For me, the companionship of a community is more important than the community content itself. Before joining Shengcai, what I most wanted to participate in was the 21-day Navigation project. The Navigation project I joined was Web Website Going Global. To better motivate myself to learn the project, I applied to be a volunteer during my first Navigation. Fortunately, I was selected. It was my first time participating in Navigation, so I had no experience, and naturally the results could only be mediocre.

1. Overall Experience

1.1 Navigation Manual Learning

When I attended a Shengcai offline meetup before, Teacher Zibai said that besides members, the most valuable thing about Shengcai is the Navigation manual. As a volunteer, I had to study the Navigation manual carefully. I read both the basic and advanced sections twice. In practice, I followed the manual to create a game site and a tool site. For keyword selection, I followed the new word list recommended by Coach Liangchen Mei on X, and I’ve now developed the habit of browsing X.

Currently, the person who has influenced me most on X is Teacher Baoyu, not only because I took his “The Beauty of Software Design” course on GeekTime five years ago, but also because his views are really useful, especially about AI programming. Although I’ve been using AI programming all along, I basically took a wild approach, somewhat at will, without method. After reading Teacher Baoyu’s shares, I realized how AI programming should really be done.

Sharing a few of Teacher Baoyu’s views on AI programming:

For experienced programmers to use Claude Code well, the biggest transformation comes from a change in thinking and development habits.

This transformation is to design first, then write prompts, and then use prompts to generate code.

“Design first, then write code” is actually old advice, but saying and doing are different. Although we claim software development involves design first, that’s usually for overall system design, but when it comes to specific implementation, few people do this, because programming details aren’t clear all at once. Even if you’re experienced, in modules you haven’t implemented before, you haven’t fully thought things through until you’ve finished writing. You have to start coding, thinking while writing, adjusting or even scrapping and rewriting parts, going back and forth like this until you finally understand it.

Many people don’t have the habit of reviewing code, let alone testing their own code. Every time I have AI generate code, I carefully look through the results, checking if the code matches my expectations—if I were to write it myself, how would I write it, is its solution better or worse, if better I can learn and happily accept it, if mediocre then so be it, if not good enough I’ll roll back and adjust the prompt, or add some requirements.

Testing is also very important. Test cases for unit testing need to be designed and reviewed by yourself. Manual testing is also essential. Try to minimize testing costs, for example by testing through command line, testing code, so that every time you generate, you can immediately test and verify. If there are problems, roll back or fix.

Doing this at first is very uncomfortable, but when you adapt, you’ll like this development method, and the results will be better.

PUA-ing AI is just a gimmick and not a truly useful technique. When AI-generated results have problems, we often don’t continue modifying, but rather re-adjust the prompt, roll back and regenerate. AI-generated content is more like AI image generation, constantly generating new versions, if unsatisfied adjust the prompt and regenerate, or switch to another model and regenerate, rather than repeatedly PUA-ing AI to modify based on an erroneous result, because once AI generates incorrectly, the wrong result will influence subsequent generation.

Don’t treat it as a truly intelligent, omnipotent creature, but rather see it as a probability generation tool. Wrong results will continue to amplify the probability of errors.

Remember one principle: Only continue fixing small errors, for slightly larger errors directly roll back to modify the prompt or switch models and regenerate. Don’t be reluctant to spend those tokens, compared to wasted time it’s not expensive.

1.2 Volunteer Status

As a newcomer who only joined Shengcai on April 18 this year, with a desire to improve, I applied to be a volunteer during my first Navigation. First, I could participate in Navigation without a deposit, and second, I wanted to use the volunteer status to better motivate myself to learn project-related content. I actually didn’t expect to pass, after all, I currently have nothing outstanding. Fortunately, I passed the volunteer assessment, which made me a bit happy.

I had no experience operating communities before, and I’m not one who loves frequent communication. Fortunately, with the help of the team leader and other volunteers, I was able to ensure the smooth progress of the group’s Navigation. This Navigation had a total of 39 people. During the 21-day Navigation, my main task was to remind crew members to pay attention to Navigation-related activities (such as live broadcasts, check-ins, etc.), and to explain Navigation-related details to some crew members. The most frequently asked question was about VPNs, since Navigation has a lot of content requiring access to foreign websites. I recommended airports and related tools to 4 community members, including one from Chengdu who wasn’t very familiar with operations and asked more detailed questions. I even provided remote assistance and helped him register a Google account.

As a volunteer, our group’s results were quite good, with a 90% landing rate. In the early stage, I also checked in every day, which better motivated team members to complete check-ins. Several team members achieved good results, with website traffic reaching over 500 per day, and one team member even had a little income from Google ads every day. But more team members were like me, basically starting from zero, so they didn’t achieve great results, but this project’s effect cycle is naturally long, so this is normal. Shengcai founder Yiren also expressed similar views when replying to community members:

Anonymous User Question: Hello Teacher Yiren, I’ve read many posts about web development on Shengcai. You said this is a super standard, including now Teacher Liu Xiaopai leading the team in the deep sea circle, but in three to six months, very few people get results. Like getting the first order in three months, three months of time cost, earning tens of dollars, what’s the value in that? And a bunch of people haven’t gotten orders in six months. Compared to Xiaohongshu, like virtual e-commerce, finding the right product selection, 1-2 weeks can get orders. I want to hear you elaborate on why you’re optimistic about web going global, and personally invited Teacher Liu Xiaopai to lead the team.

Yiren’s Answer: Precisely because most people have your way of thinking, Xiaohongshu virtual materials is just a project with very small space not worth doing as a long-term career, while making AI products is a project with huge space worth doing as a lifelong career. Various capability improvements in AI give everyone the opportunity to create web pages or APP applications, to participate in competition in huge markets, this is the fortune of us ordinary people.

Making AI products, slow to get results, this is very normal, because many people spent a lot of time just learning to make products, and then have to figure out how to get traffic, but this is a long-term compounding thing, earning more and more.

2. Growth and Shortcomings

2.1 Growth

First, regarding game sites and tool sites, I learned a little bit, knowing how to get a website running;

Second, regarding community operations, through the entire Navigation project, I saw the whole process, from volunteer selection, crew boarding, Navigation launch, Navigation sailing (expert navigation, check-ins), all the way to final docking. The role of team leaders and coaches is crucial. Our team leader was very proactive and could create a great atmosphere;

Finally, regarding the project itself, during the Navigation process, everyone discussed the same content every day, easily generating resonance. In this process, I also saw how some experts were made, which also helps with one’s own progress.

During this Navigation, the team leader made daily report images from our project group chat records and even created a website to consolidate daily chat highlights and crew projects, making it convenient for community members to view. At first, I didn’t think much of it, thinking it was no big deal, but recently I found it very useful, so I learned it and found it very interesting. I made monthly reports from an old friend group’s chat records and found our monthly chats were really worth commemorating. The team leader’s tutorial is here:

https://llzpdb9grt.feishu.cn/wiki/MlfawxNqLiZck0k5lYfcIpArnDt

2.2 Shortcomings

Before Navigation, I set a goal for myself to launch 10 websites, but unfortunately I only launched two. And my project learning wasn’t deep enough, some check-ins were a bit superficial, I couldn’t set a good example for crew members.

Due to personal personality reasons, during this Navigation, except for communicating more with that Chengdu community member, I was basically a robot, unable to connect well with community members.

But I’ll still apply to be a volunteer next time. I feel the volunteer status will motivate me to improve.

3. Summary

One person may walk fast, but a group of people will definitely walk further. With the companionship of the community, I’ll have more confidence in making global websites.

I’ve passed thirty this year. My main job is programmer, facing the 35-year-old crisis. I want to do some difficult but correct things. Global websites are what I want to do recently.

As Teacher Gefei said:

Time is magical. If you feel something is difficult, give yourself more time, so monthly tasks will be easier.

Time is magical. If you planted a fruit tree a few years ago, after years of watering and fertilizing, you can harvest fruit now. If you plant a fruit tree now, a few years later, you’ll also have a harvest.

Time is magical. If your product can bring you income every day, then multiply by 30 days, multiply by 365 days, the accumulated harvest may exceed your imagination.