I was drowning in 100+ messages per-day at work and existing AI tools just wouldn't cut it because of default sorts (e.g. recency, etc.), not by what actually matters to me.
I tried modeling it to think like me when responding to messages: belief graph with decay, trust scoring from behavior, and graduated autonomy to prevent AI from mis-categorizing my messages. Went down the path of open source so everyone can try it with a provided demo data in the repo. I'm just looking for feedback on the architecture, especially the belief decay and graduated autonomy concepts.
Nice approach, the Rust performance and single-binary deployment are compelling. Question, how do you handle contradictory facts? If John moves from Stripe to Google, does the graph resolve that, or does it store both?
I tried modeling it to think like me when responding to messages: belief graph with decay, trust scoring from behavior, and graduated autonomy to prevent AI from mis-categorizing my messages. Went down the path of open source so everyone can try it with a provided demo data in the repo. I'm just looking for feedback on the architecture, especially the belief decay and graduated autonomy concepts.
Thanks