AOrchestra: Automating Sub-Agent Creation for Agentic Orchestration

Jianhao Ruan, Zhihao Xu, Yiran Peng, Fashen Ren, Zhaoyang Yu, Xinbing Liang, Jinyu Xiang, Yongru Chen, Bang Liu, Chenglin Wu, Yuyu Luo, Jiayi Zhang
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:106017-106040, 2026.

Abstract

Language agents have shown strong promise for task automation. Realizing this promise for increasingly complex, long-horizon tasks has driven the rise of a subagent-as-tools paradigm for multi-turn task solving. However, existing designs still lack a dynamic abstraction view of sub-agents, thereby hurting adaptability: sub-agents are either context-isolated threads that lack specialization, or static roles that require human-engineering. We address this challenge with a unified, framework-agnostic agent abstraction that models any agent as a tuple (Model, Task, Tools, Context). This tuple acts as a compositional recipe for capabilities, enabling the system to spawn specialized executors for each task on demand. Building on this abstraction, we introduce an agentic system AOrchestra, where the central orchestrator concretizes the tuple at each step: it curates task-relevant context, selects tools and models, and delegates execution via on-the-fly automatic agent creation. Such designs enable reducing human engineering efforts, and remain framework-agnostic with plug-and-play support for diverse agents as task executors. It also enables a controllable performance–cost trade-off, allowing the system to approach Pareto-efficient. Across three challenging benchmarks and environments (GAIA, SWE-Bench, Terminal-Bench), AOrchestra achieves 16.28% relative improvement against the strongest baseline when paired with Gemini-3-Flash.

Cite this Paper


BibTeX
@InProceedings{pmlr-v306-ruan26e, title = {{AO}rchestra: Automating Sub-Agent Creation for Agentic Orchestration}, author = {Ruan, Jianhao and Xu, Zhihao and Peng, Yiran and Ren, Fashen and Yu, Zhaoyang and Liang, Xinbing and Xiang, Jinyu and Chen, Yongru and Liu, Bang and Wu, Chenglin and Luo, Yuyu and Zhang, Jiayi}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {106017--106040}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/ruan26e/ruan26e.pdf}, url = {https://proceedings.mlr.press/v306/ruan26e.html}, abstract = {Language agents have shown strong promise for task automation. Realizing this promise for increasingly complex, long-horizon tasks has driven the rise of a subagent-as-tools paradigm for multi-turn task solving. However, existing designs still lack a dynamic abstraction view of sub-agents, thereby hurting adaptability: sub-agents are either context-isolated threads that lack specialization, or static roles that require human-engineering. We address this challenge with a unified, framework-agnostic agent abstraction that models any agent as a tuple (Model, Task, Tools, Context). This tuple acts as a compositional recipe for capabilities, enabling the system to spawn specialized executors for each task on demand. Building on this abstraction, we introduce an agentic system AOrchestra, where the central orchestrator concretizes the tuple at each step: it curates task-relevant context, selects tools and models, and delegates execution via on-the-fly automatic agent creation. Such designs enable reducing human engineering efforts, and remain framework-agnostic with plug-and-play support for diverse agents as task executors. It also enables a controllable performance–cost trade-off, allowing the system to approach Pareto-efficient. Across three challenging benchmarks and environments (GAIA, SWE-Bench, Terminal-Bench), AOrchestra achieves 16.28% relative improvement against the strongest baseline when paired with Gemini-3-Flash.} }
Endnote
%0 Conference Paper %T AOrchestra: Automating Sub-Agent Creation for Agentic Orchestration %A Jianhao Ruan %A Zhihao Xu %A Yiran Peng %A Fashen Ren %A Zhaoyang Yu %A Xinbing Liang %A Jinyu Xiang %A Yongru Chen %A Bang Liu %A Chenglin Wu %A Yuyu Luo %A Jiayi Zhang %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-ruan26e %I PMLR %P 106017--106040 %U https://proceedings.mlr.press/v306/ruan26e.html %V 306 %X Language agents have shown strong promise for task automation. Realizing this promise for increasingly complex, long-horizon tasks has driven the rise of a subagent-as-tools paradigm for multi-turn task solving. However, existing designs still lack a dynamic abstraction view of sub-agents, thereby hurting adaptability: sub-agents are either context-isolated threads that lack specialization, or static roles that require human-engineering. We address this challenge with a unified, framework-agnostic agent abstraction that models any agent as a tuple (Model, Task, Tools, Context). This tuple acts as a compositional recipe for capabilities, enabling the system to spawn specialized executors for each task on demand. Building on this abstraction, we introduce an agentic system AOrchestra, where the central orchestrator concretizes the tuple at each step: it curates task-relevant context, selects tools and models, and delegates execution via on-the-fly automatic agent creation. Such designs enable reducing human engineering efforts, and remain framework-agnostic with plug-and-play support for diverse agents as task executors. It also enables a controllable performance–cost trade-off, allowing the system to approach Pareto-efficient. Across three challenging benchmarks and environments (GAIA, SWE-Bench, Terminal-Bench), AOrchestra achieves 16.28% relative improvement against the strongest baseline when paired with Gemini-3-Flash.
APA
Ruan, J., Xu, Z., Peng, Y., Ren, F., Yu, Z., Liang, X., Xiang, J., Chen, Y., Liu, B., Wu, C., Luo, Y. & Zhang, J.. (2026). AOrchestra: Automating Sub-Agent Creation for Agentic Orchestration. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:106017-106040 Available from https://proceedings.mlr.press/v306/ruan26e.html.

Related Material