Apple is building a personal memory system that could help Siri retain information about users' lives, based on new code findings shared by code explorer pdfu . 3 beta release, the system includes a "memory graph" that's designed to organize information about relationships, places, routines, health, workouts, and what motivates users or holds them back, according to the findings. It also identifies "memory atoms" for facts, opinions, intentions, and events, including details about the people, places, times, and topics involved. According to pdfu, the graph uses two kinds of processes to gather info. One turns structured data into insights – like reading the name fields in a user's own contact card. The other uses an on-device AI model or Private Cloud Compute to extract insights by querying sources including Siri conversations, messages, photos, and health data. Apparently, only one of these AI-driven processes is present so far: pdfu believes it's designed to identify motivations, barriers, and resources using a set of 60 built-in questions, including questions about routines that might interfere with adequate sleep. From what's been uncovered so far, it's still an unfinished system – pdfu says there's no process that creates memory atoms yet, for example. In a follow-up explanation , the code sleuth also distinguished the graph from a separate store for explicit memories. The latter store has space for records in Apple's search system, but nothing writes the records yet. Both systems reportedly use Apple's version of the Kuzu graph database. Apple acquired Kuzu in 2025 , but its plans for the technology were unclear when the acquisition emerged earlier this year.