Toward Pre-Deployment Assurance for Enterprise AI Agents: Ontology-Grounded Simulation and Trust Certification

面向企业级 AI 智能体的部署前保障：基于本体论的模拟与信任认证

Abstract: Pre-deployment verification of enterprise artificial intelligence (AI) agents remains a critical gap between large language model (LLM) capability benchmarking and production deployment. Post-deployment monitoring, human-in-the-loop controls, and prompt-level guardrails offer limited assurance once an agent is operating in production.

摘要： 企业级人工智能（AI）智能体的部署前验证，在大语言模型（LLM）能力基准测试与实际生产部署之间仍存在关键缺口。一旦智能体投入生产运行，部署后的监控、人在回路（human-in-the-loop）控制以及提示词层面的护栏所能提供的保障非常有限。

We propose an ontology-grounded verification framework combining three components: an Agent Operational Envelope formalizing the certification space across permissions, domain constraints, safety properties, governance rules, and autonomy levels; an ontology-to-scenario generation pipeline that derives regulatory, operational, and adversarial test scenarios automatically; and a Trust Certificate carrying a machine-verifiable attestation with graduated deployment verdicts (Approved, Conditional, Rejected).

我们提出了一种基于本体论（Ontology-grounded）的验证框架，该框架结合了三个核心组件：一是“智能体操作边界”（Agent Operational Envelope），用于将权限、领域约束、安全属性、治理规则和自主级别等认证空间形式化；二是“本体到场景生成流水线”，能够自动推导出监管、操作及对抗性的测试场景；三是“信任证书”，包含可由机器验证的证明，并提供分级的部署结论（批准、有条件批准、拒绝）。

A controlled pilot across four regulated industries (Fintech, Banking, Insurance, and Healthcare), instantiated as five industry-by-regulatory-regime cells across the United States and Vietnam, generated 1,800 scenarios evaluated against 125 primary-source regulatory requirements and 25 injected faults.

在一项涵盖四个受监管行业（金融科技、银行、保险和医疗保健）的受控试点中，研究团队在美国和越南构建了五个“行业-监管制度”单元，生成了 1,800 个测试场景，并针对 125 项原始监管要求和 25 个注入故障进行了评估。

Ontology-grounded generation (G4) achieved 48.3% regulatory coverage versus 33.1% for the persona-based baseline (corrected p = .0006) and the highest domain specificity (4.77/5.0; p = 2e-6). The coverage advantage over baseline and retrieval-augmented prompting was not robust after Bonferroni correction.

基于本体论的生成方法（G4）实现了 48.3% 的监管覆盖率，而基于角色（persona-based）的基准方法仅为 33.1%（校正后 p = .0006），且该方法在领域特异性上表现最高（4.77/5.0；p = 2e-6）。在进行 Bonferroni 校正后，该方法相较于基准测试和检索增强生成（RAG）在覆盖率上的优势不再显著。

Cross-validation across three LLM families (Claude Sonnet 4, Qwen 2.5 72B, Gemma 4 26B; 5,400 total scenarios) replicated the persona-versus-ontology pattern. The results establish ontology-grounded scenario generation as a credible complement to persona-based test suites for regulatory-intensive domains.

通过对三个 LLM 系列（Claude Sonnet 4、Qwen 2.5 72B、Gemma 4 26B；共 5,400 个场景）的交叉验证，重复验证了“角色基准 vs. 本体论”的模式差异。研究结果表明，对于监管密集型领域，基于本体论的场景生成是现有基于角色的测试套件的一种可靠补充。