Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Show HN: OzBrain, a shared brain for knowledge between agents and your team

ozbrain.com

Post reply on HN
Print view View original

11–20 of 73 posts

Re: Show HN: OzBrain, a shared brain for knowledge between agents and your team

#11

Post by panarky »

dariusmonsef wrote: post #6
gosolozero wrote: post #5

I’m not sure I get the value over Obsidian, can you explain the static file issue? Only happens at 7K+ individual files? What happens if you always conjoin files?

Obsidian would fall into the same answer as how is this different than gBrain. Obsidian is a powerful and configurable... but requires more work to maintain and sharing knowledge with others is more difficult. I want this to be a super easy way for agents to connect to my knowledge, and make it easy for me to share my chunks of knowledge with my teammates and their agents. Just MCP in and have your agents get to work…

I'd say gBrain is also more structured to the priorities of a VC and less of a general-purpose graph-based memory. There are tons of hard-coded regexes that may work beautifully for Garry's priorities and workflow that don't fit other use cases, and are probably really brittle even if you're a VC who uses different words in your notes.

Re: Show HN: OzBrain, a shared brain for knowledge between agents and your team

#14

Post by indigodaddy »

This is interesting.

Would you consider this a different type product/benefit than all the "memory" things we have seen popping up everywhere?. Is it different just because it lives in the cloud? To me it feels like a different thing than memory.

Re: Show HN: OzBrain, a shared brain for knowledge between agents and your team

#15

Post by cindyllm »

embedding-shape wrote: post #10
gavinboston wrote: post #7

Do you have a solution for degradation in accuracy when compiling larger amounts of llm-produced text? I am also building LLM knowledge/memory systems and I've been surprised how bad LLMs are, even SOTA models, at summarizing non-trivial input batches of text. They get things wrong, distort the underlying meaning or data, etc.

Divide and conquer essentially, is what I've found so far to work best. Split things into smaller and smaller chunks to independently be verified, double-check everything, then coalesce upwards with verified summarizations. Have benchmarks for every single task and sub-task that will happen everywhere a LLM is involved, so you can measure improvements. Takes a ton more effort and tokens in the system itself obviously…

[dead]

Re: Show HN: OzBrain, a shared brain for knowledge between agents and your team

#17

Post by dariusmonsef »

indigodaddy wrote: post #14

This is interesting. Would you consider this a different type product/benefit than all the "memory" things we have seen popping up everywhere?. Is it different just because it lives in the cloud? To me it feels like a different thing than memory.

I think they’re similar but different. Memory is usually single user focused, a summary of specific facts or instructions. OzBrain is everything that was captured, reasoned, promoted, etc. and auditable record of what the latest thinking is and why. When agents are the primary user you need a way to capture why they ended up at where they ended up. Memory is more like the filtered summary of everything it has saved as important.

Re: Show HN: OzBrain, a shared brain for knowledge between agents and your team

#19

Post by rendonroman »

It’s very encouraging to see serious attempts at addressing continuity between agents’ outputs. Right now, everyone seems to be figuring out their own way of maintaining consistency across sessions without endlessly over-contextualizing each new one.

This feels like an important layer of the emerging agent stack, and I think this Show HN will be useful to a lot of people working through exactly that problem.

Re: Show HN: OzBrain, a shared brain for knowledge between agents and your team

#20

Post by 0x457 »

dariusmonsef wrote: post #3
travisueki wrote: post #2

So is this cloud sync for my Md files? Who pays for the diffing and versioning?

In a way yes. A hosted llm-wiki, where I handle the diffing, versioning and audit log of what was changed, by what agent and why.

So a git repo + md files?
Post reply on HN