Add CXBench to the Research page

_bibliography/papers.bib

@@ -241,6 +241,20 @@ @article{lin2026agenteval
   abstract={Conversational LLM agents can cause real-world harm when their internal workflows fail, such as completing a transaction without confirmation. Testing these state-dependent failures is difficult because critical boundaries, such as identity checks and confirmation gates, are hidden behind multi-turn conversational prerequisites, rendering them inaccessible to standard tests. We present AgentEval, a black-box testing framework that discovers and stresses these stateful boundaries. AgentEval interacts with an agent to mine a conversational workflow graph, a model of its behavior. Instead of prompting blindly, AgentEval uses this graph's structure to enumerate specific guards and prerequisites as test targets, replaying the conversational path to a boundary before applying a perturbation. AgentEval then executes each test, determining whether it passes or fails using only the conversation turns. We benchmark AgentEval against a privileged, white-box auditor with access to the agent's underlying source code, which AgentEval never sees. On four tau-3-bench agents, AgentEval successfully generates tests covering 23--38 distinct boundaries per agent; ablation studies attribute the gain to the graph's structure: 23 distinct boundaries versus 12 with a prompt-only baseline, at lower duplicate and false-alarm rates.}
 }
 
+@article{yu2026cxbench,
+  abbr={Preprint},
+  title={CXBench: A Benchmark for Process-Compliant LLM Agents in Regulated Customer-Experience Workflows},
+  author={Yu, Boxi and Zhang, Yuzhong and Su, Zhenfeng and Xi, Linfu and Yao, Yinghan and Chen, Shitao and Cao, Yang and Lin, Liting and Briand, Lionel and Muñoz, Emir},
+  journal={Preprint},
+  year={2026},
+  selected={false},
+  topic={TrustworthyAI},
+  url={https://cxbench.github.io/},
+  pdf={https://cxbench.github.io/assets/cxbench-paper.pdf},
+  code={https://github.com/CXBench/cxbench},
+  abstract={Tool-using language agents increasingly perform regulated customer-service workflows, where a correct final state is insufficient: agents must also provide disclosures, obtain consent, and follow required procedures. Existing tool-agent benchmarks largely score final business states, accepting trajectories that reach correct outcomes through non-compliant processes. We introduce CXBench, a benchmark that evaluates outcome and process compliance separately across six regulated domains. Each task combines a customer request, a tool-based environment, and explicit process requirements. Rule cards link cited regulatory obligations to observable trace evidence. A hybrid grader combines deterministic checks for ordering and authority with an LLM judge for semantic requirements. We define a false-pass as an outcome-pass run that fails process checks. Across seven models and 120 tasks, 55.5\% of the 840 retained executions are false-passes, representing 80.1\% of outcome-passing runs; only 13.8\% pass both checks. Process-inclusive evaluation changes the observed model ordering and exposes differences hidden by outcome-only scoring. CXBench makes these failures inspectable through trace-linked criteria.}
+}
+
 @article{xu2026bespec,
   abbr={arXiv},
   title={BeSpec: Behavior-Level Specification Alignment for Code Generation},

_news/announcement_13.md

@@ -0,0 +1,8 @@
+---
+layout: post
+date: 2026-10-18
+inline: true
+related_posts: false
+---
+
+We released [CXBench](https://cxbench.github.io/), a benchmark built with Genesys that scores LLM agents in regulated customer-experience workflows on both the business outcome and the required process. Across seven models, 80.1% of outcome-passing runs failed a process check. See [the research summary](/research/#cxbench).

src/data/site.ts

@@ -86,6 +86,26 @@ export const projects = [
     ],
     topic: 'Automated testing',
   },
+  {
+    id: 'cxbench',
+    paper: 'yu2026cxbench',
+    name: 'CXBench',
+    year: '2026',
+    venue: 'Preprint',
+    question: 'Does a correct outcome mean the agent followed the required process?',
+    summary:
+      'A benchmark that scores business outcomes and process compliance separately for LLM agents in regulated customer-experience workflows, built with Genesys.',
+    method:
+      'CXBench pairs each customer request with a tool-based environment and explicit process requirements across six regulated domains. Rule cards link cited regulatory obligations to observable trace evidence, and a hybrid grader combines deterministic checks for ordering and authority with an LLM judge for semantic requirements.',
+    finding:
+      'Across seven models and 120 tasks, 55.5% of the 840 retained executions were false-passes: they reached the correct business outcome but failed a process check. These account for 80.1% of outcome-passing runs, and only 13.8% of executions passed both. Process-inclusive scoring also changed the observed model ordering.',
+    topic: 'Trustworthy AI',
+    links: [
+      { label: 'Project page', href: 'https://cxbench.github.io/' },
+      { label: 'Paper', href: 'https://cxbench.github.io/assets/cxbench-paper.pdf' },
+      { label: 'Code', href: 'https://github.com/CXBench/cxbench' },
+    ],
+  },
   {
     id: 'rome',
     paper: 'yu2023rome',

src/pages/research.astro

@@ -20,8 +20,8 @@ const themes = [
     id: 'ai-systems',
     title: 'Testing AI systems',
     description:
-      'AI systems produce outputs that can be difficult to assess automatically. Controlled transformations and consistency relations provide practical ways to uncover mistakes in language and vision systems.',
-    projects: ['rome', 'tin'],
+      'AI systems produce outputs that can be difficult to assess automatically. Controlled transformations and consistency relations provide practical ways to uncover mistakes in language and vision systems, and process-level checks show when an agent reaches the right outcome the wrong way.',
+    projects: ['cxbench', 'rome', 'tin'],
   },
   {
     id: 'aiops',