Yash Patel

Running AI models locally has become much easier over the past year. Whether it's for better privacy, faster responses, offline access, or avoiding recurring API costs, more people are setting up self-hosted LLMs for everyday productivity. Ollama has played a huge role in making that possible, and it's often the first tool people install. But once you spend more time with local AI, you realize there are plenty of other options that solve different problems just as well, or sometimes even better.

I've been self-hosting all kinds of services in my homelab for years. There's something satisfying about having complete control over your data, choosing your own software, and not relying on third-party providers. But over time, I realized that installing a service is the easy part — maintaining it is where the real commitment begins. Some services continue to justify that effort, while others slowly become another item on the maintenance checklist.

Obsidian has been the center of my note-taking and productivity workflow, but one thing was always missing: a proper way to organize notes like a database. I tried building my own system with Dataview, and while it worked, it took time to set up and wasn't always as intuitive as I wanted. I still relied on a separate database app to manage projects, article ideas, and trackers. That changed when Obsidian introduced Bases.