A record of the process of using Claude vibe coding to make a Discord butler robot. From the anxiety of a technical novice, to working with Claude to create a working system, and the Ramon 30 course that made me take the first step.My notes are always scattered and every time I have the intention of organizing them, I come home from work just wanting to rest.
(I really had time to sit down and forget what I was thinking. The idea just disappeared.)
Notion has been built, the categories have been set up, and the organization is neat and tidy.
But it's like a steam game brick, you buy it on a whim and then it sits in your library and you rarely open it.
I want a robot that I can get used to, that I can use, that will automatically save the text I type into Notion, and that I can save it to Notion, and that I can save it to Notion.
They are able to understand my intentions, and if they can't, they will take the initiative to ask me and record my habits until the mission is accomplished.
This will save me from having to spend time doing manual organizing as a farmer.
It's just that I can't get past the technology hurdle, and I don't even know where to start with vibe coding.
How did it start?
I can only read part of the html and css, and I can modify the kind that I can't write.
I know I need to string APIs, but I have no idea where to start.
I'm also worried about the backdoor of the installation. Or what to do if it's broken? There were so many questions in my mind that I didn't start to deal with it.
This was put on hold for a long time until Ramon's online class.
I finished it right after it was launched, and I'm not done yet, I don't think I have enough knowledge to handle more things I want to do.
The course content is progressive, sincere, and not intentionally secretive.
This class is a lot more affordable than some of the other classes that start at $2,000 to $3,000, or escort services.
My favorite part for me is that I can read it by reading the written version without necessarily watching the movie.
You can skip around to find what you want to learn, and go directly to that chapter to learn what you want to do.
I was worried that I would blow up my computer, but after a few chapters, my worries gradually disappeared.
I thought that Claude is for engineers before the class, but after the class, I realized that I can do it easily.
After two weeks of messing around on my own, I've finally gotten my idea for the Shatterbot framework up and running.
(We also built an itinerary planning robot and a paper searching robot.)

The discussion papers are sealed.
Talk to Claude, and then Claude writes, doesn't understand things.
I just acted like a giant baby asking, "What does this mean, please teach me like a high school student," and went on.
How to design the structure, whether to add that field, where to put the function ....etc, how to design the structure?
These trivial decisions have accumulated into a pile of sealed documents.
(You take every decision for granted when you make it, and only when you save it do you see how many detours you've taken.)
At first it was a total mess. Ran, broke down, asked Claude to fix it, ran again, broke down again.
They think that new ones will appear after they are done, and so on and so forth.
Features are added one by one: semantic categorization, de-duplication, automatic review, wee batch organization ....etc, and
Every time we add one, we have to re-discuss how to connect to the original structure, after connecting and testing, after testing and repairing.
Until the robot reminded me that you can use a plan or a brainstorm to plan.
I realized that I was wasting a lot of tokens.
But through this process, I also understand what I want more, and it's kind of a concept of digging into ux for myself (?).
The course will say that a good assistant needs at least one month of training.
I'm in the mood to educate and communicate as if I were a new employee, and I'm practicing how to use AI to amplify my capabilities.
Clay texted me back for the first time.
The robot's name is Clay the Butler.
("For no other reason than it's a random name.)
I said something on the # mumble channel, and it came back with, "Okay! IB-42 saved 🗂️".
Notion has really added an extra dimension, the categorization is right, the labeling is right, and the abstract is there.

"Whoa, whoa, whoa, it's starting to run the way I want it to."
In the past, I've always thought it was great to have a "second brain," to move more of my brain into my notes;
But I just don't have the time to maintain it. Life is hard enough, and my head is always full of stuff, so I don't have time to sit down and file.
But now that the idea of typing in communication software is at hand.
The robot automatically sorts, tags, and saves to Notion, all at the drop of a hat.
At 3am it will automatically finish processing and then reply to my summary.
It's also a good idea to periodically take stock of the things that need to be delved into or the thoughts that need to be sealed off.

The fact that it was working while I was sleeping is still a bit of a godsend.
Maybe I will get used to it later, but I want to record my feelings now.
I hope to be able to move towards superstardom.
The vibe coding thing made me see a different direction for my future career development.
(It must still be said that robots still make mistakes and need to be checked by different AIs. If it's wrong, fix it. If it's wrong, fix it.)
Current Status
It started with Raymond's class and the rest was polished with Claude.
Anyone who is similarly stuck can go(coll.) pretty soonI truly feel that even a beginner can get started, so don't be afraid.
(I'm not sponsored, I just recommend it from the bottom of my heart. At least it worked for me)
When I have the chance to make a simple app to solve my daily problems, or a game, it seems that I can take out the framework that I discussed with my friends in the past and continue to implement it.
Today, I am a little bit closer to the direction of AI Collaboration Architecture.
(At least I've moved on a bit from the person I was six months ago who was afraid of screwing up my computer.)
Frequently Asked Questions (FAQ)
A: I did, so I should be able to. Just have the patience to keep talking to Claude and keep asking questions, and accept that there's no such thing as a one-time fix for the process of "running and fixing".
A: It is to use natural language to talk with ai, and then make the corresponding program. I think of where to do, encounter problems and then solve them, there is no specification. For non-engineers, it may be the best way to go on.
A: For those who think Claude is great but don't know where to start. He is so easy to understand that you can read the text version without having to watch the movie all the time.
A: In the past, I had to switch states to organize my notes, but now I can say one sentence in Discord and let the robot do the rest. It saves a lot of effort.
A: It's running steadily, but it may not have detected the error yet. There are still a lot of features I would like to add, but the current state is enough.




