Inwit builds on the story you've told (values, avoidances, decisions), aiming to answer more precisely in your own words as it accumulates. So it stores your conversations and the self-model (values, dilemmas, decisions) extracted from them.
Data is stored in Supabase, isolated so each user can access only their own data (RLS). No other user can see your conversations.
To process conversations and build your self-model, content is sent to the APIs of Anthropic (conversation generation) and OpenAI (embeddings). Neither uses API-submitted data to train their models.
As a rule, no. We don't read individual conversations, and analysis for service improvement is only done in aggregated, anonymized form. Only in cases of an abuse report or a legal request may we review them in a limited way.
Yes. Just reach out anytime at yswyang@gmail.com and we'll take care of it.
Inwit is currently an alpha-stage personal project. This policy may be updated as the service evolves.