Learning to Configure Agentic AI Systems

Summary

We formulate agent configuration as a semi-Markov decision process in which each configuration acts as a temporally extended option, and introduce ARC, which dynamically selects query-specific agent configurations rather than fixing one pipeline in advance — improving average reasoning accuracy by 31.3% and tool-use accuracy by 13.95%.

BibTeX

@article{taparia2026learning,
    title={Learning to Configure Agentic AI Systems},
    author={Taparia, Aditya and Sagar, Som and Senanayake, Ransalu},
    journal={arXiv preprint arXiv:2602.11574},
    year={2026}
}