TIDE: Proactive Multi-Problem Discovery via Template-Guided Iteration

2026-06-03

AI Summary

This work addresses scenarios where multiple user issues coexist without explicit indication, proposing the TIDE framework—a template-guided iterative approach that systematically uncovers hidden problems. TIDE operates in rounds, each focusing on a small subset of issues, progressively expanding conditionally based on previously identified ones. By anchoring reasoning to predefined problem categories through reusable thought templates and integrating context-aware classification with evidence localization, it effectively mitigates biases inherent in single-pass prediction. Employing a multi-agent collaborative architecture, TIDE consistently outperforms both single-pass and parallel multi-agent baselines across four model backbones in personal workspace and software repository tasks, achieving significant improvements in issue coverage, identification accuracy, and solution effectiveness.

Abstract

Agents are widely deployed as assistants over documents, tools, and code. However, they typically act only on explicit user requests, which surface only the problems the user has noticed, while many other important problems coexist, hidden in plain sight, within the broader user context, with their total number unknown in advance. We frame this as the task of discovering multiple hidden problems from context, in which coexisting problems should be uncovered, grounded in supporting evidence, and paired with concrete actions. To this end, we introduce TIDE, a template-guided iterative framework with two complementary mechanisms. Specifically, motivated by the observation that single-pass prediction anchors on the most salient cases and yields generic claims, we propose iterative discovery, which surfaces a small batch of candidates per round while conditioning on what has already been found, so subsequent rounds extend coverage; and thought templates, reusable schemas distilled from previously solved cases that specify what contextual signals to attend to and how to connect them, anchoring each prediction in a recognizable problem class. We validate TIDE on two realistic settings, personal workspaces and software repositories, across four model backbones, showing substantial gains over single-shot and parallel multi-agent baselines on task coverage, identification, and resolution.

Problem

• multi-problem discovery
• hidden problems
• contextual understanding
• proactive agent
• problem identification

Innovation

• iterative discovery
• thought templates
• multi-problem discovery
• template-guided framework
• contextual problem identification