June 2025 - YOCA Improvements, Chat Clone, Project Management, MVP Incoming
It's summer, and I'm still building lots of things. I just haven't spent much time documenting it here on the site. I started the month by building on my existing projects, but in the end, I went back to working on a project with a friend that should be able to turn into a business.
To start the month, I was working on adding or fixing things here on YOCA—mainly the Markdown WYSIWYG editor for posting days and rewinds. I also worked on project boards, but that got me thinking about Project Sweet Corn.
Next, I got interested in @theo on Twitter / YouTube, having a t3.chat cloneathon, so I built WebLuke Chat. This was a Blazor web app that utilized LM Studio's API to run LLMs locally on a computer, which I then hosted on a web page. This includes a complete message history and login. It all works pretty well, but because Theo didn't ever link the submission site, only showing screenshots of it in dev, I missed the submission window. So I have a self-hosted LLM site. This was beneficial because I gained more experience with Blazor and how it updates data for the user.
Then it was back to Project Sweet Corn, where we created data models, databases, and laid the groundwork for building the web app. I then put this on hold after discussing our business project with my friend.
The new project, which I don't want to advertise until it reaches an MVP state, combines several things I've been working on, including multi-tenant apps, database designs, login and authentication, a public frontend, multiple management backends, and more. So far, I have extracted parts from all my projects from Year of Code, but with new features and a much more complex database.
So, about six months into YOC, with a few hiccups, it has helped me get to a better place for building my projects, as I have gained more experience starting from nothing and moving forward. Using an LLM to bounce ideas off of or help solve problems has been a big game changer. I started the year by solving things the old way, through searches, reading, and failing. However, hitting up something like GPT-4.1 or Claude Sonnet 3/4 to give you a path was so much more helpful than I had expected.
I'm still going, but I do need to catch up on my daily posts. I am also working on a non-public project, so details are limited. But by building this, I hope it will turn into a cash-positive adventure so I can focus on it and my other project ideas.
Grammarly now works in the editor, and it helped make this post more readable and understandable. The suggestions are usually good, but they have improved since receiving some bad edits a few months ago. I just accepted most of them in this post. I hope it still flows correctly.