💬 Deep Dive
In my daily work, I often need to answer product questions for overseas users. My approach is to write Chinese first, then use ChatGPT to translate it, then copy and paste it into the dialog box and send it to the user. The whole process is quite cumbersome.
So I thought about whether I could directly enter Chinese in the dialog box, and then use a plug-in to automatically translate and replace the original text, so as to save the trouble of switching back and forth.
After having this idea, I immediately tried to make a plug-in using Trae. The effect was not perfect, but it was usable.
For a person who doesn’t know how to write code at all, relying entirely on AI to generate code, piecing it together step by step, and making a plug-in that can run in Chrome in half an hour gives me a sense of accomplishment.
At the moment, it can translate and replace text in the input fields of Grok, Lovable, and X. But it still struggles to identify input fields in Reddit comments and Google Search, so I’m continuing to refine it and haven’t listed it in the Chrome Web Store yet.
In the process of optimizing this plug-in, I also encountered problems and had a few thoughts:
1. AI programming lowers the threshold, but people with zero foundation will still get stuck
Making plug-ins may be the easiest way to get started with AI programming, but even if AI can help you write code, you will still encounter various pitfalls with zero foundation. Especially when you want to optimize functions, AI may disrupt the original functions into pieces.
For example, I hope that the plug-in can also use shortcut keys to translate and replace text in the Reddit comment area and Notion, but the new version of the plug-in generated by AI not only includes the translation in the translation results, but also adds the original text. And it will also change the style I set before back to the original ugly style.
So, we can only go back. Later, after reading the tutorial, I learned that development generally uses Git for version management to avoid changes that would affect the original version. I had never been exposed to this concept before.
Moreover, AI programming is not very stable. If you start over from scratch with the same requirement, the generated version will be very different. I have reproduced it from 0 three times and still cannot reproduce the original satisfactory version.
Therefore, AI can indeed help people with zero foundation to write code, but if they don’t understand development at all and don’t know how to debug, they will still get stuck.
2. Identifying needs is more important than rushing to develop
Last year, I listed some development ideas. At that time, AI programming was not as strong as it is now, and I didn’t want to spend too much time learning JavaScript from scratch, so I put some ideas in my notes.
It wasn’t until the second half of the year that tools like Cursor and Windsurf appeared that I felt like I could give it a try. But even though AI has evolved, the several attempts I made from December last year to January this year had only mediocre results, so I put it on hold for a while.
Looking back, this is also a good strategy for zero-based users. Record your ideas and needs first, let AI evolve, and do more important things first. Maybe in the first half of the year, many development problems will have better solutions.
Identifying needs is more important than rushing to develop. Demand comes from daily details. If you find some tools that are inconvenient, uncomfortable, and need to be switched back and forth, it means there is a demand. Just like, my translation plug-in comes from my overseas customer service work.
Nowadays, there are many independent developers releasing products on Xiaohongshu, but not many good products are actually seen. Many people are still doing three-piece accounting, diary, and to-do sets. Most of the UI designs are not good-looking, as if they are developed for the sake of development, which is not very interesting.
3. Taste, personality and values are more difficult to copy than code
Lenny asked Ivan what other factors contributed to Notion’s ultimate success besides hiding the vision behind productivity tools? Ivan replied:
First of all, the product must have users and revenue. This is the basic goal, but more importantly, you must create ** based on your own values, that is, make something you think the world needs. Everyone has their own taste and aesthetic and ultimately it’s about creating a product that resonates with them.
Balint, the founder of Craft, also expressed a similar view:
It is becoming easier to create software that works. In the field of personal software, the key to standing out among many products lies in the personality and uniqueness of the product.
At the functional level, AI can help you do a good job sooner or later, but taste, personality and values are difficult for AI to copy. 61’s MD Vinyl is quite popular recently. He can make such a design because he himself is an avid vinyl lover.
Products are the embodiment of the creator’s own life experience and cognition. This is the core of product differentiation and is more difficult to copy than code. You must first have your own Vibe before Vibe Coding can have Vibe🤔
Final
AI makes the world more alive, and there are more and more useful AI tools, which is a blessing in itself. Plug-in development is the first step, and what I’m more looking forward to is polishing an app of my own with the help of AI in the next year or two.
Reference information:
Notion’s lost years, near collapse during COVID, staying small to move fast, building horizontal
Design-first software engineering: Craft – with Balint Orosz
💎 Curated Gems
1. Personality is a spectrum
In the past two or three years, MBTI has become a fashionable talking point. It is simple and direct. You can quickly define yourself by putting a label on it, and it is easy to gain social recognition. But it emphasizes group commonality and ignores individual differences. Moreover, this simple classification may also make people regard personality as a final conclusion and default to “I am this type and I can’t change it.”
In contrast, using traits to understand personality is more realistic and scientific. Traits are not fixed types, but like a spectrum, and each person’s position on the spectrum is different. For example, some people are more extroverted and enjoy social interaction; others are more conscientious and tend to plan and be self-disciplined. And when these traits—extroversion, conscientiousness, agreeableness, neuroticism, and openness to life (the Big Five personality traits)—are combined in varying degrees, they create vastly different individuals.
And traits are not static, so personality is not a static label or classification, but an adjustable spectrum.
Reference information:
“Why We Are Different: 40 Lectures on Personality Psychology”
2. There is no unified standard for successful founders
Jessica Livingston is the cofounder of YC, the only non-technical founder of YC, and the wife of Paul Graham.
She is outspoken, maternal, and compassionate, which is a far cry from the image most people have of a startup founder. But this quality of hers makes YC like a big family, and she is like the mother of this family, caring about the founders’ feelings, paying attention to whether they are feeling too stressed and eating properly, providing them with advice on relationships strained by the pressure of entrepreneurship, listening patiently, and helping them resolve disputes and breakups with co-founders.
So if you want to start a business, I suggest you think about what makes you unique. What is your unique combination of abilities and interests?
And don’t edit your answer, because my example shows that the most unexpected ingredient may just be the key to success.
Reference information:
3. Work requirements
In fact, the essence of any job is to solve problems and provide results, not to complete things.
Wang Shimu said that being a product manager requires creative problem solving.
How do you solve problems creatively? Make connections—for example, bringing vinyl records into NetEase Cloud Music’s playback interface. How do you develop that ability? The work happens outside the immediate task:
- Gain daily insight and pay attention to the problems around you;
- Strengthen basic skills, such as data analysis, user analysis, and market analysis;
- Inductive deduction, summarize the observed inspiration and put it into product design.
Reference information:
🧵 One More Thing
When I don’t feel like doing anything, typing out a passage can be a good activity. Copying long texts feels tedious, so I choose short pieces from Luo Pang’s 60 Seconds·Ten Years Collection, which is an attempt I started in 2023.
On the one hand, you can learn some new knowledge and writing methods (without forcing the content into your brain), on the other hand, it can also be used as typing speed training.