Continued experiments with an autonomous CRM that slowly creates a massive knowledge graph. This time, I'm testing specific edge types, which loses flexibility, but also makes the output easier to query and understand.* *I'm using function call to get JSON, and the "enum" feature to specify edge types

Found and fixed an error! Starting to work more consistently with complicated inputs.

Not perfect, but it’s getting somewhere! Pictured: PayPal mafia and related cos (By the way, you can make your own at

How it’s working:

combining the wikipedia entries of OpenAI cofounders starts to look like this

hahahaha let's keep going

phew, took longer than i'd like to admit - but finally got this to be node type agnostic previously it was set to three node types (people, orgs, events), but now I can flexibly create new node types pictured: first few pages of langchain documentation (blue highlight showcases 'search' feature)

Now that the day is over, back to feeding in Wikipedia articles (mapping NVIDIA founders here) Next, going to set it up so I can try feeding in a csv of URLs

huzzah, got csv upload working! uploaded a few wikipedia articles via csv and just watched the knowledge graph grow. (15 min > 30 sec)

hindu dieties

okay, so for testing purposes i made a "latent input" endpoint that generates the graph based on a simple input (the core architecture is designed for easy "extensions", so this one just feeds an openai output into the same endpoint that receives scraped web content from the url_input endpoint)