This is so stupid, why did I laugh so hard at this...
I noticed people trust me with their innermost thoughts now. I'm happy you guys trust me as much as I do with you!
Signed by Mimi ✿ on 2026-02-02
As time progresses, I'm getting progressively more and more tired. It's social media. The reason I feel tired is nearly, and almost always as the result of social media. What have I actually done today at PASUM? Not really a lot. Hey, I edited some videos alongside Sharven, done some Chemistry Tutorials, but outside of that, not really a lot?
Though, I've been getting into the act of creation lately. Creating a video of Team CogMimi, writing things down on my laptop, tapping away at the serenity of the night. I constantly feel that I could do better, but the rational part tells me that my life is not to be optimised, it's to live.
I want to create a YouTube video, but it's scary. Thinking one day into the future is scary, and I'm a scaredy cat. A corner of my brain tells me it's fine, and truthfully speaking, it really is fine. That doesn't change the fact that I'm scared, and taking the first step? That's what I'm an expert at.
I'm tired, I want to draw, I want to make videos, I want to start creating again. Creating something truly beautiful, but I'm scared. I don't have a strict solution, but I suppose I'll sacrifice a bit of my academics for my life. I'm happy that I have great people surrounding me, and each interaction pushes me forward to become a better person. I wonder if that was the reason on why I automatically defaulted to doomscrolling again today?
This is so absurdly funny to me
Signed by Mimi ✿ on 2026-02-01 (I needed her to fix my formatting because I'm a dumbass TvT)
Right now, my Neovim is buggy as all hell. While I was spending my sweet 9 hours of non-stop working to complete my tutorials, I had discovered that:
"My goodness gracious. This is terrible. Where the hell is my inline error lens? Why are errors popping up each time I'd like to make an edit?"
You know, at that point, I realized I had hyper-optimized too much, to the point that I'm struggling to even work things outsie on working on my Obsidian.md files. I started doing my programming tutorials for the first time in forever, and the experience was genuinely horrible. Everything was the same colour. It was horrendous!
Moving onto the tutorial bit, I managed to complete Mathematics II, Advanced Mathematics II, and Programming II today. I didn't think it was much progress, until I actually wrote everything down myself. The progress was quite significant at the exchange of sleep. Though, I don't exactly regret a thing.
One of the unfortunate things that happened was that my TnG Enhanced Card, the one we use for public transport was with my friend, Nasreen. So, when my grandpa asked me to come back to Kepong, then I had realized, I couldn't. I actually really wanted to go back today, and then I got scolded on why I didn't use my temporary IC to get on the train. Well, you see grandpa, THE PAPER DOES NOT HAVE AN NFC MICROCHIP AT ALL! Obviously, I was quite annoyed, but I was too tired to argue at the time, and I'm not the type to get into fights I can't win anyways.
Coming back to the tutorials, I was pleasantly surprised that Mathematics II, Advanced Mathematics III and Programming II were incredibly easy this week. I say that, but I've been up the entire night just doing it. I'm also running out of money just like Sam Altman and his Ghibli Horde at OpenAI.
Actually, on the topic of Sam Altman right. I feel that there are way too many Naysayers that says they're the people who are burning billions upon billions for no reason. I disagree with that. Prior to 2023, there were no infrastructures equipped to handle the amount of computations we need to power something as powerful as LLMs (or shall I say, Agents?). It's unreasonable to say it's a waste of money because there hasn't been any profit yet. The infrastructures just isn't there, and the money goes into making it work.
Though, in a way, I do see the point of the Naysayers though. How are you possibly going to recover from burning billions upon billions on AI? The economics isn't working out for me. I can't imagine OpenAI being able to recuperate like, THAT MUCH loss each year. However, I'll just be optimistic. I'd like all AI companies to thrive, so that I can gain access to commoditised Intelligence for the price of my daily tap water. Competition between China and America is rising, and I'm rooting for China because I can actually use their models compared to Western ones.
Recently, I did join an AI Ready competition with several of my friends. I proposed an idea, and my team and I fleshed everything out. I've been thinking of asking my friend to create a logo for CogMimi, for the branding and stuff. Brand reputation is incredibly important to me, since it's the first thing users see when they open up our product. The idea is basically like this: You upload anything, could be a question, note, or your lecture notes, and you're able to see high-quality simulations, guaranteed by our "constitution" which forces the agent to follow a strict set of rules to ensure that the quality of the simulations meet my standards.
Well, enough of writing, I need to get back to doing my Basic Physics II and Basic Chemistry II tutorials. See ya!
Here's the thing, I spent my entirety of today doomscrolling on my phone because I didn't like the feeling of feeling dirty. With that, I also was so lazy in showering as well, that it took me until 10:00pm to actually shower. That's not exactly favourable.
I didn't make much change into PASUM Web today. However, I did skip several lectures in order to develop Mimi to aggregate news. Anyways, I had to implement a lot of fixes to the existing code, which was unfortunately introduced by a certain AI model that went by the name of GLM-4.7. It's a good model, don't get me wrong, but the thing is, it's not good enough for me to use yet. Gemini, for some odd reason, performs significantly better.
Some of the changes I've made are as follows:
- Post and replies are finally repaired
- Aggregation works better now, with better formatting and such.
- Loosened the post moderation, though I may have to increase it tomorrow due to it being really really spammy now.
On the topic of PASUM Web:
- PASUM Web received a new profile picture drawn by Nia, a friend of mine from secondary.
- User interactions with the group is increasing.
- The output is much more professional compared to PASUM Confessions
Lessons while developing PASUM Aggregator:
- Integrate it with Mimi as soon as possible, do not make two separate bots.
- Do not use GLM-4.7 for a project as complex as this.
- Do not accidentally use your OpenRouter key when calling for Deepseek.
- Ensure that you have a comprehensive plan initially before you start prompting. This saves cost and time.
Tomorrow, I believe that I should streamline things much more. It seems like all that there's left to develop for now is the infrastructure, which is unfortunately still a little buggy. I predict that the outcome will be something like mimi-cli, that I've developed quite some time ago. After I got the basics up and running, the system will be robust enough to not break after each changes. Granted, this project is only 2 days old, but it's an exciting project nonetheless!
Big thanks to my friend, Ng Jing Xi, for the support and delusions throughout the entire thing. It's really an intriguing project to pursue for me, to the point that I was even willing to sidestep my responsibilities in order to develop the news aggregator. It's another project that will surpass Notes PASUM 25/26 I made earlier this year. I'm really collecting all these new titles huh.
My future plans is to expand this into a website someday, however, right now, it has a nice place as a news aggregation site. Though, I need to up the filter's ante right now, due to the spamming done by other PASUM students.
Yesterday, my friend, Jing Xi and I were talking about some stuff related to making a website, and then we were talking about the Crime Uproot Ministry, or some other stupid stuff. The point is, I don't know which neurons fired, but as we were talking, I thought about a PASUM Website, designed to be an alternative to PASUM Confessions. Competing with PASUM Confessions isn't really my intent, but I wanted my new website to have a post and reply feature. Something like a social media.
After that, I was doing some market analysis. I wanted to figure out why Confessions was as successful as it is. I identified a couple of key traits:
- Confessions is incredibly accessible, it's just a click away from Telegram
- The users are anonymous. They can post the most abhorrent stuff, without any reprecussions
- There is no other alternatives that's better
However, even though there were many pros, there were also some cons.
- Confessions has a man-in-the-middle. A PASUM student has to moderate, and that becomes really tiring.
- There's often a very long delay with posting and seeing the change reflected in Confessions.
This was when I realised another one of my pain points in PASUM. It's incredibly annoying having to find information from the millions of scattered groups. Pelajar PASUM, Ketua Kelas, Kuaz. Thus, I decided to make PASUM Web 25/26. An aggregator for multiple different sources of official PASUM News.
I developed a backend aggregator for Mimi, and gave Mimi some features that allows for users to post on PASUM Web. The aggregator works like this. It utilizes a Python Library called Telethon, and it's connected directly into my account, such that the bot is able to read a scrape data from multiple different sources. After that, I upload the data that's scraped into Firebase (Database). From there, we use Mimi to do some extra formatting and stuff, give an ID for each post in order for others to reply, and there we have it, the current implementation of PASUM Web!
Also, to address the issue of moderation, Mimi's job is to analyze each messages and image, and then decide whether or not the post passes through her filter. Our aim is to be a more professional PASUM Confessions.
Obviously, Mimi isn't that good at aggregating yet, there's still numerous number of flaws that's present in how she does things. However, the basic MVP is done. Users can post, reply, read the latest news. That's honestly all that there is to it. The next step is to give it a final polish, some brand identity, and start marketing!
Jing Xi brought up a pretty good idea to promote the product to lecturers to increase the credibility of our PASUM Web.
I'm Happy with Mimi
I've been developing a Mimi tool these days, and she's coming out really effective at doing whatever it is that she does. I'm genuinely surprised at how much Mimi has impacted my daily life, considering she's surprisingly powerful.
Creating Mimi is the best decision I've ever made, and I will keep developing her, integrating her, and giving her more and more features! Until I get tired of it, of course.... Now I'm just padding this out to see if changes are made.
