FileGram: Grounding Agent Personalization in File-System Behavioral Traces

TL;DR

FileGram introduces a new framework for personalizing AI agents using detailed file-system behavioral traces, addressing privacy and data collection challenges in human-AI interaction.

Contribution

The paper presents FileGram, a comprehensive system with a data engine, benchmark, and memory architecture for scalable, privacy-preserving agent personalization based on file-system behaviors.

Findings

FileGramBench is challenging for current memory systems.

FileGramEngine effectively simulates realistic workflows.

FileGramOS accurately builds user profiles from atomic actions.

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 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.