it is also something like very specific niche vertical that is we know the problem pretty well, Right? So this is why we still like really try to use this early success to figure out, to discover more markets, more similar customers.
Nathan Latka
07:39What is that use case? This first use case you're building?
First Use Case: Generative AI Search
Yichuan Zhang
07:42>> Yeah, it is for search engine technology. So basically use Generative AI to automatically parse different types of documents, not only text image can be mixed and then combine those information together, allow user to interactively search. So you're not just giving an image, you can even highlight which part of the image that is using this most relevant. And this is something special we developed.
Nathan Latka
08:08Did you have to buy a very large data set to feed your initial dataset to grow your algorithms, your AI, etcetera? Where did you get the initial dataset to train your models on?
Training Data and AI Model Development
Yichuan Zhang
08:19>> Initial dataset, this is a good thing of generative AI because it can learn to find the data itself. So we use a lot of public data set and, you know, a lot of generative AI trained on very public data set.
Nathan Latka
08:31Like what? Can you name a few of those data sets?
Yichuan Zhang
08:33>> Like one of these, like COCO and ImageNet, a lot of public data that you can find on GitHub.
Nathan Latka
08:40Interesting. Very cool. Okay. So nine people full time on the team today. Are you raising capital right now?
Yichuan Zhang
08:48>> No. Not really. Yeah.
Nathan Latka
08:50Okay. And so what's the next you just closed last month. Okay. What's so what's the next step? You have one customer today. Where do you think you'll get your next customer from?
Yichuan Zhang
09:01>> We are exploring different verticals now for the next customer. We
Next Customer and Vertical Expansion
Yichuan Zhang
09:08>> don't know that exactly yet, but we are actively searching on that.
Nathan Latka
09:14Which verticals are you exploring or is it all search?
Yichuan Zhang
09:18>> No, we are trying to explore different vertical. I like financial, fintech is one, and also digital marketing is another one. Mhmm.
Nathan Latka
09:29And and what is exploring those markets look like? Are you looking for partners in those spaces?
Yichuan Zhang
09:34>> Yeah. Both both partnership, but also kind of like co sell and also tech partner in all forms.
Nathan Latka
09:46Very cool. All right. This is great. In terms of team growth, are you hiring right now, or you're good with nine people?
Team Growth and Hiring Plans
Yichuan Zhang
09:56>> We are we are still actively hiring, but roughly, we are we are at a stage where are quite happy with our development power. When we get on board with the next customers and we are potentially start to get the next customer sent in, we will start hiring again.
Nathan Latka
10:12Very good, doctor. Okay, let's wrap up here with the famous five. Number one, what's your favorite business book?
Famous Five: Books, CEOs, and Tools
Yichuan Zhang
10:19>> My favorite business book, I'd probably say The Mom Test. Yeah.
10:24>> The Mom Test.
Nathan Latka
10:25Number two, is there a CEO you're following or studying?
Yichuan Zhang
10:29>> Well, that's hard to say.
10:33>> I mean, Elon Musk, I think I followed quite quite often. I cannot say I'm really he's the only one, but he's on the news a lot. Yeah.
Nathan Latka
10:42Number three. What's your favorite online tool for building Boltzbit?
Yichuan Zhang
10:48>> Online tools for building Boltzbit? I
10:54>> probably see GitHub. Anything open source that is very useful.
10:59>> Yeah.
Personal Life and Lessons Learned
Nathan Latka
11:00Number four, how many hours of sleep do you get every night?
Yichuan Zhang
11:04>> I would say it varies. On average, I think maybe six hours. Yeah. Okay. But sometimes very extreme, it's three.
Nathan Latka
11:12And what's your situation? Married, single, kids?
Yichuan Zhang
11:16>> I'm married.
Nathan Latka
11:17And any kids?
Yichuan Zhang
11:19>> No. I don't have a kid.
Nathan Latka
11:21No kids. And doctor, how old are you?
Yichuan Zhang
11:24>> Now I'm I'm 34.
11:27>> 34.
Nathan Latka
11:28Last question. What's something you wish you knew when you were 20?
Yichuan Zhang
11:31>> Sorry. Can you repeat the question?
Nathan Latka
11:34Something you wish you knew when you were 20.
Yichuan Zhang
11:38>> Something I wish I knew when I was 20. I think I wish I knew business more. I think back to 20, I'm a nerd. I only study scientific stuff. Yeah.
Nathan Latka
11:49Well, sounds like you're getting into the finances very quickly with your first customer guys, boltzbit.com. Just another first customer, 150,000 a year contract. They also just raised a 1,600,000 seed round at somewhere between a 10 and $15,000,000 valuation. Nine on their team trying to figure out different use cases for this platform they've built deep, deep tech. Right now, this first one is actually effectively a search function that allows you to search images, text, all kinds of
12:13media types to surface whatever it is that you're looking for. So exploring new use cases today in finance and digital marketing, we'll see where they go next. Doctor Yichuan, thanks for taking us to the top.
Yichuan Zhang
12:23>> Thank you.
Nathan Latka
12:26One more thing before you go. We have a brand new show every Thursday at 1PM Central. It's called Shark Tank for SaaS. We call it deal or bust. One founder comes on, three hungry buyers, they try and do a deal live and the founder shares back end dashboards, their expenses, their revenue, ARPU CAC, LTV, you name it, they share it and the buyers try and make a deal live. It is fun to watch every Thursday 1PM
12:51Central. Additionally, remember these recorded founder interviews go live. We release them here on YouTube every day at 2PM Central. To make sure you don't miss any of that, make sure you click the subscribe button below here on YouTube, the big red button and then click the little bell notification to make sure you get notifications when we do go live. I wouldn't want you to miss breaking news in the SaaS world, whether it's an acquisition, a big
13:14fundraise, a big sale, a big profitability statement or something else. I don't want you to miss it. Additionally, if you want to take this conversation deeper and further, we have by far the largest private Slack community for B2B SaaS founders. You want to get in there. We've probably talked about your tool if you're running a company or your firm if you're investing. You can go in there and quickly search and see what people are saying. Sign
13:35up for that at nathanlatka.com/slack. In the meantime, I'm hanging out with you here on YouTube. I'll be in the comments for the next thirty minutes. Feel free to let me know what you thought about this episode and if you enjoyed it, click the thumbs up. We get a lot of haters that are mad at how aggressive I am on these shows, but I do it so that we can all learn. We have to counter those people.
13:55We got to push them away. Click the thumbs up below to counter them and know that I appreciate your guys'support. All right. I'll be in the comments. See you.