Synthesis Through Simulation: Generating Coherent Enterprise Data via Scalable Agent-System Interaction
SAP's STS generates valid enterprise data by having LLM agents call policy-enforcing simulated APIs.
Synthesis Through Simulation generates enterprise data by having an LLM agent execute operations against policy-enforcing APIs inside simulated environments, so structural validity is guaranteed without database schemas. The domain-agnostic Generalist Populator reports 0.88 average marginal fidelity and 100% constraint satisfaction across all ten environments. Statistical synthesizers cannot run on seven environments that lack required seed data, and schema-privileged agents fail 82% of trajectories on the airline environment's tightly coupled workflows. SAP released the framework, environments, and generated datasets.
- Schema-free synthesis via agents calling simulated enterprise APIs
- Generalist Populator: 0.88 marginal fidelity and 100% constraint satisfaction
- Evaluated on ten environments without database schema access
- Schema-privileged agents fail 82% of airline workflow trajectories
- SAP open-sourced the framework, environments, and datasets
From the original article
1 min read at huggingface.co
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.
By Yipeng Li, Ashutosh Hathidara, Jane Lo · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.