[Annual Teammate Experience Sharing] A year later, I'm honored not to have disgraced the title of Knowledge Base Monetization Genius, a 2001 college student shares 20 months of AIGC experience

Since I shared all my project experiences and insights from Knowledge Planet on December 9th last year, I’d like to first thank all the seniors for their support and encouragement. I’m also very grateful that thanks to Brother Gou’s Knowledge Planet, I truly stepped out of it after a year and took my first step.

Before I dive into my experiences, let me reintroduce myself. Just call me Xiaoxiong. I’m based in Wuhan, an ENTJ born in 2001.

Next, I’ll continue in a timeline format, interspersed with my real experiences and accompanying pictures, to detail the results I achieved this year and what the industry + AI niche can truly bring us.

P.S.: This article was written with only three packs of cigarettes. It’s entirely ad-free and solely shares the current state of my industry (from my current understanding).

Picking up where we left off (Season 1)

Quick browse: You can go directly to 👉 AI Dad’s Douyin, the episode from March 2024, which received about 7000 likes and is titled “AI Project Teardown,” produced by a top operator from Brother Gou’s Planet.

AI Dad’s Douyin - Douyin

【Experience Sharing】9000-character long article giving back to Brother Gou! After half a year in Brother Gou’s Planet, a 01-born college student shares all project processes and experiences empowered by AIGC in 10 months

After December, although several remote prompt positions I applied for didn’t even grant me an interview, I even suspected they might have just opened a position to gather low-cost market feedback on AI interviews to see if it met their needs. Of course, I was also very fortunate to be recommended into an AI virtual girlfriend internet project for WBE2, funded by Sequoia. It was a Beijing-based company with online remote work, and I served as an operator. My salary quadrupled from 2500 last year to 8000. Due to time constraints, employees also worked online across the country, so there weren’t many cumbersome procedures or office politics. Everyone just fulfilled their duties, planned their work, and allocated responsibilities. We also planned several large-scale project activities, essentially experiencing the entire 0-to-1 process of the product, from product development, user research, product planning, platform traffic acquisition, user growth, user maintenance, to community maintenance, covering several complete project life cycles.

For example, for the New Year’s Eve event, we directed traffic from Q and V groups to our AI virtual girlfriend, the Goddess of Wealth.

During Chinese New Year, we established a 0-to-1 matrix for traffic acquisition on Bilibili, with over a thousand new registered users daily.

(Because the payment function wasn’t developed at the time, we were losing 50,000-70,000 daily.) I wonder if I caused trouble.

This is also a data point I’m proud of, the result of working on Lunar New Year’s Day, an 8x increase in conversations.

Unfortunately, good times don’t last. In April, domestic regulations began to control the scope and safety testing of virtual AI conversations. Several larger personal websites with a certain user base were “sent to the sewing machine” (shut down). So, in April, the company decided to dissolve its entire team, relocate its main entity, and venture into overseas markets. It’s a shame; there was no plot of a chosen one dominating the domestic market and then conquering the overseas market as you might imagine (if you really think that, please uninstall Tomato Free Novels from your phone first).

By July, the company’s project unfortunately ran out of funds because the second round of investment was not timely. The project ended there. During the same period, compliant AI emotional companionship projects from major domestic companies were also flourishing. I won’t list them all here (it’s interesting, it seems that last year, AI painting was also an early entrant, but ultimately, the technical barriers were vulnerable in front of big companies. One-click application was just a matter of time and information asymmetry. This year is the same; technology is always in surplus, and only products that meet user needs can survive. It also confirms the saying, “The lone runner is fast, but those who walk together go far.”) It’s a pity my three months’ salary was not paid, and I didn’t pursue legal action. I’ll explain why later.

P.S.: If earning coins from Tomato Free Novels could get my salary back, that would be great.

I’m very curious, for those of you on the planet who have read this far, do you feel like I’ve been a bit too ordinary this year?

If that were truly the case, wouldn’t it be underestimating the effort of averaging only 5-6 hours of sleep per month?

Going back to November last year, precisely because of the gap after resigning and the advantages of working from home, I started planning short video e-commerce on Video Channel with a university classmate. You can imagine it as genuinely working two jobs a day. Of course, thanks to my previous AI accumulation, I truly had an boss who previously paid for my knowledge investing in me, and it was a significant sum for the first investment. In February this year, I truly became a top-tier corporate slave: operating during the day, investing traffic at night, and then packing and shipping at home, clearing returns with a small cart as part of everyday life.

For certain reasons, on March 21st this year, I chose to venture far away and try my luck. I took a 6 AM train… arriving alone in Yiwu. Together with the investor, we secured a 200-square-meter office in Yiwu CBD. I personally assembled every computer, decorated the company to my liking, imagined harmonious coexistence with employees, bought snacks, hired endless people through Boss Zhipin, worked till 3 AM, attended countless exhibitions, collected endless samples, and processed tons of returns. Haha, if I could do it again, I definitely wouldn’t do e-commerce. AI + e-commerce still can’t beat platform rules.

So now, let me briefly explain how I empowered e-commerce with AI. The great truth is simple! Often, business logic is very straightforward, without too many cumbersome steps and details. Addressing pain points and saving costs is the purpose of businesses using AI. This is my most intuitive feeling about AI after becoming a boss for the first time.

To ensure the credibility of what I’m saying and to explain how AI can bring seemingly common but surprisingly effective results to e-commerce, I will use data to substantiate the feasibility of my methods. First, a disclaimer: the platform I’m working on is not Douyin, it’s WeChat Video Channel e-commerce, and it’s not live streaming, but purely short video product linking! Only short video product linking!

First, we need to understand short video product linking. Why can short video product linking be empowered by AI, and in what aspects can AI empower it?

Why do I say that the great truth is simple? For me, a viral product only has two most important points:

1. Recognized by users, with data verified by the market (data can be referenced from third-party platforms like Chan Mama, Kaogu Jia, and various compasses).
2. Heavy paid traffic leverage.

Here, we will explicitly focus on the first point. If a product has been validated by short video platforms like Douyin E-commerce, and the content is confirmed to be a hit, then all that’s needed is to copy and adapt it. However, this is easier said than done. How to copy? How to adapt? How to get approval? These details are where the copying can be done well. When I coached my employees, I effectively used AI to empower them in these three aspects, allowing them to transition from industry novices to qualified operators or operational assistants within a week. AI empowerment played a crucial role, but employees also needed a certain level of悟性 (comprehension/aptitude) and strong SOPs to guide their actions (I’m a black-hearted boss, 9 AM to 7 PM).

(I) First, viral content and videos have data reflected on third-party platforms. Even if you can’t extract them, there are tons of original video extraction tools on the market. Therefore, the top priority is to download multiple viral content pieces for their hot products. The quality of video content directly determines the ROI of paid traffic, and also dictates the data for short video product linking. If the data is poor, the horse can’t run, there will naturally be no traffic, leading to an endless loop. Thus, video content is the most crucial foundation for a high-quality short video product link. If the foundation isn’t laid well, an endless loop will inevitably ensue.

(II) After collecting high-quality material, editing techniques are also a critical part. However, editing ultimately depends on individual comprehension. My company’s operators handle both editing and traffic investment (this prevents blaming between editors and advertisers, and I believe operators who can’t do both editing and traffic investment are not suitable for e-commerce positions. Perhaps because I come from an e-commerce background, I think e-commerce positions only suit复合型人才 (versatile talents), not those with singular skills. The current society is also quite competitive, and those with only single skills have long given up or resigned from my company). So, how can operators improve their editing techniques? Again, it’s by copying. While sourcing material, they are also learning from their peers’ quality content. After watching so much good material, even if they draw a gourd according to a dipper, they can achieve some resemblance. Not to mention, Video Channel’s requirements for video quality are not as high as Douyin’s.

(III) After the material is edited, voiceovers and copy are purely replaced by AI. Here, I also taught my employees how to use the AI website I built last year, how to teach AI to learn the content we provide and generate corresponding high-quality copy, thereby achieving perfectly de-duplicated copy. (An information gap here: Video Channel can directly use audio from Douyin; it must be human audio. AI audio will lower the production-to-investment ratio of the entire video.)

(A small ad for a website on the verge of collapsing: AI dialogue, painting, PPT, music, video integration mirror, scan the code to browse)

(IV) Once these three points are achieved, the video is basically ready for approval on Video Channel (information gap: Video Channel rejects hard selling ads, so strong marketing ads can use a 15-30 second digital human voiceover with a picture-in-picture at the end. The digital human only needs to introduce the product in the video). At this point, simply use your collected database of corresponding products, have AI write a long-tail keyword title, and then publish for promotion to increase the weight of search traffic after traffic investment.

The great truth is simple. Only after entering an industry do you understand what AI can truly empower it with. Perhaps it doesn’t bring ultra-high efficiency, but in this recognized AI era, I will still find ways to use AI wherever possible, even if it doesn’t bring a very noticeable increase in efficiency. It’s important to develop the habit of using it. Perhaps many things, if not done with AI, would yield different results. Of course, for me, the most important thing is that employees don’t actively record important details during meetings, but I make it a habit to use AI for meeting notes every time, which helps employees identify problems and summarize meetings. For me, this is one of the key efficiency tools for improving employee productivity.

Below are some of my data.

The above are some experiences worth sharing from November last year to September this year. Why September?

Unfortunately, just like last year, I’ve given up (laughs).

Because the return rate on Video Channel went from 5% last November to 40% for the same product this year, I believe the demographic on Video Channel has been negatively affected by some merchants selling fake goods, malicious low-priced items, and human factors in Yiwu. Ultimately, I concluded that the feasibility of this project is no longer high, so I decided to terminate it.

Summary:

After a month of reflection and review, I’ve summarized several reasons why I believe it’s not worth continuing:

1: Projects with heavy upfront payments have slow capital turnover. Video Channel’s capital turnover cycle is 2-3 months, and the uncertainty of delayed risks is frightening.

2: I operate a dropshipping e-commerce model, with many products shipped directly from manufacturers. However, human nature is tricky, and within a few short months, there have been nearly 100,000 legal disputes, all because manufacturers refused to refund me. Civil lawsuits are time-consuming and costly; stubbornly fighting them is a thankless task.

3: The issue of human nature in Yiwu, or perhaps my age. My employees are generally older than me. For me, the turnover rate is astonishingly high. It seems all migrants who come to Yiwu have an entrepreneurial dream? If starting a business were really that easy, haha.

Thoughts:

Many people have told me things like, “You’re smarter than your peers, you’ll definitely succeed in the future.” But I’m someone who was so anxious about not finding a job after college that I felt depressed; I’m an extremely pessimistic person. All my experiences at this stage don’t bring me any sense of achievement. Truly smart people know how to use others and how to be used by others. “Using” isn’t manipulation or backstabbing, but leveraging one’s own or others’ examples to achieve a win-win for both. I’m not afraid of being used; I’m afraid of being called useless.

Questions:

From these past two years of experience, my relationship with AI has evolved from learning to application, and then from application to niche industries + AI. It seems I’ve preemptively laid out some directions I chose and some seemingly cutting-edge industries, yet this hasn’t brought me any achievements. Opportunities in the AI era seem to be gradually replaced by big companies and capital, and it seems increasingly difficult for young people to find a breakthrough opportunity. I feel lost again.

Finally, here’s a photo of a young aspiring individual with Brother Gou.