HippoCamp: Benchmarking Contextual Agents on Personal Computers

2026-04-01

AI Summary

This work addresses the limitations of current AI agents in effectively managing multimodal files, performing context-aware reasoning, and supporting long-term retrieval within user-centric personal computing environments. To bridge this gap, the authors introduce the first benchmark tailored to real-world, personalized file systems, comprising 42.4 GB of data, over 2,000 multimodal files, and 581 context-aware question-answer pairs, accompanied by fine-grained, structured execution traces enabling precise failure attribution. Evaluation using this benchmark reveals that state-of-the-art multimodal large language models achieve only 48.3% accuracy on user profiling tasks, exposing significant bottlenecks in cross-modal reasoning and grounding evidence from heterogeneous sources.

Abstract

We present HippoCamp, a new benchmark designed to evaluate agents' capabilities on multimodal file management. Unlike existing agent benchmarks that focus on tasks like web interaction, tool use, or software automation in generic settings, HippoCamp evaluates agents in user-centric environments to model individual user profiles and search massive personal files for context-aware reasoning. Our benchmark instantiates device-scale file systems over real-world profiles spanning diverse modalities, comprising 42.4 GB of data across over 2K real-world files. Building upon the raw files, we construct 581 QA pairs to assess agents' capabilities in search, evidence perception, and multi-step reasoning. To facilitate fine-grained analysis, we provide 46.1K densely annotated structured trajectories for step-wise failure diagnosis. We evaluate a wide range of state-of-the-art multimodal large language models (MLLMs) and agentic methods on HippoCamp. Our comprehensive experiments reveal a significant performance gap: even the most advanced commercial models achieve only 48.3% accuracy in user profiling, struggling particularly with long-horizon retrieval and cross-modal reasoning within dense personal file systems. Furthermore, our step-wise failure diagnosis identifies multimodal perception and evidence grounding as the primary bottlenecks. Ultimately, HippoCamp exposes the critical limitations of current agents in realistic, user-centric environments and provides a robust foundation for developing next-generation personal AI assistants.

Problem

• contextual agents
• personal file systems
• multimodal reasoning
• user profiling
• cross-modal retrieval

Innovation

• contextual agents
• multimodal file management
• personal AI benchmark
• step-wise failure diagnosis
• cross-modal reasoning