As the next step in building out a privacy-first AI assistant, I am adding client-side persistence to chats. These will not be synced right now, although I might add the option to enable syncing later using the same client-side encryption that is used for docs. This obviously doesn't side-step the issue of sending data to the third-party provider (OpenAI), but that is out of my control and is up to the end user to decide whether they are okay with that or not. Eventually, I hope that OSS models will become a viable alternative, and then users could just self-host their own models for full privacy.
As the next step in building out a privacy-first AI assistant, I am adding client-side persistence to chats. These will not be synced right now, although I might add the option to enable syncing later using the same client-side encryption that is used for docs. This obviously doesn't side-step the issue of sending data to the third-party provider (OpenAI), but that is out of my control and is up to the end user to decide whether they are okay with that or not. Eventually, I hope that OSS models will become a viable alternative, and then users could just self-host their own models for full privacy.