# Route Before Response: Fail-Closed Routing for LLM-Enabled Customer-Service Software

DevFeed: [Route Before Response: Fail-Closed Routing for LLM-Enabled Customer-Service Software](<https://devfeed.tech/articles/route-before-response-fail-closed-routing-for-llm-enabled-customer-service-software-120816.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.07534>)

Author: Yuchen Cao, Zhenyu Mao, Jacky Wai Keung, Yifei Wang, Guancheng Lin, Jingyu Zhang

Published: 2026-11-06T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Software Engineering](<https://devfeed.tech/sources/arxiv-software-engineering.md>)

Topics: [Large Language Model](<https://devfeed.tech/topics/llm.md>), [rule-engine](<https://devfeed.tech/topics/rule-engine.md>)

Tags: [code-checks](<https://devfeed.tech/tags/code-checks.md>), [cs-se](<https://devfeed.tech/tags/cs-se.md>), [customer-service](<https://devfeed.tech/tags/customer-service.md>), [dispatch](<https://devfeed.tech/tags/dispatch.md>), [financial-institutions](<https://devfeed.tech/tags/financial-institutions.md>), [llm](<https://devfeed.tech/tags/llm.md>), [privacy](<https://devfeed.tech/tags/privacy.md>), [program](<https://devfeed.tech/tags/program.md>), [routing](<https://devfeed.tech/tags/routing.md>), [rules](<https://devfeed.tech/tags/rules.md>), [software](<https://devfeed.tech/tags/software.md>)

## AI overview

RouteBR separates model proposals from execution: one model call proposes a primary service action and registered object, then deterministic code checks identifiers and versioned rules before dispatch or a specified non-dispatch outcome. A blocked primary route cannot be replaced by a lower-ranked candidate. In 600 privacy-preserving scenarios based on recurring confidential-request patterns at one financial institution, RouteBR achieved 97.5% final-path exact match on the 200-case CONFIRM MAIN split, versus 95.5% for a matched one-call baseline given the same rules. A diagnostic baseline without the rule program had 94.0% action accuracy but 41.0% final-path exact match. RouteBR made no false dispatches across 166 non-dispatch MAIN and FLIP cases. The abstract reports that prediction rows were paired with raw receipts and gold and reviewer fields were excluded from model requests.

## Source excerpt

arXiv:2611.07534v1 Announce Type: new Abstract: LLM-enabled customer-service software in financial institutions must choose a downstream service before responding. Classifiers and LLM routers can identify a relevant label or destination, but a composed application must still verify dispatch permission and entity registration. We study this integration problem as boundary-aware pre-response routing. We present RouteBR, a controller in which one model call proposes a primary service action and registered object, then deterministic code checks identifiers and versioned rules before dispatch or a specified non-dispatch outcome. Lower-ranked candidates cannot replace a blocked primary route. We construct 600 privacy-preserving scenarios from recurring confidential-request patterns at a real financial institution; three engineers jointly reviewed the scenarios and labels. On the 200-case CONFIRM MAIN split, RouteBR reaches 97.5% final-path exact match. A matched one-call baseline given the same rules reaches 95.5%, whereas a diagnostic baseline without the rule program obtains 94.0% action accuracy but only 41.0% final-path exact match. Across 166 non-dispatch MAIN and FLIP cases, RouteBR produces no false dispatch. The artifact pairs all 1,960 prediction rows with raw receipts while excluding gold and reviewer fields from model requests. These results support separating model proposals from code-owned, replayable execution. Our implementation is available at https://github.com/YgcCoder/RouteBR.