Synthesis Through Simulation: Generating Coherent Enterprise Data via Scalable Agent-System Interaction

查看原文 / View Original 编辑 / Edit

Computer Science > Artificial Intelligence arXiv:2610.10549 (cs) [Submitted on 24 Sep 2026] Title: Synthesis Through Simulation: Generating Coherent Enterprise Data via Scalable Agent-System Interaction Authors: Yipeng Li, Ashutosh Hathidara, Jane Lo, Harshavardhan Abichandani, Gunraj Singh, Atin Ghosh.

计算机科学 > 人工智能 arXiv:2610.10549 (cs) [提交于 2026 年 9 月 24 日] 标题：通过模拟进行合成：通过可扩展的智能体-系统交互生成连贯的企业数据作者：Yipeng Li, Ashutosh Hathidara, Jane Lo, Harshavardhan Abichandani, Gunraj Singh, Atin Ghosh。

Abstract: Tool-calling agents have become central to enterprise AI, yet training and evaluating them at scale remains severely constrained due to business and legal restrictions on enterprise systems, data, and database schemas. Tabular data synthesis offers a natural alternative, but its effectiveness is fundamentally limited by structural validity and schema availability, while procedure-based approaches yield the opposite weakness, typically lacking distributional fidelity without per-domain authoring.

摘要：工具调用智能体已成为企业人工智能的核心，但由于企业系统、数据和数据库模式方面的商业及法律限制，对其进行大规模的训练和评估仍然受到严重制约。表格数据合成提供了一种自然的替代方案，但其有效性从根本上受到结构有效性和模式可用性的限制；而基于过程的方法则存在相反的弱点，通常在没有针对特定领域进行编写的情况下缺乏分布保真度。

We introduce Synthesis Through Simulation (STS), a schema-free data synthesis paradigm in which an LLM agent generates data by executing operations against policy-enforcing APIs within simulated enterprise environments. Because data is generated through the same environment that defines what is valid, STS guarantees structural validity by construction while decoupling validity enforcement from distribution modeling, allowing each to be addressed independently.

我们引入了“通过模拟进行合成”（STS），这是一种无模式的数据合成范式，其中大语言模型（LLM）智能体通过在模拟的企业环境中针对执行策略的 API 执行操作来生成数据。由于数据是通过定义有效性的同一环境生成的，STS 在构建时就保证了结构有效性，同时将有效性执行与分布建模解耦，从而允许分别处理这两者。

The Generalist Populator (GP), STS’s domain-agnostic agent, addresses the remaining challenges of distributional fidelity and synthesis scalability: GP achieves 0.88 average marginal fidelity and 100% constraint satisfaction across all ten environments without access to DB schemas, while statistical synthesizers are inapplicable to seven due to necessary seed data requirements, and schema-privileged agents fail 82% of trajectories on airline environment’s tightly coupled workflows due to brittle task composition.

STS 的通用领域智能体“通用填充器”（GP）解决了分布保真度和合成可扩展性方面的剩余挑战：GP 在无需访问数据库模式的情况下，在所有十个环境中实现了 0.88 的平均边际保真度和 100% 的约束满足率；相比之下，统计合成器因需要种子数据要求而无法应用于其中七个环境，而拥有模式权限的智能体则因任务组合脆弱，在航空环境的紧密耦合工作流中导致 82% 的轨迹失败。