Paper Detail
Shuai Liu, Shulin Tian, Kairui Hu, Yuhao Dong, Zhe Yang, Bo Li, Jingkang Yang, Chen Change Loy, Ziwei Liu
Coworking AI agents operating within local file systems are rapidly emerging as a paradigm in human-AI interaction; however, effective personalization remains limited by severe data constraints, as strict privacy barriers and the difficulty of jointly collecting multimodal real-world traces prevent scalable training and evaluation, and existing methods remain interaction-centric while overlooking dense behavioral traces in file-system operations; to address this gap, we propose FileGram, a comprehensive framework that grounds agent memory and personalization in file-system behavioral traces, comprising three core components: (1) FileGramEngine, a scalable persona-driven data engine that simulates realistic workflows and generates fine-grained multimodal action sequences at scale; (2) FileGramBench, a diagnostic benchmark grounded in file-system behavioral traces for evaluating memory systems on profile reconstruction, trace disentanglement, persona drift detection, and multimodal grounding; and (3) FileGramOS, a bottom-up memory architecture that builds user profiles directly from atomic actions and content deltas rather than dialogue summaries, encoding these traces into procedural, semantic, and episodic channels with query-time abstraction; extensive experiments show that FileGramBench remains challenging for state-of-the-art memory systems and that FileGramEngine and FileGramOS are effective, and by open-sourcing the framework, we hope to support future research on personalized memory-centric file-system agents.
No structured notes yet. Add `summary_sections`, `why_relevant`, `claim_impact`, or `next_action` in `papers.jsonl` to enrich this view.
No ranking explanation is available yet.
No tags.
@article{liu2026filegram,
title = {FileGram: Grounding Agent Personalization in File-System Behavioral Traces},
author = {Shuai Liu and Shulin Tian and Kairui Hu and Yuhao Dong and Zhe Yang and Bo Li and Jingkang Yang and Chen Change Loy and Ziwei Liu},
year = {2026},
abstract = {Coworking AI agents operating within local file systems are rapidly emerging as a paradigm in human-AI interaction; however, effective personalization remains limited by severe data constraints, as strict privacy barriers and the difficulty of jointly collecting multimodal real-world traces prevent scalable training and evaluation, and existing methods remain interaction-centric while overlooking dense behavioral traces in file-system operations; to address this gap, we propose FileGram, a compr},
url = {https://arxiv.org/abs/2604.04901},
keywords = {cs.CV, cs.AI, file-system behavioral traces, persona-driven data engine, multimodal action sequences, diagnostic benchmark, memory systems, profile reconstruction, trace disentanglement, persona drift detection, multimodal grounding, bottom-up memory architecture, atomic actions, content deltas, procedural channels, semantic channels, episodic channels, query-time abstraction, huggingface daily},
eprint = {2604.04901},
archiveprefix = {arXiv},
}
{}