随机采访硅谷AI创业者：谁在赚钱？在招什么人？｜Inside Silicon Valley's AI Boom: Who's Actually Making Money? – Transcript

00:00The scale of our new annual revenue? Since we're a public company, we grew over 100% in the first half. What do you do for work currently? AI. I'm the CEO. What kind of people are you hiring? Those with a good intuition for AI. What kind of work do you do? We produce long-form videos and TV media content.

00:17Our recently launched model is first in the US and second globally. What kind of talent do you think will be increasingly popular in the future? Actually, whatever you learn will eventually become outdated. How many people are in your company right now? 50 to 60 people. Our ARR was $15 million last year and we're basically cashflow positive every month. You've been quite successful at the forefront of AI.

00:33What kind of talent do you think we need most right now? In every aspect, go-to-market is extremely, extremely critical. What kind of skills do they generally need? Today in Silicon Valley, I'm attending an AI conference where there will be many practitioners, entrepreneurs, and investors from the AI industry. For example, I'll chat with them about what they're doing now, their views on the future, their perspectives on the industry market, and how to make money.

00:53Hi, miss. Would you mind if I chatted with you briefly? Because I'm waiting for my car and it's almost here. Oh, is that so? Hello. Are you waiting for a car too? No. You're not waiting for a car? Would it be convenient to chat with you for a moment? Oh, no thanks. Okay. Okay.

01:08Hi, miss. Could I have a quick chat with you? But aren't we... You know, I'm a bit... Oh, alright. No problem at all. Okay. Can I have a quick chat with you? Okay. What kind of work do you do? I'm the CEO of a startup.

01:23We produce long-form videos and TV media content. We use AI to create professional-grade products, not short videos. 60 minutes, 90 minutes, 120 minutes. Is that achievable? Technically, it is achievable. The model we recently launched is first in the US and second globally.

01:39Would you like to share what that model is called? Utopiax. What did you do for work before? Before joining this company? I previously worked at JP Morgan Investment Bank. What kind of jobs or talent do you think will be increasingly popular in the next 10 years? Actually, whatever you learn will eventually become outdated.

01:55What's most important is mastering fundamental skills, as these are needed in any industry, in any era. What kind of performance do you value more? The three most important skills, first, is communication ability. No matter how AI develops, humans are still the final decision-makers and managers of everything.

02:11Another is approachability, which is also very important. At the end of the day, it's about whether you can become someone others trust. So your way of speaking and expression needs to earn their trust as quickly as possible. Third, if you are doing professional training to enter the AI industry, you might want to choose skills that lean towards the sciences.

02:30This makes it easier to transition into the AI industry later. Basic subjects like mathematics and physics are likely the most applicable. Thank you for sharing. You are very approachable. Can we chat with you for a bit? Sure. We are currently doing on-device multimodal AI, implementing all model capabilities on your PC.

02:49It's not just language. It includes various modalities, understanding and analyzing these modalities, and then using them to perform further tasks. What did you do before? I have been an entrepreneur and also worked in VC. How many people are in your company and how long have you been doing this?

03:05Our company was founded at the end of 23, so it's been about three years or so. We have five or six people. Our ARR last year was about 15 million USD and we are backed by Sequoia and Hill House in China. So you are already making money? We essentially have positive cashflow every month.

03:20We are based on a mature business model while also working on some more ambitious things with higher difficulty, but we have the ability to sustain this foundation ourselves. Who are your typical users? Over 250 of the Fortune 500 companies use our products and we also target content creators.

03:37They have a lot of videos and use these videos for a lot of their work. It generates videos, audio, images, and more, all based on the understanding of these videos, making it easier for them to execute their next project.

03:52Ultimately, we want to turn all the data we understand into a personal memory layer because for each of us, we have a lot of data on our devices, our PCs, our phones, and not everything is suitable for uploading to the cloud. So how to perform data mining and how to extract value from this data is what we do.

04:09We bring the model to your data. Rather than having you upload all your data to the cloud, we follow a privacy-first principle and it's multimodal. It can process the audio, video, and images right on your device. You're at the forefront of AI and your company has been quite successful.

04:25Looking at it now, what kind of talent do you need the most? Every aspect. Since we are dealing with multimodality, we need people in all areas, whether they are in algorithms or engineering. Besides technical engineers, what else? Of course, go-to-market is extremely crucial because how you promote a good product to the market can vary depending on the product.

04:42Some might be traditional, like Google ad spend, while others require co-creation with creators or leveraging their platforms for promotion. So for go-to-market, we also need people with very talented backgrounds to work together, which is quite difficult. What kind of skills are generally required?

04:58For things like ad spend, because that's a specialized skill, how do you use SEM or SEO to help more people find you through search engines? For instance, our large models. Now that many people use large models to search, how do you use those models so people can better find and understand you?

05:13Also, how do you manage user operations since many of your users can actually become your KOLs who can promote you? Whether it's through existing users, referral loops, or creating viral content, we need people who are very creative. How to manage your social media, how to cooperate with other KOLs, we actually need people from different backgrounds.

05:31Are these people you're hiring in-house? Usually people just find an agency to collaborate with, right? No, many of our roles are in-house because we believe we must first know how to do it ourselves before outsourcing to other teams. Otherwise, you won't know where the money is going and it won't be effective. Do startups like yours hire fresh graduates? We hire fresh graduates, university students, and even interns.

05:48We start training them from the internship stage. The company was founded in 2023, correct? Was it profitable from the very beginning? We broke even in just three months. Why were you able to break even right away? Is there anything we can learn from that? A very good market niche, polishing that specific product line well enough and being willing to spend money early on to experiment with how to promote it.

06:07They started with Go to Market very well and the product is also excellent. Yes, the founder had his previous two companies acquired, so he's a successful entrepreneur. Exactly. So he has some experience that he can replicate here. Thank you for sharing. You look very familiar. Would you mind if I chatted with you for a moment?

06:23How could I look familiar? No, no. I just feel like I've seen you somewhere. Are you an industry professional or an entrepreneur in AI? I am an AI entrepreneur, but I am based in China. What kind of work are you doing right now? We are an AI customer service company and I am the CEO. So as the founder, how long has this company been around?

06:41We've been around for many years, but AI customer service has been our focus for the last three. Previously, we provided Omnichannel software for enterprises. Oh, so you were in the customer service field. How many years did you do the original business? That original business has been running for 20 years.

06:56Transforming after 20 years to do AI customer service now? That original business is still running and it's doing very well in China. Would you mind disclosing the company name? It's called Tian Run. Did you encounter any challenges when transitioning from traditional customer service to AI? It was actually quite smooth. One reason is that our industry faced a severe crisis.

07:12We used to provide efficiency tools for customer service staff, but when AI arrived, intelligent agents replaced those staff, impacting our software. Because the people were gone and we provided tools for people, that was a major crisis.

07:27But pressure creates motivation. When the pressure is high, the motivation is even greater. So how did you manage the transformation? We had to hire completely different types of people. We formed a new team and added a new business line, AI customer service. Do you have any anxieties at the moment? We've transitioned quite well this year.

07:43Our new business revenue scale has already taken off. Would you mind sharing the revenue scale of the new business? Since we are a public company, I can't just throw out numbers. But it's doing okay, especially in the first half of the year. Yes, the first half figures are out.

07:58I don't recall the exact number, but it's quite good. It grew by over 100%. Enterprises have a very strong willingness to undergo AI transformation, and the pace is quite fast. The question is, how we can keep up with this opportunity and ensure we don't fail to capture it due to internal capacity constraints?