@inproceedings{hartley-etal-2026-deploying,
title = "Deploying {LLM} Agents for Auditable Complaint Investigation in Banking",
author = "Hartley, John and
Hamill, Conor Brian and
Din, Azam and
Siddiqui, Fayz and
Acheampong, Kingsley Nketia and
Cowan, Greig A and
Khraishi, Raad",
editor = "Atanasova, Pepa and
Haffari, Gholamreza and
Wang, Minghan",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing: Industry Track",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-industry.227/",
pages = "3802--3813",
ISBN = "979-8-89176-460-6",
abstract = "Complaint investigation in retail banking requires handlers to synthesise evidence from heterogeneous enterprise systems into policy-aligned summaries. We present a production-deployed tool-augmented LLM agent at a major UK retail bank that assists this process under human review. Given a complaint record, the agent queries 16 typed, schema-validated tools and returns an investigation summary with a source-cited chronological evidence timeline. We evaluate the system using SME assessment, LLM-as-a-Judge monitoring, rework measurement, and downstream impact testing. On 187 paired complaints, SMEs judged the agent to match or exceed historical human-written production investigations in 90{\%} of cases across faithfulness, relevance, completeness, and coherence. Median investigation time falls from 30 minutes to 23 seconds. Operational controls include daily automated monitoring, pre-defined thresholds, hallucination flagging, and periodic human audit. These results demonstrate that constrained single-agent LLMs with typed tools are viable for scalable, auditable complaint investigation in regulated environments."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="hartley-etal-2026-deploying">
<titleInfo>
<title>Deploying LLM Agents for Auditable Complaint Investigation in Banking</title>
</titleInfo>
<name type="personal">
<namePart type="given">John</namePart>
<namePart type="family">Hartley</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Conor</namePart>
<namePart type="given">Brian</namePart>
<namePart type="family">Hamill</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Azam</namePart>
<namePart type="family">Din</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Fayz</namePart>
<namePart type="family">Siddiqui</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kingsley</namePart>
<namePart type="given">Nketia</namePart>
<namePart type="family">Acheampong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Greig</namePart>
<namePart type="given">A</namePart>
<namePart type="family">Cowan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Raad</namePart>
<namePart type="family">Khraishi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing: Industry Track</title>
</titleInfo>
<name type="personal">
<namePart type="given">Pepa</namePart>
<namePart type="family">Atanasova</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Gholamreza</namePart>
<namePart type="family">Haffari</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Minghan</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-460-6</identifier>
</relatedItem>
<abstract>Complaint investigation in retail banking requires handlers to synthesise evidence from heterogeneous enterprise systems into policy-aligned summaries. We present a production-deployed tool-augmented LLM agent at a major UK retail bank that assists this process under human review. Given a complaint record, the agent queries 16 typed, schema-validated tools and returns an investigation summary with a source-cited chronological evidence timeline. We evaluate the system using SME assessment, LLM-as-a-Judge monitoring, rework measurement, and downstream impact testing. On 187 paired complaints, SMEs judged the agent to match or exceed historical human-written production investigations in 90% of cases across faithfulness, relevance, completeness, and coherence. Median investigation time falls from 30 minutes to 23 seconds. Operational controls include daily automated monitoring, pre-defined thresholds, hallucination flagging, and periodic human audit. These results demonstrate that constrained single-agent LLMs with typed tools are viable for scalable, auditable complaint investigation in regulated environments.</abstract>
<identifier type="citekey">hartley-etal-2026-deploying</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-industry.227/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>3802</start>
<end>3813</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Deploying LLM Agents for Auditable Complaint Investigation in Banking
%A Hartley, John
%A Hamill, Conor Brian
%A Din, Azam
%A Siddiqui, Fayz
%A Acheampong, Kingsley Nketia
%A Cowan, Greig A.
%A Khraishi, Raad
%Y Atanasova, Pepa
%Y Haffari, Gholamreza
%Y Wang, Minghan
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing: Industry Track
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-460-6
%F hartley-etal-2026-deploying
%X Complaint investigation in retail banking requires handlers to synthesise evidence from heterogeneous enterprise systems into policy-aligned summaries. We present a production-deployed tool-augmented LLM agent at a major UK retail bank that assists this process under human review. Given a complaint record, the agent queries 16 typed, schema-validated tools and returns an investigation summary with a source-cited chronological evidence timeline. We evaluate the system using SME assessment, LLM-as-a-Judge monitoring, rework measurement, and downstream impact testing. On 187 paired complaints, SMEs judged the agent to match or exceed historical human-written production investigations in 90% of cases across faithfulness, relevance, completeness, and coherence. Median investigation time falls from 30 minutes to 23 seconds. Operational controls include daily automated monitoring, pre-defined thresholds, hallucination flagging, and periodic human audit. These results demonstrate that constrained single-agent LLMs with typed tools are viable for scalable, auditable complaint investigation in regulated environments.
%U https://aclanthology.org/2026.emnlp-industry.227/
%P 3802-3813
Markdown (Informal)
[Deploying LLM Agents for Auditable Complaint Investigation in Banking](https://aclanthology.org/2026.emnlp-industry.227/) (Hartley et al., EMNLP 2026)
ACL
- John Hartley, Conor Brian Hamill, Azam Din, Fayz Siddiqui, Kingsley Nketia Acheampong, Greig A Cowan, and Raad Khraishi. 2026. Deploying LLM Agents for Auditable Complaint Investigation in Banking. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing: Industry Track, pages 3802–3813, Budapest, Hungary. Association for Computational Linguistics.