“One person can create an AI company with a valuation of US$1 billion.”
OpenAI founder Sam Altman’s unintentional words triggered a new The group has exploded, and this is the new profession that is currently popular - AI independent developer.
In the era of large models, AI-assisted development has greatly lowered the threshold for launching new products. Many people are trying to monetize their ideas and whether they can create a popular smart agent or APP.
Currently on the market, the most popular three-piece package for independent AI developers is: AI search, AI painting, and edge Bot. There are relatively well-known independent developers in each field. For example, in the field of AI search, there is Ai Doubi, the AI search product he made, has 160,000 users around the world; in the field of AI painting, there are many stable diffusion open source models. The most famous one is Wang Dengke, who made his debut with the coaxing simulator. His main business is It is the AI drawing website 6Pen Pro; there are many products in the edge Bot field, and many of them are going overseas, such as CrushON, the 18-ban chat product made by former Byte people.
An independent developer described to Brother Jing, “Nowadays, it is not easy for a one-person company to become famous overnight.” Nowadays, media and distribution channels are very mature. Any product that becomes popular or appears on the App Store download list will soon be noticed by short video bloggers and technology and business writers, and will be out of the circle after reporting on it.
Recently, an App called "Kitten Fill Light" has entered the Top 20 of Apple's iOS application list. It is reported that after he resigned from Dachang, he made nearly 100 yuan in AI in the past year. Ten thousand. Zhao Chunxiang, the author of the "Book of Stomach", has made three AI products by himself, insisting on 0 investment and 0 marketing fees, which is said to be enough to support himself. There is also a 17-year-old independent AI developer abroad who earned US$1 million in 4 months and was liked by Zuckerberg on social platforms. The first batch of AI developers have already achieved both fame and fortune.
Are independent AI developers really making so much money? What are the chances of it becoming a hit? How high is the ceiling for independently developed products? With these questions, Brother Jing will take you to visit these groups.
AI makes it easier and easier to realize dreamsIn the past, independence To develop a product, one prerequisite is that you must be able to program.
However, the emergence of large models and AI coding has removed this last threshold. With the emergence of ChatGPT4, the core interactive capabilities of many AI products are easily satisfied by large models; at the same time, Cursor's AI programming capabilities further lower the threshold for product generation, and products can be generated by typing input requirements.
The kitten fill light is such a product. The developer behind it, Chen Yunfei, in 2In March 2023, I resigned from a major company and started the life of a digital nomad. "Since the emergence of Claude 3.5, AI programming has been implemented. It is not an early adopter method." Chen Yunfei said to Brother Jing, which means that he can make products at any time even if he doesn't know how to code.
The development of the kitten fill light was inspired by my girlfriend. Girls often use foundation pictures to achieve special lighting effects. With this product, you can generate the backlight you want at any time. .
For a long time, Xiaohongshu has shared many such pictures, but many male developers do not understand this need, and no one develops such products. A word from his girlfriend made him aware of this market, so Chen Yunfei used Cursor to develop it. It only took one hour to complete, and it has been downloaded more than 100,000 times since it was launched on the App Store.
The explosive drama also happened to independent AI developer Zhao Chunxiang.
As a liberal arts student, he had previously worked as a product manager in a large factory. Later, he started his own business to make products. He spent 1 million in 6 months and failed. Later, he saw many people sharing records of "eating for one person" in Xiaohongshu. This inspiration inspired him. He wanted to record his daily diet by taking photos, and let AI assist in generating the calories and metabolic pathways of the diet, and analyze his daily diet. Healthy situation.
This product is "Stomach Book", a food recording app based on a large language model. "Book of Stomach" eventually gained 90,000 users, and some investors gave it to TS, which made him a little popular.
Of course, controversy also ensued. Some people mentioned that in fact, a 17-year-old high school student developed a Cali path scanning artifact abroad earlier - Cal AI, and the "Book of Stomach" idea Similarly, this product was an instant hit after its launch, with sales reaching US$1 million in just four months. Zuckerberg also praised this guy for being a bit cool.
Zhao Chunxiang is currently promoting his second AI product - the stranger alarm clock. This product is a service where users spontaneously upload sounds, and other users randomly receive wake-up sounds from strangers. The role of AI is to review a large number of user-uploaded voices to see if there are any phenomena such as marginalization.
Before Chen Yunfei and Zhao Chunxiang, there were also several independent developers who became popular.
Ai Doubi’s most popular product is AI search ThinkAny. As a former Tencent developer, he closed the development for two weeks before this product was launched. Compared with other products, search still has some barriers, and competition is relatively fierce. Both foreign Perplexity and domestic Secret Tower have raised hundreds of millions in financing. Ai Doubi is still doing AI search alone, and has recruited developers who are willing to help Ai Doubi to support it.
This product can break out of the industry because AI search is an emerging product form. The market shares of Perplexity and Secret Tower are still relatively small. Google and Baidu are not eager to use AI to revolutionize their traditional search market. This leaves room for ThinkAny to develop its productsIt has been launched in Japan, Russia and other countries one after another, and acquired new users through SEO.
One person made AI search, which is quite impressive in the eyes of many people. Many people think that search is done by big companies. Capital is also paying more attention to this field, and many technology and business media have interviewed him, which has given him a lot of exposure.
Wang Dengke developed the famous work "Coaxing Girlfriend Simulator" on GPTs. The main development threshold for GPTs is to be proficient in Prompt. This is not a problem for Wang Dengke. The prompt word is the first lesson for many AI enthusiasts after the big models started to become popular.
The difficulty of AI painting websites is not too high, and most of the core algorithms are based on open source models. Shortly after 6pen was launched, Stable diffusion was open sourced, and AI painting websites sprung up on the market, "hundreds of them appeared in a year."
Now it seems that the core capabilities of these websites are the models of different styles contributed by the community. After all, everyone has similar underlying technical capabilities. The models trained by the upper-level user data are the key to determining the quality of users' paintings. .
Betting on the chance of becoming a hitFor Chen Yunfei, the chance of becoming a hit is a matter of luck.
Within one year after leaving his job, he used AI to develop many types of projects, including locally running scripts, websites, applets and iOS applications. He has released 9 products at Station B alone.
He himself concluded: The GPTs "More Diligent and Smarter GPT4" created half a year ago became very popular. It was on the homepage of GPT Store, was interviewed by New Weekly, and was a hot search on Weibo.
At the same time, I made a strategy of implanting prompt hacker techniques in resumes to let resumes pass AI screening. It didn’t become popular, but more than a month later, a user X posted something similar (m.okjike ) exploded.
The day before ChatGPT o1 was released, I made a strategy to implement a thinking chain and use a Chrome plug-in to hide the thinking process, but it was not popular. But two months later, a 17-year-old boy did the same thing using Claude, and it became popular. "You have to make about 10 pieces of junk before you can make an out-of-the-box product," Chen Yunfei said.
The kitten fill light he made more than half a month ago became popular, not through platform recommendation or self-propagation of the product, but because he slowly learned to use self-media for promotion.
Chen Yunfei reviewed how he used traffic to activate this product, which can be said to be the promotion SOP for independent AI developers.
He posted Xiaohongshu to introduce the product with a girl's mindset and a developer's mindset. The former received little response, while in the comment section of the latter's notes for the "AI programming group", some people questioned the value of this application. : "Just go online and download the pictures. Is it really necessary to download an app for this?" Many people think that Apple's reviewers fell asleep and gaveThis is a useless product.
In fact, we are not afraid that a product will be controversial, but we are afraid that no one will pay attention to it. Many people will pay attention to this product through controversy.
The second wave of traffic is the real target users brought by the controversial traffic - girls who love beauty and selfies. Especially after many beauties posted their photos in the comment area, the traffic from the beauties pushed up the downloads of this product.
Compared with Chen Yunfei’s initiative to get out of the circle, now the media will also focus on these cases of explosive wealth. Chen Dengke’s coaxing simulator should be the one who can get out of the circle, it should be Weibo technology V Lan Xi After forwarding the recommendation, many technology businesses paid attention and conducted intensive interviews.
Zhao Chunxiang is also a beneficiary of this model. He mentioned that there are actually many users of Wei Book who are peers. These people come to experience it after seeing media reports. Pure users may only account for half.
Many people hope to get the attention of high-traffic Vs, but just to get the attention of high-traffic Vs, they also need this product to become a basic hit. Judging from Brother Jing’s analysis, after everyone has four or five products, they will have products with certain results.
Ai Doubi has made products such as GPTalk, Cicada Reader, and red envelope cover generator. Some of them are obviously for fun, but ThinkAny is an ambitious product. After his hard work, it became a global product with 600,000 monthly visits in 3 months.
The coax simulator is in the QQ ecosystem. After being played by thousands of students, everyone found it very interesting. Commercial media began to report intensively. At that time, everyone thought that GPTs would be like the era of independent website building, and promote An exploding industry. It seems that it is indeed too early to say now. The intelligent agent ecosystem still needs the continuous progress of the bottom layer of the large model.
Wang Keden has now made more than 10 AI products, such as "Accidental Synthesizer", "Modihand", "Endless Contribution", etc. He frankly has no idea whether they can become hits. The popular methodology, becoming a trend, or attracting 200,000 users a day are not part of his plan.
Zhao Chunxiang hopes to take it one step at a time and the accumulated traffic can be reused in new products. On the community forum, he recommended his second product "Stranger Alarm Clock" through his identity as the creator of "Book of Stomach". Anyone who can download it and give it a good review can add his contact information.
Relying on the reputation accumulated by "Book of Stomach", "Stranger Alarm Clock" has 8,000 registered users a week. At the same time, he is also preheating his third product - AI smart collection, which helps girls collect information cards about makeup, manicure, lipstick and other information.
It seems that what Zhao Chun wants to do is easy. In fact, he has participated in the development of more than 10 products in the mobile Internet era, so the design of the product and the control of market demand actually require a lot of effort. of exercise. Subsequent products also know how to use their own traffic to cold start and complete the breakthrough of zero delivery.
So the probability of becoming a hit depends more on the creativity of the product and the results of operations. After all, currentlyMost of the products created by independent developers do not solve a wide range of urgently needed services.
Making money is not that easyJudging from the current development, these independently developed AI products actually do not have a large number of users. It is a "coaxing simulator" that relies on the convenience of the intelligent ecosystem to gain millions of users.
Because the product functions are not particularly complex, and the core technical barriers are not too high, everyone is basically making quick money.
For example, Zhao Chunxiang insists on paying for the App on the first day it is launched. He believes that charging cannot be done at the beginning, and subsequent free changes to charges will cause users to run away. And by charging fees from the beginning, Wei Zhi Shu’s revenue reached US$12,000 five months after its launch, ranking third in Apple’s food and drink category (the first place was Xiachian).
He insists on doing all the business from coding to operations by himself, and insists on zero investment in marketing. It is relatively easy to support himself as an independent developer.
For many independent developers’ AI products, the payment point is not too strong, and it is not so easy to make money.
Wang Dengke’s coaxing simulator is a good idea. With the disappearance of the GPTs ecosystem, he independently turned the idea into a web version with more than 10 levels. He also made a "solo" product and launched iOS and Android apps. His main business should be 6Pen Pro. This product is a paid model, and it uniquely adopts time-consuming billing instead of billing based on the amount of tokens consumed.
Chen Yunfei’s million-dollar income is actually more of traffic income. Based on his identity as a self-media blogger, he has also successively worked as a large model tool developer, a corporate consultant, and a contracted lecturer on a job search platform. a role.
According to Xinbang reports, the revenue he brings from providing AI consulting services to corporate customers constitutes the bulk of his income. Customer types include educational institutions, MCNs, and overseas e-commerce companies. I hope someone can help. They integrate AI into specific business workflows.
But on the 27th, he launched “Kitten Fill Light Pro” and began to try the paid download model. Less than 24 hours after it went online, it ranked first in the photography and video category list and ranked first in all paid application lists.
The above are some ways for everyone to make money through open source. In terms of saving money, since there is basically only one independent developer People, the biggest cost is actually calling large models. After Wang Dengke’s “Coax Simulator” became popular overnight, he spent thousands of dollars on token calls. So now as a famous independent developer, the product can get support from some large model manufacturers.
Currently, the web version of the coax simulator is supported by kimi chat, and Zhao Chunxiang is currently working on the Star Project.Developers, all products are deployed on the Yuewen ecosystem.
As for obtaining financing, many independent developers choose to go to various Demo days, but they are actually of no use. An investor told Brother Jing that they regularly check the number of token calls for various products on Dachang Cloud Platform, so they have first-hand information on whether the products are growing and whether they are popular.
According to Ai Doubi, although some investors are very optimistic about the AI search track, they do not dare to invest money in him. "As long as I do other things, I will invest immediately." There is no other reason. The search track is an area that major manufacturers must compete for, and data is the core barrier of AI search.
The technology of AI search is not complicated. If it is finally completed, the financing required will not be a small amount. Investors are afraid of being dragged into this bottomless pit track.
Therefore, independent developers need to stabilize a core user group after the product becomes popular, and exploring the long-term paying ability of this group is the most critical ability. But among the current AI products, the ones that users are willing to pay for are products such as ChatGPT and midjourney. The products made by independent developers are not “heavy” enough.
But looking at the problem from a development perspective, the original function of Facebook was to rate the appearance of girls on campus. Later it evolved into a social function and now has 3 billion users worldwide.
In the AI era, the threshold for this kind of creative tipping point has been greatly lowered, and now Baidu, Alibaba, Zhipu, etc. have launched AI coding assistants. Many people who cannot program can also use it. Natural language products. What’s even better is that AI is significantly reducing the operating costs of many creative ideas. Typical chat services such as Talkie are all generated by AI.
The AI developer market should explode rapidly in the future. After all, 11 people from Midjourney have created the myth of tens of billions of valuations. More independent developers will continue to create hot hits. Will Super APP be around? Born among this group of people?