Towards Enforcing Company Policy Adherence in Agentic Workflows

Naama Zwerdling; David Boaz; Ella Rabinovich; Guy Uziel; David Amid; Ateret Anaby-Tavor

doi:10.18653/v1/2025.emnlp-industry.41

Towards Enforcing Company Policy Adherence in Agentic Workflows

Naama Zwerdling
, David Boaz
, Ella Rabinovich
, Guy Uziel
, David Amid
, Ateret Anaby-Tavor

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

Large Language Model (LLM) agents hold promise for a flexible and scalable alternative to traditional business process automation, but struggle to reliably follow complex company policies. In this study we introduce a deterministic, transparent, and modular framework for enforcing business policy adherence in agentic workflows. Our method operates in two phases: (1) an offline buildtime stage that compiles policy documents into verifiable guard code associated with tool use, and (2) a runtime integration where these guards ensure compliance before each agent action. We demonstrate our approach on the challenging τ-bench Airlines domain, showing encouraging preliminary results in policy enforcement, and further outline key challenges for real-world deployments.

Original language	English
Title of host publication	EMNLP 2025 - 2025 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Industry Track
Editors	Saloni Potdar, Lina Rojas-Barahona, Sebastien Montella
Publisher	Association for Computational Linguistics (ACL)
Pages	595-606
Number of pages	12
ISBN (Electronic)	9798891763333
DOIs	https://doi.org/10.18653/v1/2025.emnlp-industry.41
State	Published - 2025
Externally published	Yes
Event	2025 Conference on Empirical Methods in Natural Language Processing: Industry Track, EMNLP 2025 - Suzhou, China Duration: 4 Nov 2025 → 9 Nov 2025

Publication series

Name	EMNLP 2025 - 2025 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Industry Track

Conference

Conference	2025 Conference on Empirical Methods in Natural Language Processing: Industry Track, EMNLP 2025
Country/Territory	China
City	Suzhou
Period	4/11/25 → 9/11/25

Bibliographical note

Publisher Copyright:
© 2025 Association for Computational Linguistics.

ASJC Scopus subject areas

Computational Theory and Mathematics
Computer Science Applications
Information Systems
Linguistics and Language

Access to Document

10.18653/v1/2025.emnlp-industry.41

Cite this

Zwerdling, N., Boaz, D., Rabinovich, E., Uziel, G., Amid, D., & Anaby-Tavor, A. (2025). Towards Enforcing Company Policy Adherence in Agentic Workflows. In S. Potdar, L. Rojas-Barahona, & S. Montella (Eds.), EMNLP 2025 - 2025 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Industry Track (pp. 595-606). (EMNLP 2025 - 2025 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Industry Track). Association for Computational Linguistics (ACL). https://doi.org/10.18653/v1/2025.emnlp-industry.41

Zwerdling, Naama ; Boaz, David ; Rabinovich, Ella et al. / Towards Enforcing Company Policy Adherence in Agentic Workflows. EMNLP 2025 - 2025 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Industry Track. editor / Saloni Potdar ; Lina Rojas-Barahona ; Sebastien Montella. Association for Computational Linguistics (ACL), 2025. pp. 595-606 (EMNLP 2025 - 2025 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Industry Track).

@inproceedings{b279c90aee334381b898a14f7ad35d56,

title = "Towards Enforcing Company Policy Adherence in Agentic Workflows",

abstract = "Large Language Model (LLM) agents hold promise for a flexible and scalable alternative to traditional business process automation, but struggle to reliably follow complex company policies. In this study we introduce a deterministic, transparent, and modular framework for enforcing business policy adherence in agentic workflows. Our method operates in two phases: (1) an offline buildtime stage that compiles policy documents into verifiable guard code associated with tool use, and (2) a runtime integration where these guards ensure compliance before each agent action. We demonstrate our approach on the challenging τ-bench Airlines domain, showing encouraging preliminary results in policy enforcement, and further outline key challenges for real-world deployments.",

author = "Naama Zwerdling and David Boaz and Ella Rabinovich and Guy Uziel and David Amid and Ateret Anaby-Tavor",

note = "Publisher Copyright: {\textcopyright} 2025 Association for Computational Linguistics.; 2025 Conference on Empirical Methods in Natural Language Processing: Industry Track, EMNLP 2025 ; Conference date: 04-11-2025 Through 09-11-2025",

year = "2025",

doi = "10.18653/v1/2025.emnlp-industry.41",

language = "English",

series = "EMNLP 2025 - 2025 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Industry Track",

publisher = "Association for Computational Linguistics (ACL)",

pages = "595--606",

editor = "Saloni Potdar and Lina Rojas-Barahona and Sebastien Montella",

booktitle = "EMNLP 2025 - 2025 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Industry Track",

address = "United States",

}

Zwerdling, N, Boaz, D, Rabinovich, E, Uziel, G, Amid, D & Anaby-Tavor, A 2025, Towards Enforcing Company Policy Adherence in Agentic Workflows. in S Potdar, L Rojas-Barahona & S Montella (eds), EMNLP 2025 - 2025 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Industry Track. EMNLP 2025 - 2025 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Industry Track, Association for Computational Linguistics (ACL), pp. 595-606, 2025 Conference on Empirical Methods in Natural Language Processing: Industry Track, EMNLP 2025, Suzhou, China, 4/11/25. https://doi.org/10.18653/v1/2025.emnlp-industry.41

Towards Enforcing Company Policy Adherence in Agentic Workflows. / Zwerdling, Naama; Boaz, David; Rabinovich, Ella et al.
EMNLP 2025 - 2025 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Industry Track. ed. / Saloni Potdar; Lina Rojas-Barahona; Sebastien Montella. Association for Computational Linguistics (ACL), 2025. p. 595-606 (EMNLP 2025 - 2025 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Industry Track).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Towards Enforcing Company Policy Adherence in Agentic Workflows

AU - Zwerdling, Naama

AU - Boaz, David

AU - Rabinovich, Ella

AU - Uziel, Guy

AU - Amid, David

AU - Anaby-Tavor, Ateret

PY - 2025

Y1 - 2025

N2 - Large Language Model (LLM) agents hold promise for a flexible and scalable alternative to traditional business process automation, but struggle to reliably follow complex company policies. In this study we introduce a deterministic, transparent, and modular framework for enforcing business policy adherence in agentic workflows. Our method operates in two phases: (1) an offline buildtime stage that compiles policy documents into verifiable guard code associated with tool use, and (2) a runtime integration where these guards ensure compliance before each agent action. We demonstrate our approach on the challenging τ-bench Airlines domain, showing encouraging preliminary results in policy enforcement, and further outline key challenges for real-world deployments.

AB - Large Language Model (LLM) agents hold promise for a flexible and scalable alternative to traditional business process automation, but struggle to reliably follow complex company policies. In this study we introduce a deterministic, transparent, and modular framework for enforcing business policy adherence in agentic workflows. Our method operates in two phases: (1) an offline buildtime stage that compiles policy documents into verifiable guard code associated with tool use, and (2) a runtime integration where these guards ensure compliance before each agent action. We demonstrate our approach on the challenging τ-bench Airlines domain, showing encouraging preliminary results in policy enforcement, and further outline key challenges for real-world deployments.

UR - https://www.scopus.com/pages/publications/105039633362

U2 - 10.18653/v1/2025.emnlp-industry.41

DO - 10.18653/v1/2025.emnlp-industry.41

M3 - Conference contribution

AN - SCOPUS:105039633362

T3 - EMNLP 2025 - 2025 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Industry Track

SP - 595

EP - 606

BT - EMNLP 2025 - 2025 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Industry Track

A2 - Potdar, Saloni

A2 - Rojas-Barahona, Lina

A2 - Montella, Sebastien

PB - Association for Computational Linguistics (ACL)

T2 - 2025 Conference on Empirical Methods in Natural Language Processing: Industry Track, EMNLP 2025

Y2 - 4 November 2025 through 9 November 2025

ER -

Zwerdling N, Boaz D, Rabinovich E, Uziel G, Amid D, Anaby-Tavor A. Towards Enforcing Company Policy Adherence in Agentic Workflows. In Potdar S, Rojas-Barahona L, Montella S, editors, EMNLP 2025 - 2025 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Industry Track. Association for Computational Linguistics (ACL). 2025. p. 595-606. (EMNLP 2025 - 2025 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Industry Track). doi: 10.18653/v1/2025.emnlp-industry.41

Towards Enforcing Company Policy Adherence in Agentic Workflows

Abstract

Publication series

Conference

Bibliographical note

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this