"Independent Development Diary 10: Experiencing Life in Vietnam While Redefining My AI Product"

Personal Life

Last week, I went to Ho Chi Minh City for visa matters, and it gave me a very vibrant feeling. I saw young people everywhere in cafes working on their computers. I also met a friend who works there, and their Vietnamese friend taught us a bit of Vietnamese. It feels like Vietnamese needs to be spoken carefully; it seems a bit difficult to learn. However, some words have pronunciations that are the same as in Cantonese, which I found quite special. Additionally, some national heroes there were the heroes who fought against the Ming and Qing dynasties.

I personally find Vietnamese cuisine quite familiar, as it includes a lot of vegetables. Although most of it is raw, it’s not a big problem; my stomach can handle it. The rest consists of various types of noodles and some French fusion dishes. Overall, it doesn’t feel much different from Shenzhen, except that there are a lot more motorcycles. This is the opposite of Thailand, where motorcycles weave through vehicles, and actual road rights belong to cars. In Vietnam, the opposite is true; road rights are primarily held by motorcycles.

I also experienced Vietnamese massage, which is quite different from Thailand. In Thailand, it’s more about the aunties giving you a heavy punch for physical relaxation. In Vietnam, it’s lovely young ladies giving you a massage, which is more about mental relaxation. The locals can drink coffee all day long; it feels like they are drinking coffee at all times. By the afternoon, I didn’t dare to drink anymore; I can’t handle 2-3 cups a day.

sanvi.dev

Back to the main topic, this week I mainly focused on optimizing my personal website as a whole. The initial version was created using AI based on my resume, but it was too much like a resume and didn’t resemble a personal website. This week, I didn’t want to blindly add features, so I concentrated on adjusting the personal website. I mainly added some projects I’ve worked on and some experiences, and I synchronized the articles from my public account.

The website supports both Chinese and English versions. Basically, all the products I’m currently working on will be bilingual, mainly to prepare for more opportunities to go overseas in the future. I’m still quite pessimistic about the global economy, so I prefer to have more options. I also added a feature to the blog on my personal website, where the articles will be published with an additional English version. This involved some smart cutting and stitching, which took some time to figure out.

This week, I should also add a feature to see how I can upload my first album, "Unimportant Little Things." As my experience in songwriting grows, I recently did something where I selected a voice from the first album to be the lead singer for the upcoming second album. The second album will likely feature the same voice, while the third might utilize some current AI large models to create a virtual idol. These are things I want to do in the future, but unfortunately, my energy is limited, so I want to focus on things that provide positive feedback first.

If anyone is interested in discussing or collaborating in the future, feel free to reach out.

StickerAI

Last week, I didn’t make any functional updates; I was mainly handling some SEO optimizations. I’ve integrated with Ahrefs to check for SEO issues, but I’m still exploring this area myself. So far, I haven’t seen any good conversion results. This week, there’s still no growth in paid users, and I noticed some similar email addresses are exploiting new user points. This hasn’t had much impact on my current user base, so I’ll let them be.

Another plan is to create some KOL stickers that can be shared on platforms like Douyin and Xiaohongshu to see if they can bring in some traffic.

PromptPlan

Last week, after adding some statistics, I can see that Chinese users actually account for less than half, with some users from the U.S. and other countries. However, the majority of the usage language is still Chinese and English. Additionally, I made some SEO optimizations and adjusted the sitemap. After adding internationalization, it seems that search engines often fail to index due to redirection issues.

The next idea is to establish a user memory function because I’m currently facing a significant problem: large models cannot understand user intent well. For example, when a user says something, it’s unclear whether they want to create a task, organize a task, or something else. The unified prompts lead users to feel like they are drawing cards, and each return cannot be standardized. Therefore, I plan to create a function that remembers users' preferences for return results, which will allow the large model to provide results that better match user expectations.

Others

The patching method I used for the quantification program previously led to bugs that I couldn’t fix well, so I decided to redesign some design patterns, optimize, and simplify the logic. I’m currently still debugging. Recently, I’ve been discussing a new idea with friends, and I’ll share more about it later when it’s more developed.

I’ve noticed that many people continuously work on a project, but I believe everyone has projects that suit them. If a project shows growth, then invest more energy into it; if not, keep trying.