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

21–30 of 73 posts

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

#21

Post by dariusmonsef »

0x457 wrote: post #20
dariusmonsef wrote: post #3

Earlier quoted context omitted.

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?

Yeah, close. But all handled so a user can just focus on connecting the MCP and doing work. Goes to my answer elsewhere here about different between gbrain / llm-wiki… those are highly configurable and powerful, but require technical knowledge and time. I think the vast majority of tech workers will need “that for dummies” ie this just works for them.

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

#25

Post by freemindcore »

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.

[flagged]

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

#28

Post by sinuhe69 »

I think the central question for a such memory system is whether we or the agents can find the relevant information and how to organize these data as changes continues to come in. Would we miss something in the retrieval process? How do we organize the information so they stay actual and correct without piling up the garbage? Of course we can continue to concatenate the data and tag them with version and date, but then we have to face the problem of extracting the relevant information in a short time. If we delegate that problem to a LLM, long context retrieval performance will degrade and the cost will explode.

That is the reason why we condense the information in the first place. Forgetting + Synthesizing are the necessary parts of learning and basically with memory + smart retrieval we want to build a learning system.

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

#29

Post by 8lamaster8 »

sinuhe69 wrote: post #28

I think the central question for a such memory system is whether we or the agents can find the relevant information and how to organize these data as changes continues to come in. Would we miss something in the retrieval process? How do we organize the information so they stay actual and correct without piling up the garbage? Of course we can continue to concatenate the data and tag them with version and date, but th…

[flagged]
Post reply on HN