There are many digital twin companies that scrape public data to build models of individuals. But I want to include private data because I’m not famous so there is little content of me in the internet. I have exported my entire notion content and chatgpt history into a folder and I find that the cursor agent is able to surface interesting insights, i.e. the data is not just slob. I can’t just upload this data to a random company because it includes sensitive stuff (salary, insecurities, secrets, etc), yet I still want to provide digital twin companies with part of it because it is necessary to get a good representation of me. It feels like this problem needs some level of indirection.

Has anyone seen projects tackling this? Most projects I found are pre-LLM era and not that helpful tbh. I assume this has to be an open-source project where people could self-host due to how sensitive some data is, but I might be wrong.

I’ll be around for a few hours to discuss.