Training Proactive and Personalized LLM Agents
This work addresses the need for more user-centered AI agents in practical applications, representing a novel method for a known bottleneck rather than a foundational shift.
The paper tackles the problem of building effective real-world LLM agents by optimizing productivity, proactivity, and personalization, introducing a multi-objective reinforcement learning approach that achieves a 21.6 average improvement over GPT-5 on tasks like software engineering and deep research.
While existing work focuses primarily on task success, we argue that effective real-world agents require optimizing three dimensions: productivity (task completion), proactivity (asking essential questions), and personalization (adapting to diverse user preferences). We introduce UserVille, an interactive environment with LLM-based user simulators enabling diverse, configurable user preferences. Leveraging UserVille, we introduce PPP, a multi-objective reinforcement learning approach that jointly optimizes all three dimensions: Productivity, Proactivity, and Personalization. Experiments on software engineering and deep research tasks show that agents trained with PPP achieve substantial improvements over strong baselines such as GPT-5 (+21.6 on average), demonstrating the ability to ask strategic clarifying questions, adapt to unseen user preferences, and improve task success through better interaction. This work demonstrates that explicitly optimizing for user-centered interaction is critical for building practical and effective AI agents.