PASK: Toward Intent-Aware Proactive Agents with Long-Term Memory

TL;DR

This paper introduces PASK, a proactive AI agent framework capable of inferring user needs and grounding actions in long-term memory, tested on a new real-world benchmark with promising results.

Contribution

It proposes a novel paradigm DD-MM-PAS for streaming proactive AI, with a new IntentFlow model and a real-world benchmark for evaluating intent inference.

Findings

IntentFlow matches Gemini3-Flash under latency constraints.

The system effectively infers deeper user intent.

The benchmark enables realistic evaluation of proactive agents.

Abstract

Proactivity is a core expectation for AGI. Prior work remains largely confined to laboratory settings, leaving a clear gap in real-world proactive agent: depth, complexity, ambiguity, precision and real-time constraints. We study this setting, where useful intervention requires inferring latent needs from ongoing context and grounding actions in evolving user memory under latency and long-horizon constraints. We first propose DD-MM-PAS (Demand Detection, Memory Modeling, Proactive Agent System) as a general paradigm for streaming proactive AI agent. We instantiate this paradigm in Pask, with streaming IntentFlow model for DD, a hybrid memory (workspace, user, global) for long-term MM, PAS infra framework and introduce how these components form a closed loop. We also introduce LatentNeeds-Bench, a real-world benchmark built from user-consented data and refined through thousands of rounds of human editing. Experiments show that IntentFlow matches leading Gemini3-Flash models under latency constraints, while identifying deeper user intent.