Learning to Configure Agentic AI Systems

Configuring LLM-based agent systems involves choosing workflows, tools, token budgets, and prompts from a large combinatorial design space, and is typically handled today by fixed large templates or hand-tuned heuristics. This leads to brittle behavior and unnecessary compute, since the same cumbersome configuration is often applied to both easy and hard input queries. We formulate agent configuration as a query-wise decision problem and introduce ARC (Agentic Resource & Configuration learner), which learns a light-weight hierarchical policy using reinforcement learning to dynamically tailor these configurations. Across multiple benchmarks spanning reasoning and tool-augmented question answering, the learned policy consistently outperforms strong hand-designed and other baselines, achieving up to 25% higher task accuracy while also reducing token and runtime costs. These results demonstrate that learning per-query agent configurations is a powerful alternative to "one size fits all" designs.

• 1

  ARC learns query-specific configurations for AI agents, achieving 25% higher accuracy while reducing token usage and costs.

• 2

  Hierarchical reinforcement learning decomposes configuration into manageable structure and prompt policies rather than joint optimization.

• 3

  Semantic embeddings combined with hand-crafted features enable systems to match appropriate workflows to different question types.