I am building a bit of that infrastructure for All About Berlin. There is a linter that regularly updates constants from information found on other websites. I want to keep expanding it to cover more facts, and include more elaborate source checking.
I’ve done this for my health records but my Claude web app (which I want to be able to access them) can’t read GitHub as effectively as it can a huge markdown file that just collects all materials into a single 15,000 line text file I reupload every week or so.
So even though I’ve done the whole wiki / knowledge base thing. The most efficient way of handing it over as a universal file to GPT and Claude for analysis.
Yeah I ended up just adding to claude.md for it to make a single file summary of the entire vault at any new update. It works fine.
But it would be nice to directly hook it up to a GitHub repo that I can commit and push to. I’m guessing MS won’t let them build that to safe guard their own models and OAI investiment.
At this point Claude so is far ahead in “empathetic understanding” and relational understanding, it’s painful to use GPT for anything but extremely factual content. IMO both Gemini and GPT are now overwhelmingly a failure as general purpose chatbots, while Claude is increasingly resembles what I imagined an AGI system to be.
13 comments
[ 2.7 ms ] story [ 33.3 ms ] thread1. forking the repo
2. committing the changes
3. submitting a pull request
... then you don't have a wiki.
https://en.wikipedia.org/wiki/Distributed_Artificial_Intelli...
Ideally something that doesn't require rescanning all the sources and something that can verify that the wiki is correct.
So even though I’ve done the whole wiki / knowledge base thing. The most efficient way of handing it over as a universal file to GPT and Claude for analysis.
Claude Code works though. But now I can’t share this with non-technical users
But it would be nice to directly hook it up to a GitHub repo that I can commit and push to. I’m guessing MS won’t let them build that to safe guard their own models and OAI investiment.
At this point Claude so is far ahead in “empathetic understanding” and relational understanding, it’s painful to use GPT for anything but extremely factual content. IMO both Gemini and GPT are now overwhelmingly a failure as general purpose chatbots, while Claude is increasingly resembles what I imagined an AGI system to be.