Road to Cerebro 1.0
Okay, new goal to achieve by the end of March: Release Cerebro 1.0.
For the uninitiated, Cerebro is an Obsidian plugin that I have been building that enables you to chat with an LLM right from Obsidian.
The project has been in limbo until an idea sprouted in my head: What if I aimed for Cerebro winning Obsidian’s 2025 Gems of the Year?
Fun, ambitious, super challenging. Looking around the Plugin Store, there are already so many available (I count 60+ plugins related to “LLM”/”GPT”). What sets Cerebro apart from any of these?
In fact, Cerebro today is lacking in feature parity: it doesn’t have OpenAI image/PDF support, it can’t support local models, it can’t query across your whole vault. Do I have the time or capability to add those? Probably. Will doing that win me a Gem of the Year award? Absolutely not.
There’s also the question of pursuing this based on extrinsic motivation alone. Winning an award sounds cool but it’s not the full story.
To have a real shot at building a quality product, I have to clarify why I’m really doing this, then figure out what would make Cerebro stand out from the sea of other plugins.
First, why am I doing this? In rough order of importance:
- I genuinely love building Cerebro, something I haven’t felt in a while with other projects outside of work.
- I’ve been on a businessmaxx grind recently. Doing this is very similar to releasing a product if you think about it. Create something of value, and distribute it to interested people for a fair price. That’s all business is. In this project are invaluable lessons of producing a plugin people want to use, and garnering enough attention from people on the Internet. Just ignore the $0 price tag for now.
- This is a great reason to keep up with advancements in AI. And no, I don’t want to just make an LLM wrapper.
Great set of reasons. Now, what would set Cerebro apart?
Cerebro must be the most beautiful and intuitive interface for interacting with LLMs. Not just across all Obsidian plugins, but of all AI tools today.
Not the most powerful, or the most feature-packed. Just the most intuitive. Where Obsidian Copilot is Android, Cerebro should be the iPhone.
When you download Cerebro, chatting should be easy and intuitive. It should have (virtually) no bugs. It should show underlying steps (thinking, searching) from compatible models. It should let you glide across your mind and the entire Internet with ease. Exactly what features would give you this feeling would be a separate post.
The secondary characteristic is that Cerebro should be Obsidian-first, and that means to be aligned with the set of values that Obsidian describes itself to be. This means that it should be: free (at least for the base functionality); durable (conversations are Markdown-native, if embeddings are added it’ll be open file formats); private (support for local models, files should be off-limits unless given permission); malleable (this may be challenging with the Apple feeling we’re going for) and independent (open to the community, built in public, open source).