Arto Minasyan: 13:31
Yeah. Is there? First, we had an internal debate. What kind of features we want to build. Right. And we have like two groups of features. One, we called it an AI speech coach, which would tell users how to improve speech, the intonation, the accent, give recommendations and so on. And the other set of features, we call it assistant features, which will basically do the work for the user. So let’s say taking, recording, taking, transcription, taking, summarization and so on, or just transforming or removing the accent.
So basically we decided the ROI from user perspective, return on investment from user perspective is higher with the assistant features rather than coach, which just gives like recommendations. So we started from transcription.
Dr. Jeremy Weisz: 14:29
Want to be given a fish not the fish. Right?
Arto Minasyan: 14:32
Exactly, exactly. So we started from transcription, then added recording, then like this workspace which we group all the meetings across the company like AI chat there a lot of stuff. And then we brought accent conversion. And I agree this is mainly for the call centers, but like I can give a small spoiler. Here we are bringing another type of accent conversion which will allow people to change accents from which other users would have on the call.
Let’s say now we are having a call, you would be able to change my accent and I would sound neutral American for you with my voice.
Dr. Jeremy Weisz: 15:15
Can you change to other accents? Like what if I wanted to sound smarter, like with a British accent?
Arto Minasyan: 15:20
Yeah, we can do that. But now we are mainly focused on changing like Chinese, Indian, Russian, French and German accent and Latin accent to neutral American.
Dr. Jeremy Weisz: 15:34
This is super interesting. How hard is that to do though? I mean, it just seems like a very interesting, not easy problem. Where do the engineers start when you’re like, okay, we want to basically have people just erase their accent from a video call.
Arto Minasyan: 15:54
Yeah. It’s like very hard technology for several reasons. First, it’s a human voice. So when you do like a manipulation of the voice, it’s very easy to basically spoil, sorry, make the voice sound robotic, right? Basically make it not working.
And even like a small robotic voice would make people very uncomfortable and it would just delete the product. So this is like one big requirement. Second is the latency. So real time is important. So think about if like a solution which adds up latency and I sound not very natural for you, it would be also a big problem.
And then like resource consumption, it should be like very resource efficient in order to run on device and also the data. So all AI models require data. So you need like data of the same person basically speaking in Indian English and then American English. And it’s not easy to generate or find that kind of data in the internet.
So actually it was a super hard problem. Our team probably worked for three years before, before we got to the point that it’s at the quality that can be deployed for thousands or tens of thousand agents across the companies.
Dr. Jeremy Weisz: 17:20
Early on, I think eBay played a role in this journey. Early on in your journey when you were first starting this company?
Arto Minasyan: 17:33
No, I think it was Discord.
Dr. Jeremy Weisz: 17:38
I mean, when you first started it, weren’t you looking for a position at eBay?
Arto Minasyan: 17:43
Not really.
Dr. Jeremy Weisz: 17:44
Oh, okay. I thought when you moved to San Francisco, you. There was an interesting story of why, like when you first started it. Like you sold it to someone first before creating it.
Arto Minasyan: 17:56
Yeah, actually, my co-founder, Davit used to work for Twilio. And when we created the technology, the first thing that we decided to go and pitch this technology to Twilio. So they didn’t buy it back then. But they decided to invest in the company. And then after like two years after the investment, they decided to buy the technology.
So yeah, I think David being an early employee at Twilio and having all the insight in the industry really helped us to shape the company.
Dr. Jeremy Weisz: 18:36
Love it. And, you know, talk about how our, you know, getting that first call center client. I mean, you could have a like, we have this hypothesis that call centers are going to want this, but them actually buying it. What was the first milestone of a call center actually saying we’re going to we’re going to implement this?
Arto Minasyan: 18:57
It’s all happened organically. So when COVID hit, we started to get a lot of inbound leads and I was like 50% were call centers. So apparently we saw that there is a, and we didn’t have any prior knowledge or understanding of call center market. But there was like some strong product market fit and we started to see market pull. So we hired our first sales people at the beginning of 2020 and the first deal, I could remember the date. $100,000 or something deal.
We were like, super surprised to see like a company can buy Krisp for 4000 people and deployed at the scale. Right? And then we started to explore the market. We hired like advisors, people who understand the call center industry and so on. And now we are scaled to 250,000 or more agents in the industry.
Dr. Jeremy Weisz: 19:58
How was the raising money journey?
Arto Minasyan: 20:04
At the beginning it was super hard because we just had a technology and we didn’t have like any traction or revenue or any other like important achievements.
Dr. Jeremy Weisz: 20:14
Where were you living at the time?
Arto Minasyan: 20:16
We were part of the Berkeley Skydeck acceleration program. So we moved to the Bay area for the acceleration program and then after the Demo Day, Sierra Ventures gave it to us. It was our seed round, and then we did like a seed extension in 2019. And other folks also joined like before. We got real traction in 2020, it was super hard for us to pitch and raise the money. I think we met 150 plus investment firms back in 2018, before Sierra decided to invest in the company.
Dr. Jeremy Weisz: 20:53
What were some of the feedback you were getting of why people were not going to invest?
Arto Minasyan: 20:59
That’s a great question. Like features versus platform. So everyone said, okay guys, you have a cool technology, but at some point big guys will build like something similar.
Dr. Jeremy Weisz: 21:10
They’re worried about like Zoom’s going to go in and they can create some kind of Zoom or some noise cancellation or something like that.
Arto Minasyan: 21:16
Or even worse, like MacOS or Windows will bring it at the system level and give to it everyone right every other like product. And it happened actually. They were right. Like one thing which differentiates like startups is resilience. So how fast and how smart and how efficient you can bring new features, new technologies to the product to make it really valuable for the users and have the stickiness. Otherwise, like every other solution could be replaced at some point.
Dr. Jeremy Weisz: 21:52
Where do you see voice AI going now? I mean, you’ve been even early on, you know, that you spend, you know, and your team spends their waking hours thinking about this stuff.
Arto Minasyan: 22:04
A few things there. I think we should differentiate between like two use cases, like human to human and human to computer. And I think most of the exciting stuff right now is happening in human to computer part where all these voice bots are becoming like a thing. Like starting from call center to personal assistants everywhere. Now you can talk with voice AI bots.
And I think it’s super exciting because it’s like a natural form of communication between the humans. So I shouldn’t be the same with the computer, right? And Krisp is basically in almost every other voice AI bot in the industry, which is super exciting.
Dr. Jeremy Weisz: 22:50
I know you were at Cloud Fest. What were some takeaways from Cloud Fest? Any interesting speakers to talk about from that?
Arto Minasyan: 23:01
So at Cloud Fest we introduced our vibe coding for WordPress solution. I think it’s a very exciting approach, basically bringing AI cogeneration to WordPress and bringing this vibe coding thing to WordPress. We also introduced our API to the market. So it’s not only our platform that provides this service to the customers, but any other third party like hosting company, SaaS company or agency can integrate our solution white label under their brand, inside their solution and provide all this coding experience to their existing customers.
I really believe it’s like two things with coding in WordPress and API are going to level up the WordPress ecosystem and make it more robust, more modern, and more ready for a revolution.
Dr. Jeremy Weisz: 24:03
And this is 10Web.io, which I have on the screen here. What was some of the feedback you got from the conference and from people with 10Web?
Arto Minasyan: 24:13
I think we had like one of the busiest booths. All big hosting companies at C-level reached out to us. We met like a lot of demos. It was like super, super busy. Two days for the team and we got a lot of excitement.
And now is the time to materialize on these two days of hard work. But one thing we realized that with this approach of democratizing website building technology through an API like super powerful and could allow time to really scale and also allow partners to provide a better service to their customers.
Dr. Jeremy Weisz: 24:59
So I’m wondering, you know, you decided to go B2B for this, right?
Arto Minasyan: 25:05
Yep.
Dr. Jeremy Weisz: 25:06
Why B2B as opposed to, you know, other routes?
Arto Minasyan: 25:15
Two things. First, like all the big platforms today are B2C and Shopify, Squarespace, GoDaddy, all hosting companies, right? So everyone tries to get customers on their platform. But if you think about where customers live, it’s like everywhere, right? They live in their CRM system.
They live in their like domain registrar and so on. So why shouldn’t we bring the website creation closer to the customer? So let’s say if a customer buys a domain and at the same point of sales, he also gets access to one of the best website builders. Why shouldn’t it? Right. Or if a customer, let’s say a small restaurant runs on some kind of restaurant, CRM is much natural to generate the website for the customer.
Having all his data, all their older data in the CRM and connect the website to the CRM rather than having two separate platforms. So I think being closer to the customer, where the customer live is like a very essential idea. And nobody did it before for the API. And I think with this approach we can really scale.
I think about if a large SaaS company, which has 100,000 customers, decides to integrate 10Web, that means like 100,000 websites for us, right? And think about how many verticals are out there, how many hosting companies are out there who can use our solution to power their service for the customers.
Dr. Jeremy Weisz: 26:57
So who is a good customer for that? Hosting companies, you mentioned. What other type of businesses?
Arto Minasyan: 27:05
So we have three ICPs, ideal customer profiles. The hosting company would be the number one. Second would be like vertical SaaS company. Think about service titans of the world, which are providing a vertical SaaS solution for some industry.
Dr. Jeremy Weisz: 27:22
I think I saw ServiceTitan on Krisp actually on the website somewhere.
Arto Minasyan: 27:25
Yeah, I’ve got a customer of ours.
Dr. Jeremy Weisz: 27:27
Yeah.
Arto Minasyan: 27:30
And lastly, distributors and managed service providers. So. MSPs are there like MSPs which are focused on specific industries or specific cities. And they could become like kind of building platform for their niche and for their industry.
Dr. Jeremy Weisz: 27:52
Could it be like a line service for them? Another line of service.
Arto Minasyan: 27:55
Exactly. Yeah. More like it’s an opportunity for them to become SaaS product, right? They can not only provide service but also automatically generates websites. Sides bring all the features of time web to their customers under their brand.
Dr. Jeremy Weisz: 28:16
How does this happen, so I’m looking here at 10Web white label solutions. How does that work?
Arto Minasyan: 28:22
It’s an API. So our customers basically does an API call and we set up an isolated WordPress environment. And the front end, in that like WordPress, we provide this white coding solution for WordPress, which is like a kind of plugin. And they can like to build. I mean, do thousands of API calls and build a lot of websites. For the resellers and MSPs, basically managed service providers, they just upload their branding into 10Web and they can launch a copy of 10Web in like minutes. So it’s much, much easier integration.
Dr. Jeremy Weisz: 29:10
This is really cool. On the funding side, when you were raising money, you raised separately for each of those companies, right? So what did the investors say to you, if anything? Say, listen. Yeah, I want to give you money for 10Web, you already have this whole other company that you’re running also.
Arto Minasyan: 29:32
Yeah. It’s a very common feedback that I’m getting. And I usually advise like younger or first time entrepreneurs never do this. I like to emphasize how bad is the idea to start like two companies in parallel. I call it like stupid idea or the most stupid thing that I did in my lifetime.
Yeah, but it’s a matter of execution. If you can show that you can build stuff, you can bring customers, you can make them happy and make your business grow, then you get also attention and eventually funding from the investors.
Dr. Jeremy Weisz: 30:10
I do as I say, not as I do type of type of thing.
Arto Minasyan: 30:15
Yeah. And like, you know, like whereas like some random guy who is running five public companies and he’s the richest person on this planet. So it’s realistic not for all of us, but like it’s a matter of execution.
Dr. Jeremy Weisz: 30:30
What’s the most challenging part about running two companies?
Arto Minasyan: 30:36
Yeah, one thing I didn’t realize when I started this company is that the business is about business. So when the business class business requires the founder’s attention more and more, right? So now, like, Krisp is like a big company. They have like 4500 clients. We have a big team. We have new technologies, new products ongoing and so on.
And basically you don’t as a founder, you basically do time dilution and defocus. And sometimes you lose the priority, which absolutely could be like a disruption for the business. So company at the time, that’s the right approach for sure.
Dr. Jeremy Weisz: 31:25
What kind of leadership team did you have to put in place for either the companies?
Arto Minasyan: 31:32
First of all, my co-founders are amazing in both companies, so I would say I split my time 80% Krisp 20% 10Web, and in 10Web. I have like three co-founders who are like one technology guy, the other product, the other marketing. So it allows me basically just to be the strategy guy and the fundraiser for the company. And then we also have the right suite of the executives who are more like entrepreneurs rather than the classic executives. Right?
So we are very hands on. We are into the details. We are like builders. We can create stuff for themselves, very independent, and we put like goals together and they just go and accomplish things. So this kind of like mindset is essential for any startup for the success.
Dr. Jeremy Weisz: 32:31
You know, it’s really amazing what you’ve created here. I have one last question, and I want to encourage people to check out Krisp.AI and you can check out 10Web.io to learn more.
My last question is, I’m just, your favorite tools and software in the AI space. It could be apps, it could be, you know, software, whatever it is. Obviously we have the ChatGPTs, the Gemini’s. But beyond those, what type of apps or software? Maybe you don’t even use them, but you heard really cool things about them. I’m curious what’s on your radar as far as the AI apps and software?
Arto Minasyan: 33:09
Like, we use a lot of tools inside the companies, obviously, like Claude Code, Cursor for the coding. For the marketing, we use like ElevenLabs for text to speech generating AI content and for the sales we use like Gong.io and other solutions. So a lot of stuff to which can bring productivity to the teams. And I believe there is the real potential for the AI.
Dr. Jeremy Weisz: 33:47
Why did you choose Gong? I’m curious out of the other platforms out there.
Arto Minasyan: 33:54
I would assume they are the most specialized one for the sales. They call them Revenue Intelligence Platform, and they basically deliver on their message. Right. We started using Gong probably 2 or 3 years ago, and we’re pretty happy with the quality of the insights that the platform provides.
Dr. Jeremy Weisz: 34:17
Love it. Yeah. I’ve been using things like Wispr Flow a lot. I don’t know if you’ve tested that or any of those other desktop apps.
Arto Minasyan: 34:25
No, I didn’t test them out.
Dr. Jeremy Weisz: 34:27
Yeah. Well first of all, thank you, Arto. Check out Krisp.ai, check out 10Web.io, and we’ll see you next time. Arto, thanks so much.
Arto Minasyan: 34:38
Thank you. Jeremy.
