Multimodal Agent MemorySubmit a paper

EgoButler

EgoLife: Towards Egocentric Life Assistant

Jingkang Yang, Shuai Liu, Hongming Guo, Yuhao Dong, Xiamengwei Zhang, Sicheng Zhang, et al. · 2026-02-09

Read the paperPDF

Management
T1 Static Memory
Memory is built once and reused without change during deployment.
Primary modality
Video & streaming
Also: Audio & speech, text
Representation
Hybrid
retrieval-linked source evidence and records

How experience becomes memory

Each path is one memory component, traced from what is stored to how it is found again.

  1. retrieved egocentric video/audio evidence

    1. Represented asRetained perceptual source evidenceSource- and modality-specific
    2. Stored inVector-indexed collection
    3. Retrieved bySimilarity search
    4. Source evidenceSource directly available
    EgoRAG retrieves relevant ultra-long egocentric experience for life-oriented QA.Full paper: Model · PDF p. 7

  2. indexed egocentric experience records

    1. Represented asPerception-to-text records & summariesCross-represented explicit
    2. Stored inVector-indexed collection
    3. Retrieved bySimilarity search
    4. Source evidenceSource linked & recoverable
    Retrieved long-horizon records support practical recall and personalized assistance.Full-paper method / memory representation and retrieval · PDF p. 2

Related systems