pull down to refresh

Sovereign Brain is a simple blueprint for a local AI second brain that runs on your machine.

It holds your projects, TODOs, goals, thoughts, and learnings.

It pushes back on your ideas, helps you see things from new angles, and keeps you working on what matters.

2026 is the year of sovereign AI. Local AI is now fast and smart enough to run on a laptop.

It won't beat the cloud. But it's the only AI you'll share your personal thoughts with.

All local and open source: LLM + Pi Agent + Markdown.

For me this is a huge unlock: I wasn't comfortable to share this with Claude or ChatGPT. So I thought, maybe other bitcoiners are also interested and surprised that it actually works.

Watch a demo + Create your own at https://sovereignbrain.me

57 sats \ 7 replies \ @sime 1 Jun

These projects warm my heart.

More local LLMs please.

reply

Do you use one? any favorite local model that I could compare against Qwen3.6?

reply
local model that I could compare against Qwen3.6

To compare against Qwen3.6 35B? Gemma4 31B.

reply

I unfortunately get 3TPS on Gemma4 31B. I tried the 26B A4B version, which I get comparable speeds with Qwen3.6 35B A3B (20-30tps).

It seems to initially look ok, but skill use is significantly worse though from my initial tests (guessing rather than properly using the skill and understanding it - deleting text in a file that explicitly said to not delete it, and not following instructions and reading things fully).

So for me, Qwen3.6 is still a massive improvement - a game changer, which has made this project possible.

reply

Interesting.

I had problems with Qwen3.6 35B through my opencode integration (that I use in production with GLM-5) where it had instruction separation issues (i.e. it had trouble distinguishing between the files it read and the instruction given, I recorded an instance of that here: #1483366) - also happens on things like diff analysis often.

Perhaps we ought to tune more on a per-model-family basis? Not sure.

(edit: forgot to mention that it doesn't always finish the task and just quits. But I have the same issue with Gemma)

reply
105 sats \ 1 reply \ @rolznz OP 11h

Interesting that we got basically the opposite results :-)

I do see a LOT of errors, yeah. But from my observations it's very good at self-correcting and fixing its mistakes.

But this project is quite simple - not much specific technical knowledge needed. Maybe this is where the difference comes from?

I also saw you used Claude Code and OpenCode. They both use a large system prompt. Did you try Pi Agent?

reply
456 sats \ 0 replies \ @optimism 10h
I do see a LOT of errors, yeah. But from my observations it's very good at self-correcting and fixing its mistakes.

This I see too (it's better trained at this than Gemma) and I do think the self-correction is working more often than not (though what a waste of compute!) grep -i wait on thinking blocks is still "fun" too.

But this project is quite simple - not much specific technical knowledge needed.

My main use case nowadays is feeding LLMs file diffs and strace logs, mostly of third party code from npm/cargo/pubdev/mvn/pypi, to help me make security assessments. Thanks to LLMs this now only takes me a day a week instead of 4 last year, with about 5x the workload and an 100x threat increase.

Did you try Pi Agent?

Nope. The problem I'm running into is that I am swamped, but I'll try to find a moment to test your setup.

reply
15 sats \ 0 replies \ @sime 10h

No, not using any. My brief interaction was they needed internet access.

I'm missing use cases honestly.

reply
15 sats \ 0 replies \ @BITC0IN 22h

cool

reply

Very interesting projects